This page contains discussions that have been archived from
Village pump (technical). Please do not edit the contents of this page. If you wish to revive any of these discussions, either
start a new thread or use the talk page associated with that topic.
Hello. I have been successfully running a pywikibot setup from toolforge since last few months (for Marathi wikipedia (mrwiki). In the beginning, the bot was running cronjobs on grid. I think 4-5 months ago, I was told to migrate to buster grid, and I did so. Currently everything is being migrated to kubernetes. But the help/user guide pages are not very clear. Do you guys have any idea how to migrate the tool to kubernetes? I am not sure who else still runs a pywikibot on toolforge. Any help/guidance will be appreciated a lot. Courtesy ping to
Klein Muçi (another pywikibot operator), and
Novem Linguae. I have provided some help pages below. Regards, —usernamekiran
(talk)11:34, 16 October 2022 (UTC)
I've been unable to get my tools to work on kubernets there are many unix tools missing or outdated, like simple wget. Maye I need to install them in my local directory somewhere. Finding it generally difficult and limiting, designed for web applications written in a few well known languages. If the grid is shut down good chance many of my tools will be unable to migrate from what I can tell. --
GreenC15:00, 16 October 2022 (UTC)
If you're using the system version of pywikibot, it would probably be better to wait for
phab:T249787 to be resolved first. If you have installed pywikibot as local python package, migrating should be straightforward. Just remove the crontab command and
create a scheduled k8s job, passing --image parameter as tf-python39. –
SD0001 (
talk)
18:19, 16 October 2022 (UTC)
A few days ago, I had first gone with
Novem Linguae's tutorial linked above. At least on the terminal, I was not getting any errors(cronjob.batch/exp4 created), but at the scheduled time, the job was not starting. There were no .err/.out files created with this method. If I go with the command method as suggested by
SD0001, then again the job is being scheduled. But at the scheduled time, it is not running. However, I get following error in the .err file:
Traceback (most recent call last):
File "/data/project/shared/pywikibot/stable/scripts/replace.py", line 155, in <module>
import pywikibot
File "/data/project/shared/pywikibot/stable/pywikibot/__init__.py", line 43, in <module>
from pywikibot.bot import (
File "/data/project/shared/pywikibot/stable/pywikibot/bot.py", line 111, in <module>
from pywikibot import config, daemonize, i18n, version
File "/data/project/shared/pywikibot/stable/pywikibot/version.py", line 26, in <module>
from pywikibot.comms.http import fetch
File "/data/project/shared/pywikibot/stable/pywikibot/comms/http.py", line 40, in <module>
import requests
ModuleNotFoundError: No module named 'requests'
After trying these two methods, I posted here. I did some tweaking again, but no positive result. —usernamekiran
(talk)16:34, 17 October 2022 (UTC)
@
Usernamekiran You are using the system version of pywikibot. Don't. Create a local venv and install all required dependencies in that. Then, when using toolforge-jobs command, instead of invoking python, invoke .../PATH/TO/VENV/venv/bin/python. Follow steps 1–7 in
here for how to create the venv and install dependencies. It is critical that it should be done within a webservice shell as mentioned in step 1. –
SD0001 (
talk)
03:57, 18 October 2022 (UTC)
It seems that there are the following classes on the body that can be targeted: pcs-theme-black, pcs-theme-dark, pcs-theme-sepia —
TheDJ (
talk •
contribs)
19:02, 19 October 2022 (UTC)
A content page with a template redirect at top is giving some weird results
Hi all, I just noticed that using a template redirect at the top of a content page makes it being interpreted as a redirect in some cases, while in others, it does not, which is very weird. Have a look at
User talk:Cunard. It uses {{userinfo}} right at the beginning of the page, which is a redirect to {{userpageinfo}}. Now, if we use {{
is redirect|User talk:Cunard|talk=yes}} → (it's blank, as it should be). Yet, when the same is used via {{automatic archive navigator}} at
User talk:Cunard/Archive 13, it starts to populate
Category:Archive pages whose parent page is a redirect (which it should not populate). But this doesn't appear to be a template/module only problem. I am using
User:BrandonXLF/NoRedirect.js , which uses something content.find('.mw-redirect') in order to find a redirect, and put a "no redirect" link besides it. When I write
User talk:Cunard on this page, the "no redirect" link does not appear. But at
User talk:Cunard/Archive 13, the link [[User talk:Cunard|current talk page]] produced via archive navigator, starts to bring a "no redirect" link besides it, as if it were a redirect. Again, it is not. Can someone help? Thanks! —CX Zoom[he/him](
let's talk • {
C•
X})15:10, 19 October 2022 (UTC)
@
CX Zoom:User talk:Cunard was moved to
User talk:Cunard/Archive 13 today, leaving a redirect. Cunard removed the redirect two minutes later but it was a redirect when
User talk:Cunard/Archive 13 was last edited and MediaWiki caches pages for performance reasons. I don't see the category or "no redirect" link now.
Purge a page to update its content. Make a
null edit to also update link tables for the page.
Bypass your cache if your own browser has cached a page.
User:BrandonXLF/NoRedirect.js checks for redirects by searching the cached code of the page itself (where redirects have class="mw-redirect"), not by checking whether the linked page is currently a redirect.
PrimeHunter (
talk)
15:39, 19 October 2022 (UTC)
In case it wasn't clear, it's irrelevant whether the linked page uses a template redirect. No feature causes this to treat the page itself as a redirect.
PrimeHunter (
talk)
15:45, 19 October 2022 (UTC)
Colorized diffs
Sorry if this has been asked before; I'm on a break and must hurry. Can you opt out of the colorized diffs? If so how. Thanks. --
John Cline (
talk)
19:37, 19 October 2022 (UTC)
Every time your browser accesses the Wikimedia servers, it sends your login cookie. If this cookie fails to arrive, or is corrupted along the way, or is otherwise invalid, you're treated as if you were logged out. --
Redrose64 🌹 (
talk)
20:28, 14 October 2022 (UTC)
Of late, I've found that my login session is only lasting for a day. If I'm making an edit, my session expires, and then I try to submit it, I see the edit window again with a message buried before the edit box saying that I'm logged out. It can be easy to miss this and accidentally submit a change while logged out.
isaacl (
talk)
05:15, 16 October 2022 (UTC)
Under what circumstances? When I find that the current page shows that I've been logged out? Yes, I typically force a browser reload, and then I just log back in.
isaacl (
talk)
18:41, 17 October 2022 (UTC)
Any time since the problem started. I don't think that forcing a reload is the same thing. (My theory is: logins require working cookies; bypassing removes all cookies, giving you a fresh and hopefully working set.)
Whatamidoing (WMF) (
talk)
19:01, 17 October 2022 (UTC)
Since you used the BYPASS shortcut, I assumed you wanted me to look at the "Bypassing cache" section, which shows how to force a browser reload. Forcing a reload in your browser doesn't remove your cookies; it causes your browser to request the page from the server rather than serving the page from its own internal cache. Yes, I've bypassed my browser's internal cache while viewing Wikipedia pages, probably every day. I'd as soon log in again than remove my cookies (as per the "Cache clearing and disabling" section), which would definitely remove traces of my login session from my browser.
isaacl (
talk)
19:22, 17 October 2022 (UTC)
Dominic Mayers, while it won't solve the issue of staying logged in, if not accidentally spilling your IP is important to you: when you edit using
Factotum it will reject any attempt to edit in a state different from the state when the page was loaded. If you were logged in when the page loaded but got logged out somehow, any attempt to edit will also be rejected. Similarly, if you weren't logged in when the page loaded and logged in on another tab, any attempt to edit will fail. It fails with a pop-up error and no option to re-submit, pretty much impossible to miss. If you actually want to be able to re-submit (and publish your IP) the standard editors are better for this particular situation though. — Alexis Jazz (
talk or ping me)
04:08, 20 October 2022 (UTC)
Lost ability to delete?
Hi. I'm logged in as myself and I don't seem to have lost my admin rights but I can no longer do deletions on English wikipedia because the option has disappeared from my toolbar. As far as I can see, I haven't changed my preferences, so I'm confused as to what is happening. Any thoughts?
Deb (
talk)
11:44, 19 October 2022 (UTC)
PrimeHunter Yes, I'm sorry for not being clearer. "Delete" used at one time to appear alongside the "edit" option at the top of the page, and in fact it still does, but only for a split second before being replaced with a smaller range of options, including "Page", which is what I've been using more recently. Earlier today, the "Delete" option did not appear in the drop-down; now it's back. I do not think I am imagining things, but clearly the problem has resolved itself.
Deb (
talk)
13:47, 19 October 2022 (UTC)
@
Deb: You didn't name your skin but in some skins the script moves the delete link from a "More" tab to the "Page" tab. Look for a "More" tab if it happens again.
PrimeHunter (
talk)
14:02, 19 October 2022 (UTC)
PrimeHunter The problem is coming and going now, and I can't find a "More" tab. I'm using "Modern". I'll try changing the skin and hope that it has some effect.
Deb (
talk)
18:00, 19 October 2022 (UTC)
@
Deb: That's why I asked for the skin as the first thing. Modern is no longer officially supported and you can only change back to it via a
special preferences link. Modern has no "More" tab. By default it has a "delete" link at top of the page. The MoreMenu gadget moves it to a "page" link. Both work for me. I guess
Special:DeletePage and other ways still work if the interface link is missing.
PrimeHunter (
talk)
18:27, 19 October 2022 (UTC)
PrimeHunter Both of those links were missing from the top of the page, but delete sometimes reappeared. Finally figured out that I was only using "Modern" on English-language Wikipedia, which is why it wasn't a problem elsewhere. Still no idea why it should change by the minute just because it is unsupported though! Thanks for your help.
Deb (
talk)
08:41, 20 October 2022 (UTC)
Gadgets like MoreMenu are scripts which run in your own browser after every page load to change the loaded page, e.g. adding or moving elements. This sometimes fails. Many scripts run in a sometimes random order and the order can influence whether it works. I wouldn't report it to the MoreMenu maintainers when Modern is unsupported and will be completely removed at some time.
PrimeHunter (
talk)
13:13, 20 October 2022 (UTC)
Talk page FAQ not appearing on mobile
Hi, apologies if this is the wrong spot. I'm using Vector legacy. On mobile the FAQ list on talk pages like
Talk:Homeopathy doesn't appear at all on mobile. Is there something that can fix this?
Newystats (
talk)
00:39, 20 October 2022 (UTC)
MOS:COLLAPSE and tables/sections defaulting to collapsed state on mobile app
Is this something the app developers did on their own, or is this something we can correct for in our stylesheets here? See List of NC-17 rated films (
|
talk |
history |
protect |
delete |
links |
watch |
logs |
views) for example, which is mostly one large table of content, but which on the mobile app is completely collapsed by default (as are the references). The same thing also appears to happen to infoboxes. —
Locke Cole •
t •
c16:16, 14 October 2022 (UTC)
Sections have always been collapsed by default on mobile for a few reasons that basically reasonable. And infoboxes have basically always been after the first paragraph, which is what you may be identifying as 'collapsed'. If infoboxes are actually collapsed, I'd be surprised.
Izno (
talk)
16:20, 14 October 2022 (UTC)
When I look at
Avengers: Endgame in Safari on my tiny iPhone, I see the lead, the infobox, and then expandable headers for all of the sections. I don't use the mobile interface a lot, but this looks like what I see in every other article when I use that interface. If you are using a specific app on a specific operating system, state what it is so that other editors have a chance to replicate your viewing conditions. –
Jonesey95 (
talk)
17:08, 14 October 2022 (UTC)
> Is this something the app developers did on their own
The iOS app has always had this behaviour.
> The "References" sections are also, against the guidance of the MOS, collapsed.
Consider that mobile allows you to tap a reference which brings up a popup with the full reference. Also MAYBE, we should update our guidelines to take into account other form factors and their UX. My Alexa also doesn't spill out the entire article and its entire reference section. —
TheDJ (
talk •
contribs)
19:22, 19 October 2022 (UTC)
The iOS app has always had this behaviour. That's a shame, so list articles have always been fully collapsed for mobile app users? Consider that mobile allows you to tap a reference which brings up a popup with the full reference. That's true, my major concern when bringing this here initially was the "List"-style articles which often times are 95% table content that, on this app, are fully collapsed for readers. I can't possibly see how that would be desirable, and could see it causing confusion for casual readers just trying to find content and not understanding that it's been auto-collapsed. I also have to wonder if there's not accessibility concerns here. —
Locke Cole •
t •
c16:20, 20 October 2022 (UTC)
I also have to wonder if there's not accessibility concerns here. Not in principle, but I just tested it on my phone using VoiceOver, and while the collapsed table can be expanded, there is no indication that the collapsed thingie is an interactive element, so that could be improved.
Matma Rextalk23:34, 20 October 2022 (UTC)
Note that this shows the tables and infoboxes as expanded on the initial load, unlike the real iOS Wikipedia app; and it also works for the user namespace, even though the real app will actually show the normal mobile web view for them, so you can test changes with only a moderate amount of annoyance.
Matma Rextalk22:34, 17 October 2022 (UTC)
@
Matma Rex: I guess I sort of expected the behavior to be identical on Android and iOS, and I wasn't aware there were other official Wikipedia apps from the WMF. Apologies for that. I've also learned to avoid uploading screenshots of software, as those are fair-use typically and not allowed outside of article-space. At any rate, thank you for confirming that it's an issue. Is the issue limited to just iOS? —
Locke Cole •
t •
c04:10, 18 October 2022 (UTC)
Can you clarify what is the relevance of MOS:COLLAPSE to your question? I missed this, but the relevance is that an app produced and provided by the WMF shouldn't actively be undermining our style guides (unless there is a clear technical need). I'd be inclined to submit a bug report about it, but I'd like to see if someone with an Android device can confirm if it's just iOS or both. —
Locke Cole •
t •
c16:22, 19 October 2022 (UTC)
I don't think this undermines the style guide. The style guide discourages specific methods of collapsing to ensure that all interfaces for reading Wikipedia can display the content. But the app is an interface, and it can make its own decisions. As another example, the mobile web version is another interface, and it collapses all sections to make it easier to navigate articles, and I always thought that this is quite nice.
Matma Rextalk23:31, 20 October 2022 (UTC)
I have been trying and trying this morning to figure out why four sources (delineated as "Further reading") in this article are throwing "Harv error/CITEREF" messages. Can someone among all you Village pump/citation wizards take a look and tell me what the problem is?!? Maybe fix one of them and I'll take care of the rest... Thanks,
Shearonink (
talk)
15:09, 21 October 2022 (UTC)
Probably because they are cited using sfnp with the target being under the Further reading section. Further reading is meant for sources which aren't cited, and the friendly scripts that give you reference errors count that as a error.
Nigel Ish (
talk)
15:21, 21 October 2022 (UTC)
Kept getting nothing but "upstream connect error or disconnect/reset before headers. reset reason: overflow". Had to keep trying just to get it to post this.... Abductive (
reasoning)03:20, 22 October 2022 (UTC)
Thanks. I figured that it was a continuing problem, but given that it is Friday night/Saturday morning, I thought should document it here in case nobody else saw it. Abductive (
reasoning)06:08, 22 October 2022 (UTC)
Is a creative signature with 3+5 tildes allowed?
It was brought to my attention that
Paine Ellsworth has a very creative signature that consists of ~~~ <small>~~~~~</small>. This puts the timestamp within a small tag while the rest remains normal-sized.
WP:SIG doesn't seem to specifically disallow it, though it does generally suggest using 4 tildes. The issue with the creative signature is that tools and bots might get confused by it, not expecting the timestamp to be contained in its own element. Is this permissible or not? Should we add a note to
WP:SIG to clarify if/why this is(n't) allowed/recommended? — Alexis Jazz (
talk or ping me)
04:24, 20 October 2022 (UTC)
This tool appears to show that
Paine Ellsworth's signature complies with MediaWiki's current requirements.
Sometimes their usage runs afoul of
MOS:FONTSIZE (nested small tags), but I don't think putting a time stamp inside of tags will confuse any well programmed bots. Signatures often end up inside of <div>...</div>, <small>...</small>, or other tags. Is there evidence of bots being confused by such formatting? –
Jonesey95 (
talk)
04:49, 20 October 2022 (UTC)
One of my main concerns when I tested the sig many moons ago was if it would trigger notifications, and it does. I have a couple of other sigs I use sometimes, but they don't trigger noties so I don't use them very often. The main sig I use, which amounts to ~~~ <small>~~~~~</small> and which I apply with
TemplateScript, an awesome tool, does trigger noties, so I haven't hesitated to use it. And by the way, I use TemplateScript to apply
rcat templates, the {{Talk page of redirect}} template and a few other templates. It has saved me so much editing time! Highly recommend it. P.I. Ellsworth ,
ed.put'r there05:04, 20 October 2022 (UTC)
Jonesey95, that tool only checks the adjustable part (the three tilde part), the small tags for the timestamp are not considered there. Signatures often end up inside of <div>...</div>, <small>...</small>, or other tags. Whole signatures, yes, but very rarely the timestamp individually. Is there evidence of bots being confused by such formatting? I don't have any evidence of bots getting confused, though I'd imagine an archiver bot could miss these timestamps which could theoretically result in early archiving. My own script,
Factotum, got slightly confused by it but this will be resolved. (if it isn't already, I need to test a bit more to be sure) — Alexis Jazz (
talk or ping me)
05:27, 20 October 2022 (UTC)
Meh sure. I don't really like it, but don't see any problems with it. Reply-tool works with it, and if bots have a problem with it - that is the bot's problem. —
xaosfluxTalk13:50, 20 October 2022 (UTC)
Xaosflux, I consider it a matter of pros and cons. Pro: a small number of people (one?) can customize their signature in unexpected ways. Cons: can affect readability, needs to be taken into consideration/tested by every script author and bot operator that works with signatures, can be confusing because timestamps normally have no styling. And it's a small-tag now, but if this is acceptable it's a matter of time until someone puts their timestamp in sup or sub tags. Or in <span style="color:red"> which would make me unhappy as it would force me to process all span tags when looking for signatures. Currently dd, dl, ul, li and a few other bits are enough, but span can add quite a bit as it's very common. — Alexis Jazz (
talk or ping me)
17:16, 20 October 2022 (UTC)
There are
plenty of instances of signature time stamps wrapped in <sup>...</sup> and other tags, and they do not seem to have caused problems yet. What is the actual (non-hypothetical) problem here? –
Jonesey95 (
talk)
17:46, 20 October 2022 (UTC)
It is common for an entire comment, including signature to be inside of SMALL tags, and we already have a do not make your signature so small that it is difficult to read rule, a single small/sub/sup isn't generally considered "too small". We also already have a guideline that the timestamp format must be the normal format (i.e. (HH:MM, D MM YYYY (UTC)) ). I don't see any special "technical" problems with this that we need to deal with here at VPT. That the formatting of one persons comment interferes with customization that some select editors use isn't a very strong argument, a stronger argument would be that it makes things problematic for everyone. If you want to propose new restrictions to signatures, please start a discussion at
Wikipedia talk:Signatures and properly advertise it. —
xaosfluxTalk17:49, 20 October 2022 (UTC)
Xaosflux, feel free to start a discussion about that. But for what's it worth, my signature isn't split. And if anyone needs to select these with CSS it's cheap because it has a class. The goal of my span is to resolve the shortcomings of regular signatures (lack of a machine-readable date, username and pagename) while including the traditional four tildes within. Signatures with split 3+5 tildes are purely for cosmetic reasons. It is common for an entire comment, including signature to be inside of SMALL tags, That's less of an issue as the signature isn't split. If you want to propose new restrictions to signatures I'm just asking. If this is considered okay, we could consider adding a note to
WP:SIG. If it might not be, we could consider a proposal on restrictions. There are plenty of instances of signature time stamps wrapped in <sup> @
Jonesey95, The search link actually times out with incomplete results (31 results or so). Restricted to article talk
[1] I get 51 results. Excluding User:CrowzRSA I get 27. I found a 2006 IP comment on
Talk:Carl Craig, AxG on
Talk:Harry Potter and the Deathly Hallows/Archive 4 (Diff 95767573) in 2006, 8 comments by Balthazar in 2007 and 2008. None of the others are signatures. I'm not quite convinced yet there are "plenty of instances". Signature splitting isn't very common. What is the actual (non-hypothetical) problem here? It's a possible issue for developers who get one more quirk they need to test/work around and can reduce performance in some cases as the parentElement of timestamps becomes unpredictable. — Alexis Jazz (
talk or ping me)
13:09, 21 October 2022 (UTC)
FWIW, just like the one above, I don't "like" your signature, but I also don't really care that much - ensuring that they are human readable is much more important then that they are machine readable to me. —
xaosfluxTalk13:30, 21 October 2022 (UTC)
Xaosflux, my opinion would be the reverse: make them machine readable. That's the hard part, if they are machine readable they can easily be made human-readable. The other way around is much harder. But let's agree to disagree on that one. — Alexis Jazz (
talk or ping me)
17:36, 21 October 2022 (UTC)
Does it really matter as long as you can tell who wrote the message and the bot can tell when the message was written it is all good. I can't see a reason why you regularly need to read the timestamp of someone else. Let people do what they want.
Terasail[✉️]18:46, 20 October 2022 (UTC)
If such formatting were used in the statement of an RfC, it would cause Lint errors on the RfC listing pages. This is because Legobot copies the text of the RfC statement, the signature (if any) and the first valid timestamp. It takes the end of the timestamp as the string "(UTC)"; so if there is a closing tag such as </small> (as used by
Paine Ellsworth) or </span> (as used by
Alexis Jazz), that tag would not be copied to the RfC listings. --
Redrose64 🌹 (
talk)
18:54, 20 October 2022 (UTC)
Indeed, I've seen people, myself included, putting an "edited on" statement (timestamp included) inside <small> tags. In which case too, the bot will not give the appropriate results. Correct me if I'm wrong, I believe
User:RMCD bot compiles the opening statements along with the additional markup succeeding (UTC) which would be more ideal for such purposes. —CX Zoom[he/him](
let's talk • {
C•
X})11:18, 21 October 2022 (UTC)
You're right about RMCD bot, because I've relisted many move requests as well as opened some and without any timestamp malformity errors. Haven't thrown any errors with RfC closures, RfD closures,
Closure request closures, MRV closures or any other types of discussions/surveys in which I've participated, and I've been using the small tags with my timestamp for years. P.I. Ellsworth ,
ed.put'r there12:18, 21 October 2022 (UTC)
Redrose64, if I were to guess, I'd think this bot would detect/select the end of the comment by looking for a timestamp. That's not quite ideal, as mentioned people sometimes post timestamps that aren't part of a signature. They also sometimes add a "P.S." after a signature. The bot should select the whole line. — Alexis Jazz (
talk or ping me)
18:00, 21 October 2022 (UTC)
I think we should accept it as long as it meets the following basic criteria:
Displays the output of ~~~~~ in such a way that would follow accessability requirements for the user name
Our bots can understand it correctly
It's acceptable under MediaWiki's current requirements
@
Animal lover 666 your #2 is completely arbitrary; we have 296 bots, and any number of them can be created in the future. Bots should work around humans, not the other way around. Now if this was breaking the interface or even extensions, I'd be much more supportive of restrictions. —
xaosfluxTalk13:33, 21 October 2022 (UTC)
In an ideal world yes, bots should work around humans. But also humans should not go out of their way to make things difficult for bot authors - it's a partnership! Bot authors and maintainers are in short supply, I think it's reasonable to ask people to follow normal signature formats (even if it's not strictly against a written policy) to make things easier rather than forcing every bot to implement signature parsing systems.
My take on this specific case is "please don't" and a note that if tools don't work for you, that's your fault. And other people might get annoyed at you for that, especially if it affects different "easy reply" tools.
Legoktm (
talk)
18:50, 21 October 2022 (UTC)
Legoktm,
Jonesey95, after further analyzing the issue that was reported I found this and I do believe there is an issue here. I had already added an exception just for Ellsworth. (I know zero other active users who split their signatures, they might exist, but I don't know them) I had forgotten about the exception but it was in the code comments. The exception checked one link in the parentElement, the one closest to the timestamp. This was sufficient for Ellsworth, but @
Qwerfjkl (the reporter) also loaded a script,
User:Qwerfjkl/scripts/talkback that adds more links. So detection failed. I have now "solved" this by searching more links in the parentElement, but this increases the risk of false positives. Typically when people post timestamps that aren't signatures they are either encapsulated in their own tag (like small or sup) or surrounded by more plain text. When there's more plain text, the textNode in question is typically realistically too long for a signature timestamp and can be rejected on that basis. When the timestamp has it's own element, it can be rejected because there's no username in that element. That's where Paine Ellsworth comes in. The timestamp having its own element can no longer be used as a reason to reject it as a possible signature. Now for the real issue: how does one tell the difference between the following? :@[[User:Jonesey95]], this isn't a signature. Edit: but how would you know? Edited <sup>17:20, 22 October 2022 (UTC)</sup> :The following IS a signature. [[User:Paine]] <small>17:20, 22 October 2022 (UTC)</small> Without split signatures I can just reject these. Now I have two choices: get false positives or fail to detect Paine's comments. Is there any other way? FYI, DiscussionTools generates false positives in such cases. If split signatures would be disallowed the false positive rate could be reduced for DiscussionTools, Factotum and any other easy-reply tools. — Alexis Jazz (
talk or ping me)
17:33, 22 October 2022 (UTC)
ISBN checksum error
Something squirrelly seems to be going on with the Bearss 1991 cite at
Battle of Lake Providence. It's flagging an isbn error, but I'm pulling that exact isbn from the copyright information on my print copy of the book. I'm assuming the publisher used a non-standard isbn. Is there a way to suppress the error in the article?
Hog FarmTalk05:17, 23 October 2022 (UTC)
The publisher probably didn't calculate the checksum correctly, I've seen that happen with older books. You can put the ISBN between (( and )), that'll make the Cite template accept the ISBN 'as-written', without any error checking. --
rchard2scout (
talk)
07:59, 23 October 2022 (UTC)
@
Cryptic: I often want to know why a subcat or article is indirectly in a category where it seems not to belong. Is there a category route tool: something with the functionality of
this hack but more elegant and flexible?
Certes (
talk)
13:56, 23 October 2022 (UTC)
And
here's a version that shows the whole route in one query. Doing it manually was just as easy for depth 3, but depth 7 as in your example wouldn't have been. —
Cryptic14:43, 23 October 2022 (UTC)
I use the default Vector. Also, all other MoreMenu links work, and the lines 25-34 I added just today, so maybe I'm messing something up there and don't know how to fix that. —CX Zoom[he/him](
let's talk • {
C•
X})—CX Zoom[he/him](
let's talk • {
C•
X})17:34, 23 October 2022 (UTC)
config.page.encodedName is encoded by encodeURIComponent(), which encodes colons, slashes, etc. unlike mw.util.wikiUrlencode(), so the colon is %3A not :. And not sure what .join('.') is supposed to do, as it would replace a colon with a period if the template name without the namespace prefix includes a colon.
Nardog (
talk)
17:49, 23 October 2022 (UTC)
Thank you very much! Well, you see, that is what a coding illiterate gets after swifting through dozens of online material, none of which directly deal with the task in question. —CX Zoom[he/him](
let's talk • {
C•
X})17:59, 23 October 2022 (UTC)
Hi, I have a watchlist of about 8500 pages; I am sitting on it with the "500 changes, 3 days, group results by page" options. There's six active filters. I don't use the live updates. Last time I was logged in, 2-3 weeks ago, it seemed fine. Now it is bogged down and every action I take is slow. There's been a Chrome update since then; has the Enhanced Watchlist page gained any complexity or size in terms of scripting or tooling behind the scenes, that might cause the slowdown? My CPU is an i5 (2018) and I've got 16GB of RAM.
Elizium23 (
talk)
15:41, 23 October 2022 (UTC)
No idea, but I fixed a slow-down by adopting a new strategy. Keep your watchlist settings and put up with the fact that it will be slow the first time. I have a browser bookmark for the URL with the right settings. However, when you want to refresh, change the period to perhaps 6 hours and use Show. For the rest of your session, you can refresh that and it will be much faster.
Johnuniq (
talk)
01:54, 24 October 2022 (UTC)
Please, anyone who could help me solve the problem of inline references' preview - I get three kinds of previews, one of which covers ref number so I can't click on it and go directly to the reflist to check whatever interests me, regardless of what I see in preview. I tried to check and uncheck various combos in my preferences and global preferences, but I am still not able to shut down this one, annoying kind of preview, which, by the way, appeared just recently (few years ago). Also, the simplest and probably default version of preview (which is most clear and doesn't cover ref-number), actually appears extremely rarely and it does not look like in
Reference Tooltips , it look like a link preview and it lack that options/preferences link under gear icon. Anyway, if someone could offer a lifeline, I am drowning here with this annoying ref-preview.
౪ Santa ౪99°18:18, 23 October 2022 (UTC)
Hello, I am on laptop, and your link send me to Prefereces-Beta (tab) where the box is checked but locked so I am unable to uncheck it. Why is locked - maybe under some other tab it is checked sort of originally, and that is where it should be unchecked. However, I looked everywhere, and tried to uncheck, most of all under Gadgets, but nothing. I must note that many of the boxes under Editing tab are also locked, only place where all the boxes and buttons are unlocked for clicks are those under Gadgets.
౪ Santa ౪99°23:15, 23 October 2022 (UTC)
@
Santasa99 under beta, see if you have "Automatically enable most beta features" enabled at the very top, and if you can uncheck that to unlock that page. —
xaosfluxTalk23:30, 23 October 2022 (UTC)
But I haven't - "Automatically enable most beta features" is unchecked whole this time. Now I am starting to fear this is some f-up with Java, although I have enabled Java for wikipedia.org in my Edge.
౪ Santa ౪99°01:10, 24 October 2022 (UTC)
Under Beta features, my Content Translation is also checked and locked, but IP Information, which I checked yesterday is unlocked and I could uncheck the box.
౪ Santa ౪99°01:20, 24 October 2022 (UTC)
People, I have found the way to unlock References Preview under both Gadgets and Beta and to uncheck it, however, this does not help shut down that annoying kind of ref preview. It persists whatever I do. Some kind of total preferences reset, and default preview is a goal.
౪ Santa ౪99°06:55, 24 October 2022 (UTC)
And after desperate "Reset settings: Restore all default preferences (in all sections)" under User Profile tab, nothing happened, preview that obscuring ref No. is still my main preview.
Santasa99 (
talk)
07:15, 24 October 2022 (UTC)
Thanks, xaosflux. I will try this in an hour or so, and i will get back to you here, if you don't mind, whatever happens.
౪ Santa ౪99°16:37, 24 October 2022 (UTC)
OMG, @
Xaosflux, this helped :-))). Thank you a thousand times, thank you so much!!! I do not know what if anything is done to the Preferences settings by this move, if this changed anything there, but at this point I don't care.
౪ Santa ౪99°17:30, 24 October 2022 (UTC)
Yeah, I simply deleted all entries (all lines of script) in both, and I hope that's OK? I mean, it won't create some other trouble because deleting it all.
౪ Santa ౪99°17:32, 24 October 2022 (UTC)
@
Santasa99: The only issue with blanking your .js is that scripts you loaded in there such as
User:Headbomb/unreliable will no longer work for you. I would try restoring the scripts you still use one at a time, reverting any addition that causes problems, so you can enjoy as many useful scripts as possible and identify which one is breaking something for you.
Certes (
talk)
19:18, 24 October 2022 (UTC)
Thank you, @
Certes. Yes, I realized that I don't need to give up other useful scripts. I will certainly test every script individually next time, to know what exactly is doing - especially not to use one that caused me a problem with ref-preview :-). Thank you again, and thanks to all who tried and helped. Cheers.
౪ Santa ౪99°19:38, 24 October 2022 (UTC)
Tech News: 2022-43
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you.
Translations are available.
Recent changes
There have been some minor visual fixes in Special:Search, regarding audio player alignment and image placeholder height. Further details are in
T319230.
On Wikipedias, a new
preference has been added to hide article thumbnails in Special:Search. Full details are in
T320337.
Problems
Last week, three wikis (French Wikipedia, Japanese Wikipedia, Russian Wikipedia) had read-only access for 25 minutes. This was caused by a hardware problem.
[2]
Changes later this week
The
new version of MediaWiki will be on test wikis and MediaWiki.org from 25 October. It will be on non-Wikipedia wikis and some Wikipedias from 26 October. It will be on all wikis from 27 October (
calendar).
Some wikis will be in read-only for a few minutes because of a switch of their main database. It will be performed on 25 October at 07:00 UTC (
targeted wikis) and on 27 October at 7:00 UTC (
targeted wikis).
Starting on Wednesday, a new set of Wikipedias will get "
Add a link" (Assamese Wikipedia, Bashkir Wikipedia, Balinese Wikipedia, Bavarian Wikipedia, Samogitian Wikipedia, Bikol Central Wikipedia, Belarusian Wikipedia, Belarusian (Taraškievica) Wikipedia, Bulgarian Wikipedia, Bhojpuri Wikipedia, Bislama Wikipedia, Banjar Wikipedia, Bambara Wikipedia, Bishnupriya Wikipedia, Breton Wikipedia, Bosnian Wikipedia, Buginese Wikipedia, Buryat Wikipedia, Indonesian Wikipedia). This is part of the
progressive deployment of this tool to more Wikipedias. The communities can
configure how this feature works locally.
[3]
Pages with titles that start with a lower-case letter according to Unicode 11 will be renamed or deleted. There is a list of affected pages at
m:Unicode 11 case map migration. More information can be found at
T292552.
The Vector 2022 skin will become the default across the smallest Wikipedias.
Learn more.
@
Kautilya3: is inappropriately removing access-date from news citations on Wikipedia pages
Special:Diff/1117819316 On being asked to revert, they are citing document pages that do not support their action.
[8]. Is he correct in removing? Kindly ask them to self revert and restore the access-dates.
Venkat TL (
talk)
08:59, 24 October 2022 (UTC)
quoting from the page "Access dates are not required for links to published research papers or published books." @
Kautilya3 you have removed access dates from the news sites that update their articles all the time, especially Indian ones. Please self revert.
Venkat TL (
talk)
09:19, 24 October 2022 (UTC)
In the examples provided, aren't all of the citations already archived, their content links therefore rather immutable and the access date largely irrelevant?
Iskandar323 (
talk)
09:19, 24 October 2022 (UTC)
I thought that access-date was useful for tracking down versions of a website, for example at archive.org if needed. Also, if an update is issued to the website. The place to ask would be at the talk of {{cite news}} which redirects to
Help talk:Citation Style 1. If the sites are already archived, the access date is useful to confirm that the correct archive date is used.
Johnuniq (
talk)
10:01, 24 October 2022 (UTC)
The pages that have changed on the news website are not being cited. Currently it is the archived versions that are being cited, which will never change. No access-date needed, which will only serve to confuse readers. The archive-date is what matters. --
Kautilya3 (
talk)
10:08, 24 October 2022 (UTC)
@
Kautilya3 Those links are neither from "research papers or published books", your removal has no basis. News sites sometimes update the article after several months/years. Accessdate is very much needed for them. Please self revert.
Venkat TL (
talk)
12:38, 24 October 2022 (UTC)
We have two dates, Access-date for when the original URL was accessed to include the information, and archive-date, for when the archival version was created. For any online source that you've archived ( eg where url= is used), you need both, neither are optional.
Masem (
t)
12:44, 24 October 2022 (UTC)
you need both, neither are optional is a false statement per quite some discussion at
Help talk:CS1. |archive-date= is quite sufficient.
That said, I do not understand why this discussion is at
WP:VPT. The issue is not technical in nature. Either this requires personal dispute resolution, or the more appropriate place would indeed be at
Help talk:CS1.
Izno (
talk)
17:02, 24 October 2022 (UTC)
It seems wrong to not include the original accessdate, since archive creation can be automated. Eg: an article is initially created on Jan 1, and incorporated into WP on Jan 15. A substantive change is made on the article on Feb 1 which technically would require an extensive change on WP, but which is never caught. An archive is made March 1. To the reader, without the access-date, it looks like our use of the article reflects a completely different picture of the article presented at the archive-date, and thus harms our reputation. The access-date inclusion would make it clear to a reader that the info was captured before the change, and thus can flag the passage for appropriate updating.
Masem (
t)
01:37, 25 October 2022 (UTC)
That would be a behavioral issue, not an inherent flaw with removing an access date where an archived date is already provided.
That aside, even in your scenario the error can still be identified, as generally the kinds of external webpages that have substantive change that we typically need to care about (news websites, typically) have their own publication date and amendment date(s), in which case the citation date will not match the source date. Pages which don't have a publication date are suspect generally too from a
reliable source perspective.
Izno (
talk)
02:02, 25 October 2022 (UTC)
And to point out, we need the access-date info there so that if someone goes back to check and finds an update has been made post-access-date that potentially creates a problem in our summary of it, our article and that citation can be updated.
Masem (
t)
12:46, 24 October 2022 (UTC)
@
Kautilya3: When
Template:Citation#URL says Not required for linked documents that do not change, it means the documents do not change, typically because they are printed books or research papers, and the link goes to a website which displays the document. The citation should have enough information to uniquely identify the actual document in such cases. Any news site, archive site or other website can close, change url's or remove content so we want access dates for those to help track down disappearing or changing content.
PrimeHunter (
talk)
13:19, 24 October 2022 (UTC)
Can somebody kindly collapse all the sections of Projects founded, Contests and challenges, Featured content, Good articles and DYKs and shiny things at
User:Dr. Blofeld so they're neatly in column sections shrunk by default and just the headers showing? ♦
Dr. Blofeld13:07, 25 October 2022 (UTC)
I converted the top section in a way that somewhat replicates the previous look. If you want to preserve every little style choice, like the font and the borders, there are probably additional style changes you can make using the parameters available at {{cot}}. If you like how it looks, you can convert the other sections using the same method. –
Jonesey95 (
talk)
16:02, 25 October 2022 (UTC)
Cheers Jonesy, I've replicated that for the others but it's now shrunk into a smaller frame, can you set it at the width of the page but in default shrunk as it is now? Also there's a gap between Contests and featured sections for some reason..♦
Dr. Blofeld16:51, 25 October 2022 (UTC)
I don't know a technical fix but you might have to split the results somehow (by country?) and then transclude them into the article.
Nthep (
talk)
14:40, 24 October 2022 (UTC)
The post-expand include size is about how much you transclude in total including transclusions made by called templates without being in the output. It becomes larger if you move content to other pages and transclude it from there, unless you use partial transclusion to only transclude some of it like a summary. The page is currently 14% over the limit. The England section alone uses 91% of the limit so it could be split to another article. 9% isn't much room if more content is added or the used templates become more complex.
PrimeHunter (
talk)
15:27, 24 October 2022 (UTC)
It rendered fully at the latest Internet Archive snapshot on 21 August 2020. The wikitext was similar and a preview of the version at the time now breaks at the same spot after Hamilton, so the used templates did increase their footprint since then.
PrimeHunter (
talk)
15:41, 24 October 2022 (UTC)
One solution here would be to re-write those templates to be entirely lua. Using a template that calls a lua module results in double-counting the output of the module. --
Ahecht (
TALK PAGE)
17:38, 25 October 2022 (UTC)
I've seen this a lot on older pages. I'm not exactly sure why it shows that however I notice that it usually appears on articles that have been around for a while. ―
Blaze WolfTalkBlaze Wolf#654513:43, 25 October 2022 (UTC)
I have been getting new and unregistered users not responding to any of the questions and then pressing Publish changes without giving any meaningful feedback. I would like to implement an editnotice that encourages users to give feedback, but cannot leave everything blank and that it will be removed if it is posted without doing anything. If you look at the history of the page
HERE, you will see what's been going on. It is frustrating that people are not using this page for its intended purpose.
Interstellarity (
talk)
20:30, 25 October 2022 (UTC)
@
Xaosflux: Thanks, I have put in an edit request. I was hoping that you could look over the editnotice to see if it looks good or otherwise be improved. I want it to grab the attention of newcomers so that they are discouraged from implementing blank feedback.
Interstellarity (
talk)
12:33, 26 October 2022 (UTC)
Looking at the page history, I'm creating it at the earliest. But feel free to implement and/or propose more improvements to it. —CX Zoom[he/him](
let's talk • {
C•
X})12:56, 26 October 2022 (UTC)
{{Cite web}} is showing an error where one doesn't exist
On
Solo (Frank Ocean song), a citation from {{Cite web}} includes an article from someone named "Chris Author", hence: last=Author is one of the parameters. {{Cite web}} doesn't like this and claims that it's a generic name and identifies this as an error. I don't see how to fix this. Evidently, there is some error-correcting mode for identifiers like DOIs or ISBNs, but not names. Can anyone help here? ―
Justin (koavf)❤
T☮
C☺
M☯00:13, 26 October 2022 (UTC)
I've recently started being active again on the Icelandic Wikipedia. We only have 55.000 articles. So I had the idea, would it be possible to check every article to see if there are obvious mistakes or outdated information. I've already found a few by using "random article" feature. I did a quick search for a tool that would help me be more systematic and also would allow other people to join the effort. The tool should be able to cache all the articles in the Wiki, list them up, and then users can mark when they have checked the quality of the article. Features that the tool could have is to list them in priority, articles with only one edit could have higher priority, also articles with no edits for 5 years could have priority ect. Maybe users can mark if the article is time sensitive, meaning it's about a person that is alive or a city ect. That way it can be checked again once all 55.000 articles have been checked. Is this a crazy idea? Is a similar tool available? Is anyone crazy enough to program it?
Steinninn16:15, 26 October 2022 (UTC)
I'm not a coding guy but I think this could be done as a bot+javascript task. A bot compiles list of all articles, their creation timestamp, last edit timestamp and with a column for verification timestamp. When, an article isn't verified at all or the last edit time is after verification time, it shows up as out of date. The "verification time" can be manually edited or updated via javascript. The list page can have appropriate protection level to ensure that they aren't vandalised. —CX Zoom[he/him](
let's talk • {
C•
X})17:53, 26 October 2022 (UTC)
This use case is kind of why
mw:Extension:FlaggedRevs exists and is deployed at a few wikis in its "review everything" mode. The wikis where it is deployed like that have issues keeping up, but it is an option and is basically already developed, it just needs deployment. It won't help review what's there today of course, but would take care of future changes to review.
Izno (
talk)
18:37, 26 October 2022 (UTC)
Thank you for your replies. It looks like FlaggedRevs is not being maintained. I will try this PetScan query, looks promising. Let me know if anyone has any other ideas. --
Steinninn18:59, 26 October 2022 (UTC)
PetScan's "Page properties" tab has a "Last edit" field, to limit the output to recently changed pages. It may be useful in future, after you have checked all pages once and want to look only at pages which have changed.
Certes (
talk)
19:26, 26 October 2022 (UTC)
I don’t have experience with wiki software, but it seems to me that there are many superficial similarities between MediaWiki and a multi-user transactional database system. If this is so, a back-end for what you are asking already exists (assuming pages are treated as records and page edits as record revisions), the related table is obviously live and accessible by database admins. If you ask kindly and persuasively, someone may decide to make a facility to extract a read-only subset available. I understand you don’t want a dump, but an updatable list of certain records (pages). I believe that is the most efficient way, although I have been involved in technology long enough to know it may be unattainable. Off-topic: Years ago, I had an idea to ask MediaWiki devs to include a Citation engine, a back-end in the core software itself. Seemed like a logical idea: if this is an anonymously edited, multi-user data entry system, it should have in its core some basic validation tools. A short time at VP and the related talk pages quickly disabused that notion. But who knows? You may have better luck.
65.88.88.93 (
talk)
19:39, 26 October 2022 (UTC)
If a wiki has few articles then it probably also has few editors and the average work per editor to check all articles may be the same or more than a large wiki. It doesn't sound realistic to me to check 55,000 articles. If other methods to distribute the work are too technical then the low-tech version is to use
is:Special:AllPages and make a coordination page where users can reserve an interval and report when they have completed or abandoned it. But it would be better for motivation and quality if users could choose topic areas to check, or individual articles.
PrimeHunter (
talk)
20:45, 26 October 2022 (UTC)
Toolforge database replication has stopped
Just a heads-up: the
replication lag to the toolforge database is currently around 1:30h, and increasing. That means that database reports etc. will be outdated until this is fixed. I'm assuming the database team is aware of this already, but pinging
Ladsgroup anyway, just in case. And maybe he can give an estimate to when it'll be resolved? --
rchard2scout (
talk)
15:13, 25 October 2022 (UTC)
Yes, now it is at 8:24. I always come here to check because I can never find my link to the replag page. Maybe it should be put in the FAQs at the top of this page. Half of the time I come to this noticeboard, it is about checking on the replag. LizRead!Talk!22:04, 25 October 2022 (UTC)
I think they meant that's how long it will take to fix, not how much of a lag it will go up to. But it is up to 29 hours now. LizRead!Talk!18:45, 26 October 2022 (UTC)
I don't know for sure but that would make sense. I'll stop keeping count. Just figure out how many hours between now and ~13:45 25 Oct. UTC. LizRead!Talk!20:49, 26 October 2022 (UTC)
It only happens if "Enable responsive mode" is enabled at
Special:Preferences#mw-prefsection-rendering and the window is narrow enough to trigger the responsive mode in the screenshot to the right. It also happens on a desktop PC with a narrow window. I would first have tried to report it at
Template talk:If mobile where
DLynch (WMF) wrote: "I added a specific class to MobileFrontend that we could trigger from rather than having to guess based on Minerva".
PrimeHunter (
talk)
15:35, 27 October 2022 (UTC)
My goal is some very basic, ballpark descriptive statistics about how many refs are in the median/typical Wikipedia article (e.g., any non-redirect page in the mainspace). I expect the answer to be a single digit number. Unfortunately, my question doesn't appear to be amenable to a Quarry query, so I'm bringing it here for help.
WhatamIdoing (
talk)
23:56, 24 October 2022 (UTC)
Hello! After seeing a former admin (ClockworkSoul) return (because of a personal attacking vandal on their request for admin rights), I've been working on fixing all the lint errors on their user pages and relevant subpages. However, I"ve run into an issue while attempting to fix one of their subpages. On their subpage
User:ClockworkSoul/userpage/panel left, there's a line of code which is: width=95% align=center style="background-color: #F3DE94; text-align: left; border: 1px solid #CC9;". The issue comes with align=center, which according to
mw:Help:Lint errors/obsolete-tag should be replaced with style="text-align:center", however that causes issues as the text is already aligned to the left side with text-align:left. So does anyone know what I should replace align=center with? ―
Blaze WolfTalkBlaze Wolf#654513:35, 27 October 2022 (UTC)
Wikipedia talk:Linter is a good place to ask about Linter errors. I do not see any reference to align=center being a Linter problem on that Help page, but I may have missed it. There are no remaining Linter errors on
User:ClockworkSoul or its subpages except in one or two pages that have no transclusions. –
Jonesey95 (
talk)
17:44, 27 October 2022 (UTC)
Ah I didn't know that was a place I could ask. Thanks! Will ask there, and it talks about "align-center" being a linter problem where it says "Table cell horizontal alignment" however it says that Linter doesn't detect it. ―
Blaze WolfTalkBlaze Wolf#654517:59, 27 October 2022 (UTC)
Comment: It was noted on its talk page, the article was getting
Post-expand include size too large warnings in February. The article is now at over 235K bytes. I assume it's the extensive use of
Template:Graph:Chart that's bogging it down. It loaded fine (albeit slow) for me, but that page is pretty big. --
DB1729talk08:13, 27 October 2022 (UTC)
The table with daily metrics is transcluded. Make those weekly or monthly metrics and I expect that will fix the issue.
Wikidata items can now be linked to Wikipedia redirects, solving the "Bonnie and Clyde" problem. When editing the "Wikipedia" section, type "en" and the redirect title in the usual way then click the rosette(?) icon to the right to add an "intentional sitelink to redirect" badge, and the "✓ publish" button will be enabled. This should only be done in accordance with Wikidata's rules: basically, when the redirect matches the Wikidata item but its target doesn't. Further information:
phab:T54564.
Certes (
talk)
09:50, 28 October 2022 (UTC)
speech recognition
Apache OpenOffice doesn't provide this feature. If Wikipedia was able to integrate something similar to Microsoft 365 Dictate, the the users' productivity would be highly speed up. Is it possible?
87.8.120.3 (
talk)
19:32, 28 October 2022 (UTC)
Technically yes, however we can't trust voice-recognition software to correctly interpret the speaker's words (it's a lot better than a few years ago, but I would not trust it to edit on Wikipedia). ―
Blaze WolfTalkBlaze Wolf#654519:36, 28 October 2022 (UTC)
@
JoeNMLC: The default in {{Bump}} is 30 days. On 11 September you added (via template substitution) [[User:DoNotArchiveUntil]] 14:52, 11 October 2022 (UTC).
[9] The archive instructions on the page say algo = old(7d) so the section could be archived 7 days after 11 October. It was archived 19 October
[10] so it's working as intended. You could have added a longer or new delay if you still didn't want it to be archived.
PrimeHunter (
talk)
21:13, 28 October 2022 (UTC)
Hi, is there a tool on here which can be used to reverse order a list, his filmography needs to be put in chrono order in the director and screenwriter sections but it's backwards. Can somebody do it? Gracias. ♦
Dr. Blofeld12:41, 28 October 2022 (UTC)
I don't think this applies for all people, but when I go into the Velociraptor article on my computer, the image shows a
Jurassic World image of Velociraptor which is not what the file is about. It's weird considering that some people do not experience the same issue, and my phone also don't have this issue as well.
Junsik1223 (
talk)
12:14, 29 October 2022 (UTC)
Has anyone else noticed that the font size for edit previews in Vector (2022) seems to have increased to be larger than the normal size? Does anyone know what might be causing that?
Graham (
talk)
04:40, 30 October 2022 (UTC)
Template:FlagIOC between Malay Wikipedia and English Wikipedia
Hello, I'm not sure where I should address this problem because our community at Malay Wikipedia can't solve it. There is a problem with the module in this template on the Malay Wikipedia. The country flag changed to a new one, instead of an old one in the articles in Malay Wikipedia. The flag used should be the same as in the English Wikipedia, but it is different in the Malay Wikipedia. An example is the Canadian flag in this article;
Malay version for1924 Winter Olympics.
@
CyberTroopers: Sorry, I misread the article code, and it took me a while to track down the actual problem. The issue is the regular expressions in the flagIOC function at
ms:Module:Country alias. I think the following will fix the problem:
Perennial topic, I know, but is there an alternative to Dabsolver? I know people moved from Dispenser's original site to using
[11] but now that version of Dabsolver has a error connecting to a MySQL server.
Nthep (
talk)
15:43, 31 October 2022 (UTC)
The software used for dabsolver is not freely-licensed, which is what more or less caused this issue where dispenser is off on his own website/location.
My understanding is that there was a secondary issue of the size of the data the tool needs to operate, but this is from memory and I may be confusing it with
BamBot.
Izno (
talk)
18:56, 31 October 2022 (UTC)
Tech News: 2022-44
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you.
Translations are available.
Recent changes
When using keyboard navigation on a
Kartographer map, the focus will become more visible.
[12]
In Special:RecentChanges, you can now hide the log entries for new user creations with the filter for "⧼rcfilters-filter-newuserlogactions-label⧽".
[13]
Changes later this week
The
new version of MediaWiki will be on test wikis and MediaWiki.org from 1 November. It will be on non-Wikipedia wikis and some Wikipedias from 2 November. It will be on all wikis from 3 November (
calendar).
The
maps dialog in VisualEditor now has some help texts.
[14]
It is now possible to select the language of a
Kartographer map in VisualEditor via a dropdown menu.
[15]
It is now possible to add a caption to a
Kartographer map in VisualEditor.
[16]
It is now possible to hide the frame of a
Kartographer map in VisualEditor.
[17]
I would like to receive notifications when The Signpost is posted. This would be better than talk page messages that I then have to delete or Watchlist changes that I usually don't look at.
Lectrician1 (
talk)
16:09, 31 October 2022 (UTC)
Having a quick look at
T119083 (and the phab board in general) suggests there's at least
some outstanding work to be completed. Although already in production, it may be worth filing a task there to see if the developers of that extension consider it suitable for deployment to the English Wikipedia
—
TheresNoTime (
talk • they/them)
16:15, 31 October 2022 (UTC)
Here's what I do: find somebody who subscribes, and watchlist their user talk page. This works for other newsletters too: by watchlisting this page, I get the Tech News without needing to have it on my user talk page. --
Redrose64 🌹 (
talk)
23:44, 31 October 2022 (UTC)
I recently encountered a bug [18] while adding footnotes to a table I am working on. I combined between VE and source for stylization and when I published the edit the footnote was missing. Cleared my cache several times as well as deleted and re-added the table, but to no avail. This I have no idea why is happening. Said efn is in the source but couldnt't figure how to have it display during or after saving to save my life. Hopefully, this minor bug, for the lack of a better term, gets fixed. Best,
Qwerty284651 (
talk)
04:08, 1 November 2022 (UTC)
@
Qwerty284651: I think you can't use a number ("2001") as the name for the footnote. I edited (but didn't save) the version of the page you linked to and changed the name to "2001x" and the footnote appeared.
DH85868993 (
talk)
04:18, 1 November 2022 (UTC)
Note that identifiers used in the name attribute require alphabetic characters; solely relying on numerals will generate an error message. The quotes are optional unless the name includes a space, punctuation or other mark. It is recommended that names be kept simple and restricted to the ASCII character set.
{{ec}}WP:REFNAME describes the syntax and behavior of the name=... portion of references. It notes: "Names must not be purely numeric; the software will accept something like ":31337" (which is punctuation plus a number), but it will ignore "31337" (purely numeric)."
DMacks (
talk)
04:30, 1 November 2022 (UTC)
Did you use the reference editor in VE ? If so it would be helpful if it validated the name for this condition. And that should be pretty easy to implement if it did not validate. —
TheDJ (
talk •
contribs)
09:27, 1 November 2022 (UTC)
As mentioned previously, I switched both VE and source in the same edit session, to place multiple efn's with the same efn name param. I guess I overlooked the invalid name warning, for some reason. :shrugs
Qwerty284651 (
talk)
16:41, 1 November 2022 (UTC)
@
Jéské Couriano: "PermaLinks" only faithfully reproduce the wikitext of a page as of a certain revision, not the state of the rendered page as of that timestamp. The link you have above for example has 72 templates/modules/other transclusions on it - which would import their current states. —
xaosfluxTalk20:15, 1 November 2022 (UTC)
Error
Our servers are currently under maintenance or experiencing a technical problem. Please try again in a few minutes.
See the error message at the bottom of this page for more information.
If you report this error to the Wikimedia System Administrators, please include the details below.
Request from [IP address redacted, I'm in Denmark] via cp3053 cp3053, Varnish XID 990781289
Upstream caches: cp3053 int
Error: 429, Too Many Requests at Thu, 27 Oct 2022 20:42:22 GMT
Hmm. My error message details are as follows, in case it's useful: Request from 199.208.172.35 via cp1080 cp1080, Varnish XID 530773433 Upstream caches: cp1080 int Error: 429, Too Many Requests at Thu, 27 Oct 2022 20:55:15 GMT.
At the help desk, another user posted: Request from <ip-address redacted> via cp3055 cp3055, Varnish XID 745869069 Upstream caches: cp3055 int Error: 429, Too Many Requests at Thu, 27 Oct 2022 19:01:45 GMT199.208.172.35 (
talk)
20:59, 27 October 2022 (UTC)
I havent checked, but the original is probably broken and the png scaler became more strict. Running it through pngfix tool and reuploading is probably the way to go. —
TheDJ (
talk •
contribs)
14:51, 28 October 2022 (UTC)
Noting that while the image displays fine for me at 40px, the link you gave throws:Request from 92.220.8.171 via cp3057 cp3057, Varnish XID 513967930Upstream caches: cp3057 intError: 429, Too Many Requests at Fri, 28 Oct 2022 18:52:06 GMTThis is the same if I find the image source with inspect element and navigate there. Purging my browser cache does not seem to change anything. Also,
Convenient Discussions is malfunctioning when trying to reply to this thred. No idea if related. ■ ∃
Madeline ⇔ ∃
Part of me;18:59, 28 October 2022 (UTC)
Oh, I may understand the difference now. MediaWiki uses the srcset attribute
[19] in the generated HTML to offer an image at other sizes so the browser can choose what to show.
File:MusicalnotesFinland.png is so small (79px) that the largest srcset size 2x is the original unscaled image for sizes at least 40px. The HTML in {{Finland-musician-stub}} says: <a href="/info/en/?search=File:MusicalnotesFinland.png" class="image"><img alt="Stub icon" src="https://upload.wikimedia.org/wikipedia/en/thumb/d/df/MusicalnotesFinland.png/40px-MusicalnotesFinland.png" decoding="async" width="40" height="26" srcset="https://upload.wikimedia.org/wikipedia/en/thumb/d/df/MusicalnotesFinland.png/60px-MusicalnotesFinland.png 1.5x, //upload.wikimedia.org/wikipedia/en/d/df/MusicalnotesFinland.png 2x" data-file-width="79" data-file-height="52" /></a>. Maybe nobody is actually seeing the 40px version and The DJ is right that MediaWiki is simply unable to scale the original. Compare 39px and 40px . I don't see either but does anyone see an image at 39px? There the largest srcset size is a scaled 78px while the largest for 40px is the unscaled original 79px.
PrimeHunter (
talk)
20:26, 28 October 2022 (UTC)
For the record, I noticed this the other day with an italian stub template so it appears that the issue is not confined to Finland. I can see the Finland images at 40px and larger in preview. Before I started this comment, all are broken except the unscaled. Here are Italy, Brazil, and Japan:
– 30px
– 40px
– 50px
– 60px
– 70px
When I previewed all of these, none worked until I added an unscaled row . Then, in preview, all worked except 30px Italy. Removing the unscaled row, in preview all work except 30px Italy. Don't know what will happen when I save this comment.
The ones that I see correctly are: Finland - 79px; Brazil - 50px, 60px, 70px; Japan - 40px, 50px, 60px. The others (including all of the Italy ones) show a file link only (the text of this link is definitely coming from the alt= attribute of the <img /> tag), and the problem images are consistent regardless of whether it's a preview or a saved edit, and edits by other people make no difference, nor does a
WP:BYPASS so it's not images cached on my machine. Perhaps the problem is an incompatibility between image file format and browser? I am using Firefox 106.0.2. --
Redrose64 🌹 (
talk)
12:34, 29 October 2022 (UTC)
Since this is a problem with old PNG headers, another way to fix this PNG thumbnail problem, when you encounter it, is to download the image, open in mspaint (or paint.net to keep transparency), save as a new png file, and upload as a new version. It takes hours to take effect on wiki pages containing the image, though.
Wikipedia:Village_pump_(technical)#PNG_thumbnail_problemMichael7604 (
talk)
18:17, 31 October 2022 (UTC)
@TheDJ: You are correct; pngfix complains about a bad zlib header (causing a too-far-back error). I have uploaded a corrected version.
BentSm (
talk)
04:33, 30 October 2022 (UTC)
PNG thumbnail problem
c:File:Manchot empereur carte reparition.png, a range map image that used to be on
Emperor Penguin, for some reason only displays as a broken inline image. Trying to load the image file gives
an odd error message: "Request from [IP] via cp3053 cp3053, Varnish XID 751884773 / Upstream caches: cp3053 int / Error: 429, Too Many Requests at Sun, 30 Oct 2022 14:51:22 GMT". The same problem arises from any specific thumbnail size - it's not just the default 220 - but loading the image unscaled seems OK.
I can confirm that the image has the same issue as mentioned there (the 'too-far-back' issue). I'm guessing that something changed with the MediaWiki software (or some part of it) that is causing it to refuse to scale images with the 'too-far-back' issue (which, as I understand it, was caused by a bug in old libpng versions). I could fix this image, too (and am willing to, if desired); however, I don't think that finding (somehow) and manually fixing all affected PNGs is a workable solution over all of Wikipedia... A bot to find and fix such images or a change to the software seem like the only sensible options here. (Also, the 529 error does seem rather a strange way to indicate that the image has issues.)
BentSm (
talk)
00:58, 31 October 2022 (UTC)
It looks like another way to fix these png thumbnail issues as you come across them, if you don't have technical knowledge of PNGs, is to download the image, open it in mspaint (or paint.net/GIMP to keep transparency), then save as another PNG, then upload that as a new version. I reverted my reupload because I didn't see the change on this talk page immediately, but looking at the version history on the Wikimedia Commons page I see that it did work.
Wide image of "Version history of emperor penguin range map"
Here's what the problematic PNG images displayed as before, for posterity.
And now that I reverted yet again to the 117 kb version that appeared to give good thumbnails, it has the thumbnail problem in the version history again. What pngfix program did you use to fix it?
Michael7604 (
talk)
03:28, 31 October 2022 (UTC)
pngfix comes with libpng (as of version 1.6.3 or so). (It is not the first few items one sees (assuming that you see what I do) when you Google "pngfix".) As far as the thumbnail issue, it indeed takes some time for the thumbnails to be generated. — Preceding
unsigned comment added by
BentSm (
talk •
contribs)
21:42, 31 October 2022 (UTC)
Hello. I have discovered a disturbing trend, and loophole in modern Wikipedia where templates can be used with impunity to circumvent our policies and procedures simply because they are not moderated or don't go through any kind of vetting process like our articles or policies do. Why is this so, and why are template creators/editors given such enormous freedom over something that has so much control over the actual operations of how our policies and procedures are technically enforced or enacted? I first noticed the problem
here, but the creator has since recognized the issue and modified the template so that it is compliant before the discussion has even completed. I then stumbled onto
this notice where it forbade red links in conflict with the
common list selection criteria #1 guidance that specifically allows redlinks. The wording was later tweaked to make it appear the conflict does not still exist. More recently, and probably most importantly, I discovered
this notice at:
Wikipedia talk:Inline citation, where there appears to be a built in form of control over the page with tricky language that can easily be misinterpreted thus enabling a type of local consensus and
WP:Ownership to occur. For example: this part of the notice; Please use this talk page to discuss issues specific to this information or how-to page—issues related to the consensus described here should be raised on the talk page of the associated policy or guideline (typically identified at the top of the information page)., seems to have been
recently interpreted to mean that any consensus for change on [to] the information page needs to be done at the policy level, and if that was not the intended meaning of the interpretation, then the problem still exists with this notice that it very well easily could be. I think we need to talk about having some kind of checks and balances in place for all templates and all template editors or even maybe some type of maintenance system to catch this stuff because simply just running into these by accident one by one is not cutting it. Thanks.
Huggums537 (
talk)
11:04, 30 October 2022 (UTC)
Good point. Perhaps fortunately, I can't remember the page titles, but I've come across a few edit notices which basically said "I own this page; here a a list of
LOCALCONSENSUS rules I invented for you to follow". Edit notices have their place, and warnings are sometimes appropriate. However, these notices can be hard to correct unless one is (or has the ear of) a template editor, and their wording may suggest that anyone attempting to amend them is heading for a fight.
Certes (
talk)
13:14, 30 October 2022 (UTC)
If you see a problem with a specific edit notice template, you should raise it on the talk page of the page it applies to or at an appropriate wider forum. Just like any other page, and as seems to have happened at
Template talk:People from place list criteria. About the only thing different about edit notices is that they're protected by default so you can't just fix them if you're not a templateeditor.As for the edit notice for
Wikipedia talk:Inline citation, I'd have to agree that policy should be argued at the policy page rather than an essay that's supposed to summarize it. The "recently interpreted" you link to seems like avoiding your attempted
WP:FORUMSHOP to me.
Anomie⚔14:13, 30 October 2022 (UTC)
I think you have misunderstood the post, and I modified it to correct the misunderstanding. Nobody is saying the edit notice is wrong for saying policy should be argued at the policy page. That is obvious. It doesn't even need to be said. Of course policy should be argued at the policy page. There's only one reason for even mentioning that in the first place, and it isn't good. The problem is with the way the last part is worded where it could be misinterpreted, and used to mean that edits to the information page should be taken to some policy page rather than being discussed there where they should be.
Huggums537 (
talk)
23:45, 30 October 2022 (UTC)
One idea I thought of to hold template editors accountable and make these notices transparent to everyone is to require that all edit notices be placed into the edit notice templates category so that they can be easily identifiable and monitored by any editor who cares to refer them to an admin or template editor if they require modification or review. As I explained earlier, this idea of one by one dealing with the problem by raising it on a talk page if you just accidentally happened to see a problem is probably the most inefficient way to reduce the problem. Also, if you try to deal with it by looking at existing categories it will fail since most templates seeking to control pages through stealth technical means most likely won't be making themselves easy to discover by announcing their presence in any categories...
Huggums537 (
talk)
22:28, 30 October 2022 (UTC)
Thank you for that very helpful link! However, not all of the problematic templates I mentioned appear in that list, nor do all the problematic ones that exist would appear either.
Huggums537 (
talk)
23:58, 30 October 2022 (UTC)
I have a comment, and it's that, for the most part, template editors don't come up with the text at issue, they simply add it (as an edit notice) at some user's request. The real issue is the general "some editors need to be educated about policy". I guess it's unfortunate that sometimes template editors don't catch the problem themselves. --
Izno (
talk)
23:00, 30 October 2022 (UTC)
I think regardless of where the source of the text stems from, edit notices, and really all templates should have a way where they are easily sortable, and available for community review. My idea about a mandatory requirement for placing edit notices into a category should probably extend to most all templates since edit notices are not the only ones that contain errors where editors who need to be educated about policy have inserted text through template editors who have not caught the problem themselves. If we had a mandatory category or a few subcategories where the whole community could assist these template editors in reviewing the text, then the problem would greatly be reduced.
Huggums537 (
talk)
23:25, 30 October 2022 (UTC)
This request is nothing more than forum shopping. The first example is not a problem since, as the creator says, it was resolved ages ago. The second example was kept at TfD, so is by no means unmoderated or a defiance of consensus. The third example had a discussion at
Wikipedia talk:Project namespace/Archive 1#Edit notice for information talk pages and furthermore the editnotice itself is not the problem since every other human taking place in that discussion agrees it's in the wrong place. Perhaps the reason you keep running in to editnotices you think are problematic is because the rest of the community does not agree with you.
* Pppery *it has begun...23:44, 30 October 2022 (UTC)
All of the templates I mentioned are exactly what you said they are; first, second, and third examples of my previous experiences with edit notices. If I can't share my past experiences (just to demonstrate edit notices have in fact been a real world problem) without being accused of forum shopping for doing so, then maybe you are the one who should reconsider yourself. The first example is no longer a problem because the community did agree with me and corrected itself accordingly, and Certes also mentioned running into problematic templates as well in addition to Izno being in agreement that there are textual problems related to policy errors that are not caught by template editors. I should also like to point out that you failed to recognize the fact that in spite of the deletion discussion ending in "keep" for the edit notice in example 2, the community still acknowledged my concern there was a conflict in the guidance and agreed to correct that problem as evidenced by the diffs in my post, even if I didn't think it was the best solution. So, your idea that perhaps the reason I keep running into editnotices I think are problematic is because the rest of the community does not agree with me might perhaps be just as misplaced as bad faith suggestion I'm forum shopping for simply recounting my past experiences to use as examples of an ongoing problem that others have also agreed is apparent.
Huggums537 (
talk)
00:36, 31 October 2022 (UTC)
Also, I feel like the link you provided to the discussion about the information page regarding the third example is a perfect specimen of the exact type of local consensus I'm trying to prevent with these edit notices. The fact you are presenting it here as if it were any type of strong evidence there was a consensus for that edit notice is a prime reason why these types of localized discussions where an extremely limited number of editors are supposedly representing the whole community are more part of the problem than part of the solution, and the so called consensus
at the TfD you mentioned was not all that much better.
Huggums537 (
talk)
01:23, 31 October 2022 (UTC)
Another idea I had was maybe some kind of project page where editors interested in helping each other solve this problem could share techniques, and resources with each other, especially with newer, or less advanced users to empower them with options to help with this problem when they come across it and they know something is wrong, but maybe just not sure what to do about it or where to go from there. It would also be helpful for advanced users to share techniques among each other to broaden the collective knowledge of ways how best to resolve the issue. It would also create a central hub of activity and awareness about the issue as well.
Huggums537 (
talk)
23:15, 1 November 2022 (UTC)
{{
Infobox magazine}} uses {{
ISSN link}} to render a url for an ISSN search at
WorldCat. Because WorldCat returns the message that you are seeing, it would appear that WorldCat hasn't got a record of that ISSN in its database.
It is unusual, but not unheard-of for periodicals with assigned ISSNs to not appear at the OCLC catalog. Usually they are either brand-new assignations, or very limited circulation, or fringe items. This is a Russian publication. Worldcat has over 15000 Russian member institutions, but apparently none are carrying it (search by title brings up nothing for me). Perhaps because this seems to be a magazine on UFOs, magic and mysteries? (from a Russian resale dealer page and according to Google Translate). This would fall under "fringe". Infobox usage is ok, but I would be careful in citing this, although anything can be cited depending on the context.
65.254.10.26 (
talk)
00:36, 2 November 2022 (UTC)
Hi, to synonym two names in an Infobox, we can use {{{firstName|{{{secondName|}}}}}} as in {{{developer|{{{owner|}}}}}} in
Template:Infobox file format. This technique should implement this scenario:
if (firstName exists) then
show firstName
else
show secondName
But the problem arises when the firstName parameter exists but has null Value, and secondName has another non-null value. This way, the if satement returns true and then in this branch we should show null value in Infobox but the Infobox rendering algorithm shows nothing in the final rendered Infobox.
This scenario is not correct. For example "owner" and "developer" parameters are synonym by this code
| data10 = {{{developer|{{{owner|}}}}}}
at line 26 of
Template:Infobox file format.
Here the bug arises when the "developer" parameter exists but has null value. For example, consider this usage for the above example:
{{Infobox file format
| name = Scalable Vector Graphics
| icon = SVG Logo.svg
| icon_size = 200px
| owner =[[World Wide Web Consortium|W3C]]
| developer =
}}
This null valued scenario for the "developer" parameter produces:
Scalable Vector Graphics
Which developer value (or its synonym owner value) is not shown in the final rendered Infobox. But if we remove the "developer" parameter from the above codes, then the developer value will be rendered truly by the "owner" parameter. So, please correct this "null valued bug scenario". The correct scenario for {{{firstName|{{{secondName|}}}}}} is
if ((firstName exists) and (firstName ≠ null)) then
show firstName
else
show secondName
@
PrimeHunter Yes, the usual scenario (using only one alias) works well, but in an unusual scenario as I mentioned above, the result is unusual too. In my opinion making this {{#if: statement implicit for {{{firstName|{{{secondName|}}}}}} is more reasonable. For example, in Java or C++, if(null) statement always returns false, conversely to here that it returns true. I.e., it automatically checks that firstName is not null, or equivalently, the second if i.e., (firstName ≠ null) is implicitly checked automatically. Doesn't this action make coding in Wikipedia more convenient?
Hooman Mallahzadeh (
talk)
09:17, 2 November 2022 (UTC)
@
PrimeHunter This is a bug of system, because the expected behavior for a normal person is:
The statement {{{firstName|{{{secondName|}}}}}} ::should mean:: Show the firstName if has value otherwise show the secondName
I have run into this MediaWiki quirk with some frequency while modifying and troubleshooting infoboxes. I have resigned myself to putting in if statements when I run into this problem. If you have the energy, file a bug in Phabricator. I think it would be great (and logical) to have {{{firstName|{{{secondName|}}}}}} return the populated value of |secondName= if |firstName= is present but blank. –
Jonesey95 (
talk)
14:59, 2 November 2022 (UTC)
Blank is a value like any other. What if you want the blank value? If firstName is blank then should {{{firstName|{{{secondName|}}}}}} be secondName if it's set but blank if it's not? That seems inconsistent. And what about default values which are not parameters like {{{year|{{CURRENTYEAR}}}}}. Should that be CURRENTYEAR if you intentionally give a blank value? The current method {{#if:{{{firstName|}}}|{{{firstName}}}|{{{secondName|}}}}} is rarely coded, probably because it's much harder to write than {{{firstName|{{{secondName|}}}}}}. Maybe it would be best to just have an easier way to do the former, like an extra character somewhere in {{{firstName|{{{secondName|}}}}}} to specify that secondName should also be used if firstName is blank.
PrimeHunter (
talk)
16:03, 2 November 2022 (UTC)
I'm currently away from home, and here I have no network for most of the time. I'll be back home few days later, and by that time, I'm pretty much sure that my watchlist would have more than 500 unread entries. Problem is that the watchlist settings only allow a maximum of 500 entries in it. Is there a way to increase that limit so I don't miss anything potentially important when I get back to it? Thanks! —CX Zoom[he/him](
let's talk • {
C•
X})09:38, 31 October 2022 (UTC)
Besides the noted preference to increase the number, when I go on vacation I usually work through my watchlist by namespace.
Izno (
talk)
18:31, 31 October 2022 (UTC)
In case Izno's note is unclear, you can select a namespace at the top of the Watchlist page, which will show you 500 changes (or your preference) just in that namespace. It works well. One thing that helps me after short wikibreaks is to temporarily remove busy pages (like VPT) from my watchlist when I return. I read those pages separately, then re-add them to my list when I am caught up with everything else. –
Jonesey95 (
talk)
20:14, 31 October 2022 (UTC)
Don't forget that the watchlist has an unseen changes filter, you can force it with watchlistactivity=unseen.
RAN1 (
talk)
21:17, 31 October 2022 (UTC)
Extending Izno's post — You can also filter the watchlist to only include the main namespace. Save that as the "Main" filter. Then enable the main namespace again and disable all of the other namespaces. Save that as the "Other" filter. Now open two tabs in your browser. Load "Main" into one tab and "Other" into the other tab. You can also save differently filtered sub-views as bookmarks and open each bookmark in a different tab. Better still, save each watchlist subset as a bookmark in a specific group, then open the whole group in a new window —
GhostInTheMachinetalk to me22:44, 31 October 2022 (UTC)
- {{AFC submission}} is good at catching drafts that are already in the pending submissions queue with premature categories on them, but seems to be inconsistent in catching drafts that haven't actually been formally submitted for review. ‹The
templateCategory link is being
considered for merging.›Category:AfC submissions with categoriessometimes catches drafts where the AFC submission box is grey ("Draft article not currently submitted for review") instead of yellow (awaiting review), but sometimes does not -- today alone, I've caught numerous categorized drafts which have existed for weeks or even months as unsubmitted but categorized drafts, with the grey AFC box on them, without ever having shown up in the cleanup category. (Examples if needed:
Draft:Indianapolis 8 Hours (created October 8),
Draft:Hans Geisler (Politician) (categories added to it on September 28),
Draft:Fox Netball (categories on it had been disabled, but then got re-enabled on September 10). But, of course, since the rule about not categorizing drafts as articles applies the same way regardless of whether the draft is in the pending submissions queue yet or not, that template needs to catch categorized drafts without regard to whether the box is yellow or grey. So could somebody check the AFC submission template to see whether its code needs to be revised to make it more consistent in catching greyboxed drafts?
- With the increasing reliance on moving underdeveloped articles into draftspace as an
alternative to deletion, there's also been a massive uptick in users who do that without removing or disabling the categories that were on the page as an article. These, similarly, can end up sitting in mainspace categories for weeks, depending how long it takes for somebody to catch them, because there's no dedicated tool to find them without coming across them organically — so I wanted to ask if it would be possible to add category-detecting code to {{Drafts moved from mainspace}}, similar to what {{AFC submission}} already uses, so that moved pages which haven't already had their categories turned off can get added to a maintenance queue (either ‹The
templateCategory link is being
considered for merging.›Category:AfC submissions with categories or a similar new dedicated maintenance category) for quicker attention?
Item 2 cannot be done in the way you requested. What might be feasible with a
query of some sort would be to find draft space pages not in some set of management categories, ignoring their membership in hidden categories.
Izno (
talk)
23:13, 3 November 2022 (UTC)
I was also going to suggest a regularly updated database report based on a query. Someone with the necessary skills should be able to set one up. I don't know much about draft space, but the query logic could be as simple as "show all pages in Draft space with categories, with the exception of the following categories: X, Y, Z". Human editors would need to go through and remove the cats, or maybe a bot could be programmed to do so. –
Jonesey95 (
talk)
23:24, 3 November 2022 (UTC)
Here's a query; there's currently 66 drafts with 83 total offending categories. The fancy new {{database report}} template should be able to translate that into something with clickable links and that gets automagically bot-updated on a subpage. (I suppose I should look into how it works.) —
Cryptic04:20, 4 November 2022 (UTC)
I checked the bot source (surprisingly easy to read, for a programming language I don't know!), and it's supposed to add the colon to Category: and File: links. So I added the link to the categories, and the bot updated the page, and it worked! So now it's also to see which categories are redlinks. --
rchard2scout (
talk)
09:05, 4 November 2022 (UTC)
I've noticed that for the last couple of days when I go to my Watchlist there's a notification that "The new issue of The Signpost is out now". I "dismiss" it (as usual, without following the link to the Signpost), but if I close my browser and re-open it, the notification is there again. Previously I could dismiss it and I did not see a notification until there was another edition. So far as I know, there have been no changes to my account or browser (
Pale Moon on Windows 7). Is it just me, or is there something wrong with the notification system?
Mitch Ames (
talk)
01:09, 5 November 2022 (UTC)
Dismissals are stored in your browser's localStorage, presumably that's being cleared or reset somehow whenever you restart your browser.
Legoktm (
talk)
01:42, 5 November 2022 (UTC)
Thanks. I checked all my browser profile settings but couldn't find anything wrong, so I deleted caches, deleted history, deleted stored data etc, tried again, still not working . So I reverted to a backup copy of the profile from a month ago, and now it is working again. Manually comparing settings in the two profiles I still can't see any difference, but presumably there must have been a corrupt file somewhere. Hooray for backups.
Mitch Ames (
talk)
06:10, 5 November 2022 (UTC)
How can I replace my sandbox shortcut in the top tab navigation menu with my /subpage page and have a shortcut attached to it: let's say Alt + Shift + O? I am far more active on my /subpage, have the default sandbox for projects put on the backburner.
Qwerty284651 (
talk)
03:39, 3 November 2022 (UTC)
@
Qwerty284651: There ought to be support for this in the PHP, because doing it in JavaScript is hacky and the text change can be distracting, but here is something you can add to
Special:MyPage/common.js:
In spite of searching the archives, I cannot believe this hasn't been asked. Maybe I missed it.
If I make a mistake in my edit summary, anything from a minor typo to a missing or extra word that completely changes the meaning and intent of the summary, I would like to have the ability to correct it. Perhaps such an ability could be granted to extended-confirmed users or administrators, to edit one's own summary and nobody else's.
I know I can "correct" an error in a summary by making a dummy edit with a corrected summary, but this seems like a kludge.
A larger concern came from an incident that just happened today, when I blocked a user and made a mistake in the block summary (see
block log). I had to re-block with a correction. But now that user account has two blocks on its record instead of one. That isn't really a problem for a sockpuppet account, but it would be a concern for a legitimate account to have an extra block that doesn't need to be there.
There's also the issue of when making a dummy edit, the fixed edit summary is usually lost in the undoing of the dummy edit. ―
Blaze WolfTalkBlaze Wolf#654520:23, 4 November 2022 (UTC)
That was something I had actually stated when Anachronist brought it up on the Teahouse. I had stated that issue would be it's basically
paradoxicalthat's probably the wrong word to use, but it's basically a
fractal as it can basically be an infinite loop of editing edit summaries (i.e you need to edit the edit summary that is editing the edit summary of your edit summary that is editing your edit summary of the edit summary that is editing the edit summary....) and is practically impossible to do. Theoretically you could just make it so you can only edit the main edit summary and the edit summary of that edit summary would be replaced with your new one, but then there are issues with attribution. ―
Blaze WolfTalkBlaze Wolf#654522:02, 4 November 2022 (UTC)
This wouldn't actually be that hard to solve. Only allow edit to the summary within, say, the first minute of the edit being up. Would allow fixing typos without causing other issues. (I suppose you could run into the situation of someone using abusive ES's and quickly editing them out... but that's not really an effective method of abuse.)
Elli (
talk |
contribs)
22:06, 4 November 2022 (UTC)
That is actually a fair point. It seems that we do tend to overthink things like this. But again it's not possible in mediwiki currently. ―
Blaze WolfTalkBlaze Wolf#654522:08, 4 November 2022 (UTC)
It’s not just about abuse. It’s about transparency, having and keeping a consistent record. And we can correct edit summaries if really needed, because revisions can be deleted and edit summaries can be suppressed. So ultimately (as almost always), the question mostly becomes; who is going to put in a lot of work to make edit summaries editable, those edits transparent, show up into the logs and possibly even in watchlists and then make those suppressable etc etc etc. Or do these same ppl put their time into other? —
TheDJ (
talk •
contribs)
14:09, 5 November 2022 (UTC)
Yeah this isn't about it not being technically possible -- of course the software could be modified to allow editing of edit summaries. This is about it not being desirable. We want a log of all changes. So if edit summaries can be changed, then we'd need a log of changes to the edit summaries. And if the edit-summary-change-log could be changed, then we'd another log tracking the changes to the edit-summary-change-log, an edit-summary-change-log-change-log. This is what xaos means by "turtles all the way down". At some point, we need a log that cannot be modified.
Now, there is no reason that edit summaries (or, the other common example, block logs) must be the unchangeable log. We could just go one step down with the turtles, and have a log that keeps track of changes to edit summaries and block logs, and make that log unchangeable. That brings us to TheDJ's point: who is going to code this and do we really care enough to spend the resources?
In short: we need an unmodifiable log, edit summaries are that log, and while we could change that, there is more important work for devs to do than to make this change.
Levivich (
talk)
17:45, 5 November 2022 (UTC)
No, you don't need a log of changes to edit summaries. All that is needed is a small tag like (edited) appended to the summary to indicate that it has been edited by the author of the summary. That is sufficient for many other sites that indicate when a comment has been edited, and it should be sufficient here. I don't see what's undesirable about that, and I also don't see what's desirable about keeping a log of edit summary changes. ~
Anachronist (
talk)
22:46, 5 November 2022 (UTC)
It would be too easy to abuse if we allowed anything on this website to be changed without a record of the change (who changed it, when it was changed, and what changed). An edit summary on Wikipedia is not comparable to a comment on other websites.
Levivich (
talk)
01:00, 6 November 2022 (UTC)
Non-sequitur, and also assumes bad faith. There is no proposal here to allow just anybody to change any edit summary. Only the author of the summary should be permitted. If needed, it can be set up as a user right for extended-confirmed or administrators, still restricting it to being editable only by the author. It can also be time-limited. I've seen other websites that give you 5 minutes to edit something you posted. That could work here. The timestamp and time tracking would be internal and need not be exposed. ~
Anachronist (
talk)
02:28, 6 November 2022 (UTC)
It would be too easy to abuse if we allowed editors to change their own edit summaries within five minutes without recording what was changed. For example, edit summaries are sometimes sent as pings, and if they were changed after being sent, other editors wouldn't be able to see the original, which would make abuse harder to detect and prevent.
Levivich (
talk)
05:08, 6 November 2022 (UTC)
Well, a version that's not easy to abuse isn't inconceivable. Make the summary editable just once. Once edited, an (edited) indication shows up next to the summary on which if you hover, it shows the original summary. Revision deletion, suppression, etc would hide both original and revised edit summaries. –
SD0001 (
talk)
17:24, 6 November 2022 (UTC)
@
Anachronist: This discussion is sort of going in circles. The answer to your original question is (a) because the mediawiki software doesn't support this. If you would like to help update software that can do that, see
mw:How to contribute for how to get started on that. If such software gets created, it will likely be opt-in/permission-based, etc. At that time, a proposal to opt-in to it can be discussed here on the English Wikipedia. —
xaosfluxTalk17:34, 6 November 2022 (UTC)
Error 503
I frequently encounter "Error 503" when editing Wikipedia with Norton Smart Firewall installed on my computer. However, as soon as I deactivate Norton, editing becomes normal. Debug result points to proxy issues. Is there a fix?
RPSkokie (
talk)
16:48, 6 November 2022 (UTC)
Something removing ANI board from my watchlist & it's not me.
In the last few weeks. I've noticed that
WP:ANI was twice removed from my watchlist & not by me. Has anyone else, been having this problem?
GoodDay (
talk)
15:44, 6 November 2022 (UTC)
At Preferences →
Watchlist, do you have "Expand watchlist to show all changes, not just the most recent" set? If not, are any of the six "Changes shown" options set? If so, that may be the problem. For example, if the most recent edit was a bot edit, and you have "Hide bot edits from the watchlist" set, ANI won't be listed at all - a non-expanded watchlist won't back-track to the most recent non-bot edit. --
Redrose64 🌹 (
talk)
18:44, 6 November 2022 (UTC)
GoodDay should be able to eliminate one possibility by going into "Watchlist" → "View and edit watchlist" and searching for "Incidents".
Certes (
talk)
19:11, 6 November 2022 (UTC)
I put a check-mark into the box for "Expand watchlist to show all changes, not just the most recent". So far, the problem hasn't reoccurred.
GoodDay (
talk)
21:55, 6 November 2022 (UTC)
Do you mean that it was already set prior to 15:44, 6 November 2022 (UTC), or that you have set it since 18:44, 6 November 2022 (UTC)? --
Redrose64 🌹 (
talk)
23:12, 6 November 2022 (UTC)
All images on page
Gomal Zam Dam opens in MediaViewer for me, except the last image, an image in footer template. I saw a code of template and doesn't see any reasons for that. What's the problem?
MBH (
talk)
06:01, 6 November 2022 (UTC)
The navbox module adds noviewer to its images, which causes MediaViewer to ignore those images. This is by design.
Izno (
talk)
06:27, 6 November 2022 (UTC)
Media Viewer has arrows to browse through the images in an article. Navbox images were not selected for the article and often have low relevance (and low resolution) so they are omitted. The three portal icons in
Gomal Zam Dam#See also are also omitted
PrimeHunter (
talk)
13:42, 6 November 2022 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you.
Translations are available.
Recent changes
An updated version of the
Event Registration tool is now available for testing at
testwiki and
test2wiki. The tool provides features for event organizers and participants. Your feedback is welcome at our
project talkpage. More information about
the project is available.
[20]
Problems
Twice last week, for about 45 minutes, some files and thumbnails failed to load and uploads failed, mostly for logged-in users. The cause is being investigated and an incident report will be available soon.
If I save this Village Pump page on my hard disk, open it in Firefox and click the link then I stay on the pdf url and see the pdf file. If I click the link here at wikipedia.org or upload the saved page to my own website (not linking the page here for copyright reasons) then I get the redirect you mention. I guess the site uses
HTTP referer info to decide whether to serve the PDF or the redirect. It's their choice and I don't think we should try to bypass it when the redirect gives access to the PDF.
PrimeHunter (
talk)
01:03, 8 November 2022 (UTC)
If I go to
Cloud computing, I am unable to view the history of the page as that is completely obstructed by two icons an a link "Unaccepted". I am using Vector-2022 and have twinkle enabled. 0x
Deadbeef→∞ (
talk to me)
14:02, 8 November 2022 (UTC)
It positions for legacy vector on the title, but with the switch to tools below title on vector 2022 it covers the view history button. Simplest fix is to just change the top value to -0.8em for v22 or something but it isn't the best solution.
Terasail[✉️]14:53, 8 November 2022 (UTC)
Can someone update the articletalk TOC trigger to (2 or) 3 threads?
It's extremely annoying when there are three threads on an article talkpage, especially when they, or at least the first two, are long and involved. This means the third thread will very likely be completely ignored, because there is no TOC. Articletalk TOCs get triggered only when there are four threads.
How does this possibly make sense? We need articletalk TOCs when there are even two or three threads present. It's very difficult to attract participation to a thread at the bottom unless there is a TOC. I'm really tired of having to force a TOC (as I just did
here) in order to get one.
Since most articletalk pages are underused, this is getting more and more crucial. Especially when there are important discussions on the talkpage such as RFCs or Merge proposals or Split proposals.
It's also crucial when one wants to link to an articletalk thread from another page; this is easy when there is a TOC, very hard to figure out or do when there is none.
Could someone either please lower the TOC trigger to 2 or 3 threads, or make TOC default on all articletalk pages even if there is only one thread?
@
Softlavender: Yes, that's not what you asked. But IMHO it is an answer to problems you want to solve: it makes a talkpage navigable, it exposes the second and third thread when the first one becomes very long, and it adds a visible link to every thread, even if there's just one. Last but not least, you can easily apply it immediately at each page you need right now, without waiting for anybody to modify the Wikipedia configuration or even the Wikipedia software. Additionally, it won't affect talkpages which do not need it. --
CiaPan (
talk)
07:54, 9 November 2022 (UTC)
CiaPan, you clearly didn't read my post either the first time or when I specifically asked you to, so please do not respond to me again. Let someone who reads and understands my post answer.
Softlavender (
talk)
08:44, 9 November 2022 (UTC)
Contributions page jumps around while loading
I opened a user’s contributions page and clicked on a diff link that looked interesting. But the page was still loading stuff and it jumped around as it did so. Before I knew what happened, I notice that the link I tried to click on had been displaced by a Twinkle rollback link that proceeded to
revert an innocent edit without warning. I have reverted my edit (thus restoring the original innocent edit).
Since I am no longer active here, I have now disabled Twinkle. But that doesn’t fix the root problem of everything jumping around. Someone should work on that.
Brianjd (
talk)
07:02, 9 November 2022 (UTC)
Was this with the new vector or the old vector skin ? I noticed that the Twinkle gadget doesn't reserve space for itself properly yet on vector-2022 pages where it is the only portal bar item. —
TheDJ (
talk •
contribs)
09:17, 9 November 2022 (UTC)
It isn’t just Twinkle. The Contribsrange gadget (which I didn’t even remember was a gadget until I checked just now) doesn’t reserve space either, or at least it doesn’t do it properly.
I feel like the page jumped around more than that, but maybe I was just confused by Twinkle adding so many links at once.
Brianjd (
talk)
09:37, 9 November 2022 (UTC)
Yes this is what gadgets generally do when they modify the page. That's one of the big downsides of gadgets, and you can't really solve it (you can reserve some space, but its really bothersome to do).
BTW. I don't think most people realise this, but that contribsrange gadget isn't really needed any longer. Range searching is a built-in feature since 2017. The only functionality that contribsrange provides, is wildcard prefix searches, but most people don't actually use this part of it. —
TheDJ (
talk •
contribs)
10:17, 9 November 2022 (UTC)
String search for stripping with AWB
Hello. Please, How can I search the string "[[" in a generic text line starting with "{{nota disambigua2|" using regular expressions?
Pierpao (
talk)
15:01, 9 November 2022 (UTC)
@
Pierpao: I suspect you also want to find uppercase Nota: ^\{\{[Nn]ota disambigua2\|.*\[\[. Maybe you also want regex to remove brackets but if you want to achieve something then say what you actually want.
PrimeHunter (
talk)
16:01, 9 November 2022 (UTC)
Shortcut keystrokes
I use Safari 16.0 on macOS Monterey. The keyboard shortcuts for various interface items are usually engaged with the Control and Option keys plus the letter, so Control+⌥ Option+e to engage the edit button for a page. Last week, on Thursday, this switched to Control+e, and now this week on Thursday, it's switched back. This isn't the first time I've had this happen. It's usually only temporary, but it's very annoying to have to relearn the keystrokes only to have switch back just as I've gotten accustomed to the change.
Does anyone else have this problem?
Does anyone know why this switches back and forth?
This last time, it's been on Thursdays, so is this a
WP:THURSDAY issue?
Hi all! We wanted to give you a follow-up regarding the topic of including non-free images in the recently developed feature of article thumbnails on Special:Search.
As I
already said in one of my previous messages, since such images were sourced from the project and already used in other similar places as well (such as the Go bar autocomplete), it was assumed that they were ok to use. After the reactions we read in the previous thread, we see that the community believes the inclusion of those images does not comply with the
existing policy about non-free content on this project.
In light of the discussion, we decided to precautionarily disable fair use/non-free content for thumbnails in the search results and in the Go bar, to avoid any potential friction with the policy and the community. The change will go live in the next few days.
We are open to restoring this functionality at any moment, but we want first to have a conversation with the community about this topic. Before turning it back on, we want to be sure that all potential friction is clarified and a community consensus can be reached on the point.
Hello. I have a concern with the {{expand x}} templates as x seems to be a separate template. To me, this makes little sense and does not follow the commonly-used DRY (Don't Repeat Yourself by Repeating Yourself by Repeating Yourself) programming methodology. If a change to the whole template group is needed, in the WET environment here you have to change each template manually or put boilerplate after a few other templates.
This is obviously rather bad and is not a good thing and causes bad things to happen like what i'm about to say here in just a few secands I will say what bad things will happen: more maintenance and pain for users, even if only the language name is there. What if more languages exist then the current few? What if the template is deleted?
Instead I propose a better solution. Instead of having each template for a separate language, why not have the language, as an ISO code, as a argument. Why not try {{expandFrom|es}}? That way if the language name changes or there are more languages or you need to delete the template, it is quite easy to just remove or retrofit. 56independent/notacoworcatTalk17:43, 10 November 2022 (UTC)
It seems to require you to create a new template for each language. This has the same problems as I described. I wanted to see if turning the language into an argument would work. That way users don't have to play the template lottery when an obscure language has a better article. 56independent/notacoworcatTalk20:38, 10 November 2022 (UTC) — Preceding
unsigned comment added by
56independent (
talk •
contribs)
Umm ... didn't
Special:Search use to work differently? Now, if an existing article exists for a search term, it goes to the article instead of providing search results. For example,
Special:Search/Apple goes to
Apple instead of providing
these search results. I mean, I know going straight to the title match of an article/redirect has always been how Wikipedia's search bar works, but I thought it worked differently if
Special:Search was called directly with a "subpage". Am I mistaken? If not, I find it rather troublesome that there's either no way or a different way to use internal links to provide search results for a string that has an existing title.
Steel1943 (
talk)
22:03, 10 November 2022 (UTC)
Different wiki's (namely Fandom) handle it differently but on Wikimedia wiki's this has been the standard behaviour for the last decade and a half. Seddontalk00:12, 11 November 2022 (UTC)
No, because it is sent from Wikipedia, not from your email box. What you can do when sending an email in
Special:EmailUser is check the box that says "Email me a copy of my message." Then Wikipedia will send two emails, one to the user you are trying to contact and one to your email box.
Izno (
talk)
10:47, 11 November 2022 (UTC)
Charles III - signature shows black background on firefox.
For me, on
Charles III, both logged-in and logged-out, but only on Firefox, the signature in the infobox is displayed over a black background, making it impossible to read. It is basically just a black box with some smushes. Both Safari and Chromium render correctly, that is a black signature over the light blue background of the infobox. I don't know what causes it, but it is very strange. - I reported this on
Talk:Charles_III#Firefox:_Signature_has_black_background and was referred here. (Please note that I might not be able to reply to this in the next few days, depending on how my day goes tomorrow.) --
Lommes (
talk)
21:06, 10 November 2022 (UTC)
I have Firefox Version 106.0.5, Modern skin. There is no black background on mine. Could it be something under your personal "Preferences"?
— Maile (
talk)
21:30, 10 November 2022 (UTC)
Some systems can display transparent PNG images with a black background. It's generally preferable for images like this to be uploaded as SVG images.
Thebiguglyalien (
talk)
21:38, 10 November 2022 (UTC)
I'm getting the same effect when I tried it on FireFox, latest version 106.0.5, default settings for everything. I use Chrome normally and it's fine there. Bypassing the cache did nothing.
Jameel the Saluki (
talk)
22:25, 10 November 2022 (UTC)
Can report that I am also seeing it (Firefox 106.0.5, Ubuntu, logged out & private window so should be nothing unusual). Interestingly it is not showing with the weird black background when looking at the frwiki article in the same browser, although it also uses an infobox.
Looking at the code, the enwiki infobox seems to be loading a
125px thumbnail, and when that is opened separately it has the black background. But the frwiki infobox is loading a
110px thumbnail, which renders cleanly. Likewise, Wikidata shows it as a
220px thumbnail which also works fine. So maybe something to do with the way the thumbnails are being generated meaning whatever weirdness Firefox doesn't like is only introduced into some of them? I have a recollection that Commons pregenerates some standard thumbnail sizes (including 220px) but unusual-sized ones are done when first requested, so it's possible some discrepancy could come in there?
Andrew Gray (
talk)
23:50, 10 November 2022 (UTC)
Purged the Commons file and the 125px thumbnail is now showing correctly. Weirdly, Firefox recognized the previous 125px thumbnail as a
WebP rather than a PNG (I downloaded it before purging and it indeed appears to be a WebP despite the extension).
Nardog (
talk)
00:24, 11 November 2022 (UTC)
Could you please elaborate. I am uncertain what you mean by "thumbing system". Are you saying the problem lies in Firefox? All of the thumbnail sizes described by Andrew Gray work fine, whereas the
Charles III signature is still failing. It seems to me that if this is a problem that can be fixed by Wikipedia then it should be.
Jameel the Saluki (
talk)
01:17, 11 November 2022 (UTC)
No, the problem is with MediaWiki. We don't know why it happens as much as it does. I'm sure there's a task on
phab: if you want to hunt. --
Izno (
talk)
04:28, 11 November 2022 (UTC)
Disable images in search results, only partial implementation
At
Phab:T320337 they have deployed a preference for logged-in users to always not load images in Special:Search, but have declined the other requested aspects: on-page toggle, persistence for logged-out users, and URL parameter. How do we get this decision re-considered? and is there strong community support? (I suspect re-opening a ticket is not viewed favorably, and a new one may just be closed as duplicate.) ⁓
Pelagic (
messages )
19:46, 9 November 2022 (UTC)
A URL parameter does exist (?search-thumbnail-extra-namespaces). Why do we want persistence for logged-out users? Showing images seems like a good improvement for the general reader, who wouldn't be using workflows (such as requesting too many search results or copy-pasting results to Excel) with which the images interfere. –
SD0001 (
talk)
03:51, 10 November 2022 (UTC)
Glad to hear about the param. Good points; what about users with limited data speed/allowance? Though I admit I haven’t compared page size for search-results versus typical articles. Perhaps the thumbs don't make such a big difference? ⁓
Pelagic (
messages )
11:05, 13 November 2022 (UTC)
Is it possible to undo a move of an article at Wikipedia. I think it would be a very useful possibility. Namely, I tried to undo the move of the article, which I was unable to do. I had to move the article again. This can be problematic, for example, because of the spelling of the name, or probably if the article has been moved more than once, I would not be able to make that move.
Governor Sheng (
talk)
10:38, 13 November 2022 (UTC)
(
edit conflict) @
Governor Sheng: If you go to
your contributions, at the top you should see a "
logs" link. Follow that. and in the first drop-down box select "Move log", then click Show. Locate the log entry for the move concerned, and towards the end of the entry there may be a "revert" link. You can use that to undo a move that was made by mistake. This works by initiating the move process with one or two items pre-filled that are normally presented as blanks.
Whether you use the "revert" feature or not, if a page is moved more than once (i.e. A → B → C) and you want to move it back to the original name, the technique is to backtrack along the chain (C → B → A) rather than try to move directly (C → A). --
Redrose64 🌹 (
talk)
13:08, 13 November 2022 (UTC)
...after checking that no one has helpfully retargetted the redirect A directly to C because, if they have and you're not a
page mover, the second move will, fail leaving the page stuck at B.
Certes (
talk)
14:27, 13 November 2022 (UTC)
Per
Wikipedia:Moving a page#Moving over a redirect, if you aren't a page mover or administrator then you can only revert a move if the move is the only entry in the page history of the old title. There are often other entries like adding redirect categories or retargeting the redirect. A move is not an edit but a log action. It's shown in the page history of both titles but only in the page log of the old title and can only be reverted from the log entry. This makes it such a hassle to use revert that it's usually easier to just make a new move. One way to use revert starting from the page history of the new name: Click the old name in the move summary, click "(Redirected from ...)" at top of the page, click "View history", click "View logs for this page", click "revert" at the move.
PrimeHunter (
talk)
16:08, 13 November 2022 (UTC)
But now I converted
File:FW DeKlerk sign.png to
File:FW DeKlerk sign.svg and his signature is used in about 50 different Wikipedias. I most certainly don't have the attention span to do this mass replacement by hand; or if I had to do it, only in a lot of smaller chunks. This would likely take a week or more. Is there any way to automatically replace them? I have thought about this multiple times before, but I never found a way. Technical people, please help me out... Thanks so much in advance! Cheers! --
Lommes (
talk)
16:40, 13 November 2022 (UTC)
@
Lommes A lot of these were being pulled across automatically from Wikidata - many smaller wikis are set up to generate infobox values automatically based on WD content. I've
changed the WD signature value to point to the SVG not PNG; the SVG is now being used on 30 wikis and the PNG only on 15. I guess those ones will still need to be switched by hand but at least it's a smaller problem :-)
Andrew Gray (
talk)
16:46, 13 November 2022 (UTC)
@
Lommes without changing the pages, you could just upload the new file over the old file - that's prob not desirable here though. A similar option would be to move the old file, then upload the new file to the old name. —
xaosfluxTalk16:49, 13 November 2022 (UTC)
In this case (where the file names indicate their file types as well, those are probably bad ideas, now that the count is lower based on the WD changes Ansdrew made, you could use something like
WP:AWB to find/replace on just one proejct with a lower volume. —
xaosfluxTalk16:50, 13 November 2022 (UTC)
Thank you so much,
Andrew Gray. 15 wikis was already a tall order (and i need to rest now) but it was a manageable amount. 50 was not. As to
WP:AWB, it doesn't really work on Linux, at least not good enough to actually save time. Thanks everyone! --
Lommes (
talk)
17:24, 13 November 2022 (UTC)
Currently, the following error appears in the infobox in the article
Signal (software):
Error: first parameter cannot be parsed as a date or time.
I failed to fix the problem by editing the article, so it appears to me that the problematic data in the infobox is loaded from Wikidata (the documentation only mentions that the official website (P856), source code repository and software engine (P408) are loaded from wikidata). How can the error be fixed?
Kallichore (
talk)
19:24, 11 November 2022 (UTC)
I noticed that Github-wiki-bot made 125 changes on 26 October 2022 at the Wikidata entry (
difflink). The problem is still unclear to me, but I think there is a connection to Wikidata and maybe to the edits of Github-wiki-bot. --
Kallichore (
talk)
07:29, 13 November 2022 (UTC)
Thanks,
Matma Rex! I didn't even know that another team had taken up the mobile improvements baton. I admit that I often live under a rock, but it would be nice if these kinds of programmes were announced more widely, or easier to find. ⁓
Pelagic (
messages )
08:30, 13 November 2022 (UTC)
It's the usual dilemma of being too quiet or too distracting… But it looks like this project has been rather quiet, so @
Samwalton9 (WMF) please consider this a request to brag about your team's great work more ;)
Matma Rextalk09:24, 13 November 2022 (UTC)
Thanks for the ping @
Matma Rex! Oops, it really has been a while since I provided an update on that page. Our work has been progressing fairly slowly but I really should have posted something before now, I'll make sure I do that this week. @
Pelagic This is only live for users with Advanced Mobile Contributions enabled and we're always happy to receive feedback on the project page linked above. In particular we implemented a new main menu design (the submenus are still in-progress) and would love to know what you think about it.
Samwalton9 (WMF) (
talk)
10:17, 14 November 2022 (UTC)
Changing {{increase}}/{{decrease}} etc. to be ignored in sortable tables
Templates like {{increase}}/{{decrease}} that display icons like / currently affect the sorting order of sortable tables in which they are used: all decreases are sorted first, followed by all increases. For example:
Value
1
2
5
8
Real example:
Visa policy of mainland China#Visitor statistics (in this article, this also causes the numbers to not be detected and be sorted alphabetically, which breaks the ordering even more thoroughly).
I'd like to add the following markup to these templates, so that they will be ignored when sorting: <span data-sort-value="">…</span>. Example with this markup:
@
Matma Rex Hmm. Looking for earlier discussions, there's an old thread at
Template talk:Increase#Can table cells using / be made sortable? that seems to suggest that these are sometimes used to indicate the magnitude of the change, rather than the value which has been changed - so a list like your example would be intended to read as as "-1, +2, +5, -8", rather than "1 (decreasing), 2 (increasing), etc". And in that case, you would presumably want it to sort as [-]8, [-]1, 3, 5"
However, I looked through more uses of the templates, and there are unfortunately some tables where they have been used in lieu of plus/minus signs; one example is the "Electoral Performance" table in
Janata Dal#List of prime ministers (although it's not sortable). This is probably a reason not to do this…
Matma Rextalk23:20, 14 November 2022 (UTC)
If the numbers were sorted correctly then I would also prefer to sort increases and decreases separately in
Visa policy of mainland China#Visitor statistics. The direction of change usually matters so much that I don't think an increase and decrease around a million belong next to eachother in a sorted list which has many smaller increases and decreases.
PrimeHunter (
talk)
23:55, 14 November 2022 (UTC)
Reply Tool on mobile
I thought that was deployed, but I'm not seeing Reply links any more on w:en mobile. I do see them on French Wikipedia (with the new visual styling). ⁓
Pelagic (
messages )
19:34, 9 November 2022 (UTC)
It has not been enabled on English Wikipedia yet. It is currently enabled on several Wikipedias, including French; and we will soon run an A/B test comparing it to the current mobile talk page interface on several others (see details at
mw:Talk pages project/Mobile#Status Updates). Assuming that doesn't suddenly reveal some negative effects (the results from the projects where it's enabled were promising), it should be enabled everywhere soon. I don't know the timeline, but I hope it'll be before the end of the year. (You can try it out on a one-off basis here by using the URL parameter dtenable=1, for example like this:
[21].)
Matma Rextalk20:12, 9 November 2022 (UTC)
I'm writing this reply with ?dtenable=1 and have used this in the past also. I actually liked it, but there is no beta preference yet to enable it, and having to append ?dtenable=1 to the url everytime I visit a page isn't a sustainable option. Although, I have to say that the average page on mobile version seems to be opening a little slower. —CX Zoom[he/him](
let's talk • {
C•
X})10:06, 13 November 2022 (UTC)
Hi,
Whatamidoing (WMF)! (Pinging although you might also be subscribed.) I’m more usually on tablet, but replying on a phone now. (Thanks to MR's link above, I didn’t have to fiddle with the address bar. 😉 ) As someone who’s already using the desktop Discussion Tools (often on the "mobile" tablet), Reply in Minerva/mobile is comfortably familiar, and I would recommend it.
The five-button toolbar fits nicely across a phone screen, and I like how bold and italic have been moved into the style dropdown to reduce the button count. With the new appearance, other sections are collapsed so it’s obvious if I scroll outside the section I’m replying in. Hints such as "Return to reply V" and "Show 1 new comment" appear to be working fine. Can’t tell yet about page performance (seems fine on this device, which is at the newish/faster end of the spectrum).
The blue bendy-arrow is a bit heavy and distracting where there are a lot of short comments, is that still under discussion? (Is this what you mean when asking if it’s annoying?)
Minor observations: (a) I just noticed that the on-screen keyboard gets out of the way when I tap the A drop-down, and reappears when I dismiss it, without having to tap again in the edit box / editing surface. It may seem like a silly thing to notice, but that doesn't always happen elsewhere with touchscreens. (b) Though the gutters outside the edit box are narrow, I can just manage to tap there to deselect the box and hide the keyboard. (c) Because the edit area expands and doesn’t have its own scroll bar, I can drag-scroll the page without having to find somewhere safe to grab outside the box, c.f. classic source editor. (d) Small text is rendered normal size in Minerva, so in visual mode I can’t tell where I've applied that style.
I accidentally swiped back just now while trying to position the cursor near the left edge of the edit box, but miraculously didn't loose my writing after a page-back-forward cycle.
Apologies for the long reply, I got carried away as usual. Short answers: Like? Mostly yes. Annoying? A little. Recommend? Compared to section-source-edit on mobile, yes. ⁓
Pelagic (
messages )
10:30, 13 November 2022 (UTC)
Thanks, @
Pelagic. Editing on a phone has been so difficult that I tend to think anything will be an improvement, but it's nice to have someone who has used this with a phone confirm my guess. These tools are
out for all mobile users at 9 Wikipedias, including the French Wikipedia. Nobody has tracked me down to complain about it, or even to ask how to turn it off, which is good but not a strong signal. The next step is an A/B test at 15 Wikipedias. It'll take about a month to run, and then probably another month to analyze the data and figure out what happened.
The swooshy arrows are expected on mobile and are being proposed to a couple of other wikis, because the old [reply] is being replaced with the new Reply, and that's kind of awkward when your language's word for "reply" is two characters. It needs to be big enough to tap on it, if you're on a small phone. But perhaps it could be made a little smaller? It's not set in stone.
Do you think that any of your observations are things that you'd like one of the engineers to think about? Please don't hesitate – there are thousands and thousands of mobile users, and you're the first to mention these. I'd rather ask them to look now than to wait until even more people have been affected by a problem.
Whatamidoing (WMF) (
talk)
03:00, 15 November 2022 (UTC)
Collapsible table that won't collapse depending on other text
It sounds odd if a script makes the difference unless it depends on a partially random loading order. Did you bypass with Ctrl+F5 (on common Windows browsers) and not F5 or the reload icon alone? Your "after" link has a different url structure. Does
this after link work? What is your skin? Does it help to change it? Does it work to copy the code to another wiki?
PrimeHunter (
talk)
15:30, 14 November 2022 (UTC)
Ctrl-F5 doesn't seem to make a difference, either on the link you give or the current page without the revid in the link. In Monobook the "[hide]" is no longer even a link, just black text, in the revid link; in the live page it's still a link but doesn't work. I'll try another language wiki next.
Mike Christie (
talk -
contribs -
library)
15:41, 14 November 2022 (UTC)
No, that code is for ' p.' including a leading space. The problem is in the revision with 'pp.' which other parts of the script has code for.
PrimeHunter (
talk)
18:27, 14 November 2022 (UTC)
My JavaScript is very limited. I could make a version where the whole code for "pp." is disabled. Would the script still be good without it?
PrimeHunter (
talk)
20:18, 14 November 2022 (UTC)
I think I'll stick with it as it is in that case, but thanks. I do a lot of source reviewing, and the script highlights various kinds of source formatting errors, such as "p. 15-20" or "pp. 15". Now I know it interferes with table collapsing, I'll just disable it or log out if I find I need to hide/show tables. Thanks for the help, though.
Mike Christie (
talk -
contribs -
library)
20:23, 14 November 2022 (UTC)
In case it's ever relevant, you can use
mw:safemode to disable all your scripts for a one-time page load. That will save you the trouble of editing your common.js or needing to open the article in a private/incognito window.
Whatamidoing (WMF) (
talk)
03:09, 15 November 2022 (UTC)
The reason
User:Lingzhi2/reviewsourcecheck-sb.js breaks collapsible tables is because it changes the innerHTML property. This causes the affected fragment of the page to be parsed as HTML and rendered from scratch (even when you're just appending HTML, not changing it), which loses any interactivity (event handlers) that has been added after the page was loaded, such as the behavior of collapsible tables.
I have problems with most Wikipedia pages on mathematical topics when using dark mode. Some expressions succeed in changing to white color but many others do not.
I have observed that the ones that do not work are those having a 'display=block' inside the 'math' tags. Do not work as well those using the tag 'equation'.
I think the issue is important because many people use dark mode, specially in winter when there are so many night hours. Besides, dark mode saves energy and reduce CO2 emissions.
@
Altasperlas to confirm, you are only having this problem with a mobile device, with the mobile app installed? Are you using the IOS or Android app? What version of the app are you using? A bug can be opened on this in Phabricator (
WP:BUG), screenshots will be helpful. —
xaosfluxTalk15:54, 15 November 2022 (UTC)
(copied from my talk —
xaosfluxTalk18:26, 15 November 2022 (UTC)) Hi, I am using the latest version of the Wikipedia app with an android mobile phone. The problem is recurrent in most pages that use mathematical content, whenever the option 'display =block' of the math tag is used; also the expressions that use the tag 'equation' are badly handled, but this tag is less used in wp articles;
There are a few app dark mode math bugs out there,
phab:T252904 for iOS,
phab:T284327 for Android (specifically mentions block display in comments),
phab:T255038 for both,
phab:T268279 for both (also specifically mentions block display in comments), and
phab:T182128 for both.
This should be an easy fix; their math handling just picks the wrong selector, so I don't know a) why it's not fixed, and b) why there's not a specific task documenting the issue.
Izno (
talk)
19:17, 15 November 2022 (UTC)
phab:T182128 has been around for almost 5 years now, just doesn't seem like there's any sense of urgency.
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you.
Translations are available.
Recent changes
At Wikidata, an interwiki link can now point to a redirect page if certain conditions are met. This new feature is called
sitelinks to redirects. It is needed when one wiki uses one page to cover multiple concepts but another wiki uses more pages to cover the same concepts. Your
feedback on the talkpage of the new proposed guideline is welcome.
[23]
The
new version of MediaWiki will be on test wikis and MediaWiki.org from 15 November. It will be on non-Wikipedia wikis and some Wikipedias from 16 November. It will be on all wikis from 17 November (
calendar).
There will be a new link to directly "Edit template data" on Template pages.
[25]
Future changes
Wikis where mobile
DiscussionTools are enabled (
these ones) will soon use full CSS styling to display any templates that are placed at the top of talk pages. To adapt these “talk page boxes” for narrow mobile devices you can use media queries, such as in
this example.
[26]
As, bizarrely, no one outside the WMF interpreted it in that way, the Wishlist announcement has been retracted, after needlessly upsetting people. With the WMF budget skyrocketing, one would expect that for CommTech to get more work done, they would get more staff, not this.
Fram (
talk)
14:41, 17 November 2022 (UTC)
Thank you for the clarification. Rather than wait up to two years even to have our wishes registered, I'm sure the community can do most of the work of discussing and voting on the wishlist. I realise that some technical input will be needed: if two requests are equally useful but one needs a day's work and the other a year's, obviously the former takes priority.
Certes (
talk)
14:53, 17 November 2022 (UTC)
Can we get a disambiguation redirect when a disambiguation page is moved to its base page name?
Currently, when a page is moved from any "Foo" title to any "Foobar" title, a nice redirect is left behind at "Foo" (unless suppressed) with the following bonus content:
{{Redirect category shell|
{{R from move}}
}}
Common at
Wikipedia:Requested moves are requests to move a disambiguation page to the base page name, on the theory that whatever title was there already is not the primary topic.
Can it be made so that whenever a "Foo (disambiguation)" title is moved to a "Foo" title (i.e., the same title except without "(disambiguation)" we get the following left on the redirect page:
{{Redirect category shell|
{{R from move}}
{{R to disambiguation page}}
}}
Is that literally text, or can it be something like {{subst:#if:this looks a R to dab|proposed text|current text}}?
Certes (
talk)
14:36, 12 November 2022 (UTC)
I feel like this is solvable. An alternative would be to get a bot to just add this template where the circumstances are met ("Foo (disambiguation)" is a redirect to "Foo", which is a disambiguation page) just as we have bots that fix double redirects.
BD2412T17:38, 13 November 2022 (UTC)
It's approved for doing what we want (though I'm not sure how well it supports the Rcat shell) but doesn't seem to be doing it despite running regularly to do related good work. Pinging
R'n'B.
Certes (
talk)
20:28, 13 November 2022 (UTC)
That discussion also seems to be about creating redirects, rather than tagging existing redirects as R to disambiguation. We seem to have
1358 untagged redirects.
Certes (
talk)
21:04, 13 November 2022 (UTC)
Unless anyone objects to tagging unlinked redirects, that seems very reasonable: there aren't many pages to tag, and as redirects to dabs they do deserve the template.
Certes (
talk)
22:18, 13 November 2022 (UTC)
┌───────────────────────────┘ @
Certes, I've finally gotten round to writing some code for parsing the pages. I've started the run, it should be done in a few hours. —
Qwerfjkltalk21:39, 16 November 2022 (UTC)
Is this task necessary? Sure, rcats are nice (I place them whenever I can on new redirects), but their usefulness is at best marginal. And there's a cost here: the edit that adds the rcat will make it impossible into the future for users without advanced permissions to revert the move. –
Uanfala (
talk)
13:24, 17 November 2022 (UTC)
My initial request followed from the fact that a redirect category is already automatically made when a page is moved. If the appropriate disambiguation category could automatically be added during the initial move in the cases specified, that would not implicate future moves.
BD2412T13:27, 17 November 2022 (UTC)
"Languages" cross-reference on left side of Wikipedia window
If I recall correctly, until a few years ago, all the language cross-references in an article (on en.wikipedia) were listed in the left panel under the Languages heading. Now, the list is abbreviated, and there is a "## more" button, with additional languages listed by region.
I think the old way was much more user-friendly. Navigating way down to find the Japanese version (for example) of the article within the "## more" dropdown is a pain.
In my experience, most pages are long enough to allow all the cross-referenced languages to be listed. Conversely, stub and other short pages usually don't have many other languages to display. And if the language list alone would lengthen the page, so what?
But if the issue is that the Languages list should not cause a page to lengthen, couldn't the feature be programmed to lengthen and shorten dynamically, to fit on the page, with the "## more" button used only for overflow?
(Please let me know if this is not the correct Village Pump/Help/Contact page to be using for this suggestion.)
Have you tried typing in the name of the language you want, rather than scrolling?
The "new" system (which was introduced on 2 July 2013) gets used a lot more than the old system. The number of languages shown at a time was determined through user testing and based on what most users found easier to use.
Whatamidoing (WMF) (
talk)
01:27, 18 November 2022 (UTC)
Can we get an option to suppress redirects in User contributions?
In searching user contributions, we have options to "Only show edits that are latest revisions", "Only show edits that are page creations", and "Hide minor edits". I'd like to see added to that an option to suppress redirects in the results. Can that be done?
BD2412T01:35, 18 November 2022 (UTC)
Yeah, and it was already off when you asked about the hoax category, too. The page that broke it isn't restorable. —
Cryptic00:48, 17 November 2022 (UTC)
I asked someone with Moar Power to take a look, and they kindly fixed it for us. (Interested bystanders: the problematic page had been oversighted, so ordinary admins couldn't restore it.) —
Cryptic12:16, 17 November 2022 (UTC)
Pinging
EmanWilm who updated some of the lists this year. In general there will be a lot of templates with redlinks. Indeed, template in some categories consist almost entirely of redlinks.
Certes (
talk)
16:09, 9 November 2022 (UTC)
If there are a lot, then it makes sense to lay them all out, so we know the scope of the task, and perhaps whether there are patterns of issues that can be resolved with some large-scale thinking.
BD2412T16:36, 9 November 2022 (UTC)
The thing about having a directory at
WP:TRL is that it is (hypothetically) much easier to browse the lists and parse them out by area, so that editors can see what in their area has templates with red links. When we do have a newer list, we can also do a better job of grouping templates by subject matter, and informing corresponding Wikiprojects and likely interested editors of those subsets.
BD2412T04:53, 11 November 2022 (UTC)
I could easily put up a list of the 2,319 templates with 1–5 red and >100 total links, or even the 46,888 templates with 1–5 red and >10 total links. Classifying them would be harder. We could attempt an initial pass semi-automatically, for example by splitting off templates in any category containing the word "football", though it might require following a chain (Template:Foo FC squad → Category:Foo FC templates → Category Barland football templates) and each template can have dozens of other chains which are unhelpful. (Which code of football is being played is left as an exercise for the reader.)
Certes (
talk)
12:24, 11 November 2022 (UTC)
From the 46,888: the 20 most popular words in template titles are:
2576 in
2528 of
2428 football
1873 squad
1547 men's
1271 team
1146 standings
1137 champions
1021 navbox
1017 women's
987 the
980 basketball
931 conference
870 summer
854 world
823 cup
759 olympics
712 games
709 footer
684 roster
Obvious noise such as "in" can be ignored. There is some overlap and correlation: predictably, removing the "football" templates reduces the count for "team" but not for "basketball". We may want to make a list of words (or short phrases such as "American football") and test for each, starting with the most specific and working to the more generic, so the "team" templates exclude anything with a better hint such as "basketball". Exploring keywords in categories too might help; I've not tried that yet.
Certes (
talk)
16:14, 11 November 2022 (UTC)
@
Certes: This is excellent, and I think is something we could work from. I had intended to classify manually, but this is another level.
BD2412T17:36, 13 November 2022 (UTC)
Hello. About a year ago
I started having problems with my watchlist returning very limited number of results.
Just as it was starting to get back to normal, today it suddenly and completely failed. No changes showing. I have 1,934 pages on my watchlist. My settings are 30 days, 500 changes, and "Hide my edits from the watchlist" is the only ticked box. --
DB1729talk01:48, 17 November 2022 (UTC)
Xaosflux. No, same thing. However, I have now discovered that if I select an individual namespace, I can get some changes to display. That is, until I check "Category", then it goes empty again. I'm wondering if there was a massive series of adding/removing of one of my watchlisted categories today, and my watchlist can't handle the volume for some reason? --
DB1729talk02:14, 17 November 2022 (UTC)
@
DB1729 are you sure there should be results? Pages in the "Category" namespace doesn't get edited often. To test, make an edit to a Category page in your watchlist and see if it appears. Note, placing other pages in to or out of a Category is not an edit to the category page. —
xaosfluxTalk11:08, 17 November 2022 (UTC)
Yes. Absolutely sure. I have a lot categories on my watchlist and I get that type of edit all the time. It looks a little different. It has a description that says something like "Article X was removed from Category:Foo. This page is included in other categories." The "diff" button is grayed out (because no change was made to the cat itself) but there is a link to the article where the change was made so you can click on that if you want to view the diff.
Update: Strangely, although I'm not surprised really, my watchlist is currently displaying exactly one change this morning. This behavior is very similar to back in December '21 when this all started as
I described here. Whatever happened back then, it's been affecting my watchlist ever since. It has not displayed a full compliment of changes for 11 months now.
And I'm not sure you fully understood when I said when I click on "Category", it all goes empty. I can get results by clicking a combination of article, template and module for example and get (limited) results to display. It will show a mix of article and template changes. It would show a few article changes mixed with some template changes, and no module changes as expected, because I think have only one module page on my watchlist. But as soon I click category, it all goes empty, just like for the default "all content" when you first click on your watchlist. Empty. Until this morning when it displayed a single article.
DB1729talk11:33, 17 November 2022 (UTC)
Would it be a sensible wishlist request for "display most recent edit only" (i.e. when Preferences→Watchlist→"Expand watchlist to show all changes, not just the most recent" is unticked) to mean "display most recent unfiltered edit only"? E.g. if a page is edited by (oldest) Human1, Human2, me, Bot1, Bot2 (current) and I hide bots and own edits, I see only the edit by Human2. By the way, I group all changes for each page together into one row per page, sorted by the time of last update, which I find much more helpful than seeing only the last update.
Certes (
talk)
14:28, 17 November 2022 (UTC)
"Article X was removed from Category:Foo" is not a change to a category it is a change in categorization; make sure that "page categorization" is not in the hide option. —
xaosfluxTalk16:27, 17 November 2022 (UTC)
Update: Just got home and now, without changing anything, my watchlist is displaying six changes from today only, nothing from any earlier than that. Very similar to to how this all started last year. It's slowly repopulating my WL with a few more results each day. I am going to select "Expand watchlist to show all changes" now and see what happens... --
DB1729talk22:57, 17 November 2022 (UTC)
@
Redrose64: So now, after I set to "Expand watchlist to show all changes" it is showing nine changes from six different pages. Still nothing from any earlier than today.
@
Xaosflux: I don't see anything in preferences that says "hide page categorization". Never knew that was an option. I just assumed you always get categorization changes, as well as changes to the category page itself, if you put a category in your watchlist. Where is that option to hide categorization located? --
DB1729talk23:11, 17 November 2022 (UTC)
Wow! I forgot to look there. That did it!!! My watchlist immediately filled up with results and seems to be behaving normally now. I'm now convinced the problem had to be flooding from categorization edits. Thank you MANdARAX, Xaosflux and everyone for your help.
DB1729talk00:49, 18 November 2022 (UTC)
@
Xaosflux: And again I will try rephrase so you can see my point about the categories: If I select article namespace and template namespace, I get over 100 results from the last 30 days from both article and template namespace. But if I select both article and category namespace, it all goes empty, including the article namespace changes. It all just disappears. That's why I think it might be a category page that has flooded my WL; and if can hide categorization changes, it might restore my WL to a more normal set of results.
And now after just clicking the regular default watchlist button, only five changes are displayed instead the nine I just got 30 minutes ago. Four changes just disappeared for no apparent reason. It's acting crazy; it's just f'ed up.
All I want is to be able to click "watchlist" and it show me any changes from the last 3 or 4 days and I will be happy. That's what I had before, for two years. It was fine, I changed nothing, and then late last year it just started acting goofy and its never been quite the same since. Extremely frustrating.
DB1729talk23:41, 17 November 2022 (UTC)
Unchecked "Category changes" in filters (unchecking effectively hides those changes) and all seems normal now! So my watchlist was indeed flooded with categorization changes. Filtering those out fixed everything. Thanks to all for helping solve the problem and the mystery:) Cheers!:)
DB1729talk00:52, 18 November 2022 (UTC)
@
DB1729: You state If I select article namespace and template namespace also if I select both article and category namespace - I don't understand how you can select two namespaces and exclude all of the others.
Special:Watchlist has a namespace dropdown, you can select "all" or a single one - having selected one, if you then select another one, this replaces the first one that you selected. --
Redrose64 🌹 (
talk)
01:25, 18 November 2022 (UTC)
That's not the behavior I've seen. I can select multiple namespace and it combine the results for any combination unless I included category with it.
DB1729talk01:38, 18 November 2022 (UTC)
I just confirmed what I just stated above. I can use the namespace dropdown. It is selected at the right, next to the tags dropdown button. A ticked box never disappears when selecting addition namespaces. However, sad news. I playing around with this and checking if the category namespace works now, it did the same thing as before and now my watchlist went back to screwed up state it was in before. I'm gonna try a couple things and see if it helps. Stay tuned...
DB1729talk02:16, 18 November 2022 (UTC)
Ok. Fixed it again. It put back that same categorization filter that was causing the problem and so it put the same whammy back on my WL. X'ing it out, again fixed it. But yeah, I can 100% confirm, I can select multiple namespaces from the watchlist dropdown and it will show combined results for just those selected and exclude all other namepaces.
DB1729talk02:23, 18 November 2022 (UTC)
@
Whatamidoing (WMF): I think that you know about various interfaces. Apart from
Special:Watchlist, is there another way of displaying the watchlist, one that allows more than one namespace to be selected? There is nothing at
Help:Watchlist#Options, so I am really confused by this description of I can use the namespace dropdown. It is selected at the right, next to the tags dropdown button. A ticked box never disappears when selecting addition namespaces.. I don't even have a tags dropdown. --
Redrose64 🌹 (
talk)
17:11, 18 November 2022 (UTC)
I don't think it's meant to work that way. @
DB1729, it might help if you could provide a URL for this, assuming that it gives you one of the (really) long ones and not just "Special:Watchlist".
@
Whatamidoing (WMF):I have no idea how to pull up a url for such a thing. I merely changed a filter setting. Would that show up on a log somewhere?
It sounds like a bug to me only in the sense that my watchlist was rendered unusable when my filters were set to show categorization changes. My theory is, like I said, yesterday someone probably added or removed an enormous number of pages to/from a category in my watchlist, and it couldn't handle the volume. I don't know this for certain of course, because my watchlist trying to display those edits, is in my theory, what broke my watchlist to begin with.
DB1729talk01:36, 18 November 2022 (UTC)
When viewing an article in my default vector-2022 skin, footnote numbers are superscripted; they appear a little higher than the text. When editing an article in Visual Editor mode, the footnote numbers are subscripted! I have just been through a number of Chrome/Windows 10 updates, so there's a lot of potential in the mix. This change just became noticeable this week. I haven't been doing a lot of visual editing, but I'm fairly sure it's new as of a week or two.
Elizium23 (
talk)
08:55, 18 November 2022 (UTC)
I was using Recent Changes to look for IP vandalism. I have several highlighting options, including highlighting "Very likely good faith" edits green and "very likely bad faith" edits black. I also set the "Live Updates" option on. Suddenly, after a live update, the Very likely good faith and very likely bad faith options went away for no reason and I could not add them back, even by reapplying the filter set nor by refreshing. It only worked when I closed the tab, opened a new one via the "Recent Changes" link on the left sidebar, and then applied the filter set again. I'm not sure why this happened.
RPI2026F1 (
talk)
13:01, 18 November 2022 (UTC)
I've mentioned this bug in the past. I think a phab was filed for it but I can't remember. I also dont remember where the link to the discussion is. ―
Blaze WolfTalkBlaze Wolf#654518:22, 18 November 2022 (UTC)
Putting citations inside a note ref group and also reusing that note
I can't figure out how to (a) use a separate notes ref group, while also (b) putting citations inside that note, and also (c) re-using the note. Here's what I tried.
This is a sandbox with a note done using <references group ="refgroupnotes">...</ref>, and another note done with {{tag:ref|...|group=tagnotes}}, plus a regular <references /> at the end.
Then in
this version I've added a citation inside each of the notes groups. It works for the tag:ref but not the refgroup one. So it looks like I have to use the tag:ref method if I want to have citations in my notes.
So I reverted that and tried
this, in which (in VE) I copied the note and applied it to a later sentence, for both the tag:ref and the refgroup note. That works for refgroup, but for the tag:ref it just generates a second copy of the note instead of two links from a single note.
So I can have a note used twice and a note that has internal citations, but not one that will do both. I confess I don't really understand what the tag:ref method is doing; I've just been copying and pasting it for many years. Is there a way to put an identifier on it so it can be re-used like a ref tag?
Mike Christie (
talk -
contribs -
library)
03:03, 19 November 2022 (UTC)
That does look easier. I think I never tried efn because I use VE and I seem to recall VE couldn't handle it at all at one point, but now I see VE handles it and tag:ref about the same. The main issue now (for both) is that with a re-used note you can only edit the note text in VE by clicking on the first instance of the note. I have to use source mode for tag:ref anyway so I might try switching to efn. Thanks.
Mike Christie (
talk -
contribs -
library)
11:21, 19 November 2022 (UTC)
Accordingly, I wanted to ask if there's any way to make that template suppress the generation of any categories if the month and year specified in the template is still in the future, and generate categories only when the month in question actually arrives.
Bearcat (
talk)
05:09, 15 November 2022 (UTC)
If someone is going to change the math expressions in that template, I would recommend a catch-all category called something like "Articles containing potentially dated statements from the future" as an error-tracking category. The category could be listed on on the template's documentation page. A typo like {{as of|November 2202}} would generate an error category (which can then be fixed) rather than no category at all (which will be silently ignored but still an error). –
Jonesey95 (
talk)
05:35, 15 November 2022 (UTC)
I like the idea but not the proposed name. Using {{as of}} for a future date makes no sense so it usually means the template should be simply stripped.
Nardog (
talk)
05:41, 15 November 2022 (UTC)
I've implemented Jonesey95's suggestion in the sandbox. @
Nardog: would you explain what you don't like about the name? Personally, I don't think there's anything inherently wrong with using the template for future dates,, as the one use that currently uses a future date is
Pennsylvania State Senate#Current session, where the information is both about the future (the session that will start in 2023), not obviously problematic, and likely to become out of date. I think that shows that the template cant try to guess what you mean when you give it a future date, as several people's guesses about what to do there would have been wrong.
* Pppery *it has begun...23:39, 19 November 2022 (UTC)
I think the template documentation is quite clear about prohibiting use of future dates, and I agree. {{As of}} is supposed to be used for statements that were true at some point in the past but may become outdated or inaccurate as things change in a future we cannot predict. Using "As of" for future dates encourages people to break the
WP:CRYSTAL ball prediction prohibition, and this is nasty stuff. I wish there were fewer future predictions here on Wikipedia, and so ensuring that "as of" can't be abused for them is a step in the right direction.
Elizium23 (
talk)
23:43, 19 November 2022 (UTC)
I don't see a problem with using it for future dates. The template says to use it "to provide the most current information available and will need a future update". With proper sourcing, a future date could be fine. For example, if an organization passes a budget for future spending, what is wrong with saying "As of Jan 1, 2023, the organization spends xxx per year on yyy", instead of "As of Nov 15, 2022, the organization plans to spend xxx per year starting in 2023". The template says not to use it for future dates because of the categorization problem - but that can be fixed.
MB00:34, 20 November 2022 (UTC)
I think that future events are counter to the intent and purpose of the template; the miscategorization is just an example of how it malfunctions when misused.
Because in the example I gave above, if you put an {{update after}} template on the Jan 1 date, the update on Jan 1 would be just to change {{update after}} to {{as of}} (and slightly adjust the wording). That's really not what update-after is for.
MB01:11, 20 November 2022 (UTC)
Actually, the possibility of extreme typos is a very good point I hadn't considered. Maybe a catch-all "error tracker" category is a better idea than just suppressing generation completely. That's almost certainly possible, because any date before 1990 just autogenerates one "before 1990" category instead of individual categories by year.
Bearcat (
talk)
13:07, 16 November 2022 (UTC)
Yes, I'd say this is in the realm of possible. You would need consensus of course. Jonesey suggests one possible differing end appearance.
Izno (
talk)
05:42, 15 November 2022 (UTC)
From today I am seeing a Learn more about this page button at the top of every project and project talk pages in mobile web, which doesn't do anything. I have tried it in a lot of pages and haven't found any where it works. I'm guessing it was meant to be a new
WP:ITSTHURSDAY feature. Does anyone know what this is supposed to do?
ಮಲ್ನಾಡಾಚ್ ಕೊಂಕ್ಣೊ (
talk)
04:09, 18 November 2022 (UTC)
I think this is meant to provide access to "top matter" (all the coffee roll boxes) on talk pages. That has traditionally been hidden from editors on the mobile site.
I'm using Firefox on Mac OS, and clicking on that gray text does nothing for me. Sometimes I think they really don't know about the ten-year-old bugs that are in phabricator, some of which are literally "set up a cron job", the fixing of which would make editors' work so much easier. –
Jonesey95 (
talk)
15:57, 18 November 2022 (UTC)
That´s another WMF win! Thoroughly tested obviously, with the support bought with donation money I suppose. Making the poor mobile environment even worse...
Fram (
talk)
15:00, 18 November 2022 (UTC)
I was trying to create the redirect
Ethnologue: Languages of the World ->
Ethnologue. The thing is, I can't. See the first blue link in the previous sentence? It's actually an external link, even though in wikicode it's [[Ethnologue: Languages of the World]]. It turns out, if you try to get to a Wikipedia page whose title is Ethnologue: followed by whatever, you'd be shown a soft redirect that takes you to www.ethnologue.com/language/whatever. –
Uanfala (
talk)
21:59, 20 November 2022 (UTC)
Per
WP:NC-COLON, Article titles cannot begin with an interwiki or interlanguage prefix. The top of that page suggests some workarounds, which basically amount to choose a different title for the page. Omitting the colon out and adding a {{correct title}} tag may be the best compromise.
Certes (
talk)
22:30, 20 November 2022 (UTC)
With JavaScript disabled and when at enwiki,
Special:Notifications shows notifications and thanks from enwiki. That has worked for years but recently it shows nothing other than a talk page notification if "you have a new talk page message" is present. Was a change about how this works announced?
Johnuniq (
talk)
02:36, 20 November 2022 (UTC)
@
Johnuniq There were no (intentional) changes to the no-JavaScript Special:Notifications recently (or within the last few years). It's working correctly for me right now, although I don't really understand the problem you're having. Can you elaborate and possibly provide screenshots of the bug? (And file a task at
https://phabricator.wikimedia.org/ if you can.)
Matma Rextalk22:50, 20 November 2022 (UTC)
@
Matma Rex: I have some screen shots but I'll start with a description. Using Firefox 106.0.5 with NoScript active, I have had no trouble receiving notifications or thanks from the current project (e.g. while viewing enwiki, an enwiki notification would be visible at Special:Notifications). I would have to enable scripting to see a notification from another project such as bnwiki or meta. I haven't been very active lately and I'm not sure when it started but around a couple of weeks ago something changed. For example, at 23:30 19 November 2022 UTC with scripting disabled, refreshing my watchlist showed the orange "you have a new talk page message" and a red [2] and blue [1]. Doing a ctrl-click on a link to my talk showed the new talk message with no orange bar and a red [1] and blue [1]. That is expected. The problem is that a ctrl-click on the red or blue shields took me to Special:Notifications and that page showed "Today xxx left a message on your talk page" then older notifications which I had previously seen and dismissed, and nothing else.Enabling scripting and refreshing Special:Notifications showed two additional items under "Today", namely a thanks and a mention. In the past, I would not have needed to enable scripting to see those extra items.
Johnuniq (
talk)
01:58, 21 November 2022 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you.
Translations are available.
Recent changes
The display of non-free media in the search bar and for article thumbnails in Special:Search has been deactivated. Further details are in
T320661.
Changes later this week
There is no new MediaWiki version this week.
Some wikis will be in read-only for a few minutes because of a switch of their main database. It will be performed on 22 November at 07:00 UTC (
targeted wikis) and on 24 November at 07:00 UTC (
targeted wikis).
My watchlist is pushing 9,000 pages. I struggle to keep it under control. I have left on the preference that adds every edited page, and so there are often unwanted pages being put on my watchlist as I edit them (I know, that's avoidable.) Are there any gadgets or tools that could bring the watchlist under control? I would like something with AWB-level logic that could, say, remove every page in a category from the watchlist, or remove pages in my last 500 contributions, or just allow me to manipulate it en masse in new and different ways. As it stands now, I have found no gadgets for this; is this an un-programmable part of the project?
Elizium23 (
talk)
02:38, 20 November 2022 (UTC)
9,000 is perhaps a large number for working with, but it's certainly not large enough to break the software. My watchlist is presently around 16,000 pages and I have no problems - it's been right up to 22,000 in the past and
Special:Watchlist has worked just fine. Above about 18-19,000 however,
Special:EditWatchlist hits problems - either it fails to list pages later on in the list (so e.g. Category: might be incomplete and Draft: might be absent) or it just times out altogether. --
Redrose64 🌹 (
talk)
10:07, 20 November 2022 (UTC)
@
Elizium23 ^-- this. I suggest you open raw mode, copy all the text to your computer, save as a text file. (This is a backup if nothing else). Then make a copy of that file and use any text editor you want to update it, suggest sorting it then you may find large sections that you can just delete. Then open /raw again and paste in the replacement to update the entire thing at once. —
xaosfluxTalk18:30, 20 November 2022 (UTC)
Elizium23, much of what you want to do can be accomplished with AWB. For example, to remove every page in a category from your Watchlist, do the following: From the menu at the top, select "Tools" then "List comparer". For "List 1" Source, select "My watchlist" and click on "Make list". For "List 2", select "Category", enter the name of the category, and click on "Make list". Then click on "Compare". Copy the results shown as "Unique in List 1" and paste it as your raw Watchlist, as mentioned above. (Or continue by pasting the "unique" list as your new List 1, clear List 2, and enter your next category to create your next List 2, and Compare again.) You can do the same to remove pages in your last 500 contributions, except for List 2, select "User contribs (user defined number)". Enter your username, and click "Make list". A dialog pops up asking for the number of contribs. Enter your desired number and click "OK". Note that if you don't want to request permission to edit with AWB, you can still install it and use the list comparer. MANdARAX •
XAЯAbИAM03:05, 21 November 2022 (UTC)
@
Elizium23 It might not do everything you want, but
User:Ahecht/Scripts/watchlistcleaner has a number of features for cleaning up your watchlist (it can remove redlinked pages, redirects, pages you haven't edited for a certain amount of time, etc.). Some of the options will be a bit slow for a watchlist of your size, but you could let it run overnight. --
Ahecht (
TALK PAGE)
06:33, 22 November 2022 (UTC)
Changing the reply button's appearance
The Editing team wants to change the reply button's appearance. For now, this should only affect the folks with the Beta Feature enabled (but that's a lot of active editors, if you remember the last change). If it goes well, then the change would presumably be rolled out to everyone later.
I've been using it for weeks on other wikis, and after the usual initial "Huh, that's different" adjustment, I find that I just don't really care. It works the same, and the exact appearance just doesn't seem that important either way to me. But if any of you have any concerns, please let me know.
In particular, if you use any of these languages: Arabic (ar), Chinese (zh), Japanese (ja), Northern Thai (nod), or Yue Chinese (zh-yue, sometimes called Cantonese), the word "reply" is only two letters long. While a wiki with any of those content languages could choose to use the mobile-style button (which adds an arrow – bigger buttons are preferred when every button has to be large enough to be touched by a finger), I understand that the extra formatting to accommodate the very short word ("reply" is just "ردّ" in Arabic) is wiki-specific, rather than UI-language-specific, so it won't appear here.
Whatamidoing (WMF) (
talk)
06:16, 21 November 2022 (UTC)
@
Whatamidoing (WMF) Is there an explanation somewhere of the reasoning behind this change? By accident or design there's a pattern of interface links within content being in square brackets, e.g. [ edit source ], [ subscribe ], [ reply ] so I'm not sure why we would want to break with that unless there's a reason.
the wub"?!"12:19, 21 November 2022 (UTC)
Note that the reply and subscribe links were added as part of the reply tool features, so it's the same team that is planning to alter the appearance. (I'm not sure where an "edit source" in square brackets appears.)
isaacl (
talk)
15:50, 21 November 2022 (UTC)
The [ edit ] link appears at the right side of each section (at least for me, with my prefs, in Vector) so that you can edit a section. The [ subscribe ] link appears to the right of the edit link (although a bit lower for me; it's not pretty). The three links are currently consistent in their design.
Whatamidoing (WMF), is there a redesigned UI pattern that this change to the reply link is trying to follow, and will the edit and subscribe links be changed to follow that pattern? What other "action" links share this bold formatting? Will the Read/Edit/View History links at the top of the page be changed to bold as well? –
Jonesey95 (
talk)
16:11, 21 November 2022 (UTC)
Indeed this breaks away from the pattern of links in square brackets, and uses the pattern of "
quiet buttons". The subscribe link is also planned to be changed (preview it here:
[28]). We're not planning changes to the edit / edit source links, or the tabs at the top of the page (although if you
try this on Vector 2022, the buttons in its sticky header use the same pattern).
Matma Rextalk17:01, 21 November 2022 (UTC)
Thanks for the link. When I follow that link, I see a large, normal-weight (not bold) "Subscribe" link with a hollow bell icon next to it in place of the current normal-sized [ subscribe ] link. At the end of each comment, I see Reply in bold, and slightly subscripted. So you have one link ([ edit ]) that is plain and in brackets, one link that is large plain text with no brackets and a bell icon, and one link that is bold with no icon and no brackets and slightly higher than the line of text it follows. That seems a lot less consistent than what we have now, and makes little sense to me from a UI perspective.
Also, when "Reply" wraps to the next line after the signature time stamp, it has a space in front of it, which is undesirable and does not comport with normal behavior of text. It looks like this needs a lot more testing as well as more thinking about UI design consistency. I do not envy the developers this task; I know it's not what I would choose to work on, since it seems enormous. On the other hand, if the developers would like to work on some actual many-year-old bugs and feature requests from volunteer editors instead of making currently consistent links into a UI mishmash, I'm sure that people here at VPT would be happy to link to some phabricator tasks. –
Jonesey95 (
talk)
17:22, 21 November 2022 (UTC)
@
The wub, it's a button, so I gather that they want to make it
look like a proper button. This will (eventually, when everything's converted over) result in much less of the "UI mishmash" that @
Jonesey95 noticed.
This will be going into the Beta Feature. If you have that turned on, then you don't see the old [ subscribe ] link any longer; you see a button with a bell icon in addition to the word. Of course, if you don't like it, then you can turn it off in
Special:Preferences#mw-prefsection-editing-discussion. It'll be bundled with "Show discussion activity" (which I believe you can only see now if you have the Beta Feature enabled).
Thanks for that link. By my reading of that design document, "Reply" in this case is a "quiet button", which explains why it is bold and which may explain why it sometimes appears to be slightly subscripted instead of vertically aligned with the text that precedes it. This design document appears to conflict with the
link provided above that shows the "Subscribe" button/link as larger text that is not bold, and the section edit links still as bracketed plain-text, normal-weight links. If you're going to change the WP interface so that these links become buttons, at least do it all at once, please. –
Jonesey95 (
talk)
05:20, 22 November 2022 (UTC)
I think bold might be more prominent than I'd personally prefer. I agree that it would quickly fade into the background and I wouldn't care that much. However it might have a knock-on effect of causing me to pay less attention to other bold text. This probably isn't a big deal on talk pages, though.
isaacl (
talk)
15:50, 21 November 2022 (UTC)
How often do you have a the timestamp of your signature both linked and also bolded? I'm guessing that's going to happen to maybe one in a million comments.
Whatamidoing (WMF) (
talk)
23:09, 21 November 2022 (UTC)
@
Whatamidoing (WMF) in a word: Yuck. The link currently appears to be a "control" / "action element" via the bracketing, just making it plain text isn't very distinguishing and just results in a random word floating at the end of sentences. Additionally, stylistically this would be identical to wikilinks, and the expected behavior when clicking on a wikilink is to follow it to a page, not to open up an editing dialog. Link styling may still be a good enough way to differentiate this control - perhaps a different color, background shadow, etc? —
xaosfluxTalk16:11, 21 November 2022 (UTC)
The real appearance will be slightly different than presented here (updated 17:14, 21 November 2022 (UTC)), with a different shade of blue and hover effects compared to normal links. You can preview it here:
[29].
Matma Rextalk16:54, 21 November 2022 (UTC)
I think it would be better if the new button had a very light border on it. Either that or a fill, like the Reply button on the reply editor.
RPI2026F1 (
talk)
22:41, 21 November 2022 (UTC)
If the whole button were blue, I think that would be really heavy.
I'm not a big fan of the current styling, so am amenable to a format that doesn't have square brackets. To me, though, the links have undue prominence in the preview. Perhaps a somewhat faded version of the normal button appearance (such as the one at the top of
the button design style guide) would be better. The borderless style of quiet button, to me, isn't a great fit for UI elements contained inline within text, as it can blend in too easily. The primary saving grace for this particular use is how it gets peppered across a section at the end of each comment, and that pattern can signal that something unusual is going on.
isaacl (
talk)
00:32, 22 November 2022 (UTC)
How would it look with a hollow swooshy-arrow, that echoes the hollow (not-subscribed) bell? I take the point that the reply-link's position after the time stamp is unique; even so it might benefit from some kind of distinguishing decoration. Either way I'd probably get used to it. ⁓
Pelagic (
messages )
12:42, 22 November 2022 (UTC)
{| class="sortable wikitable"
|+Median Household Net Income by Country
|+(Purchasing Power Standard)
<templatestyles src="Template:Static row numbers/styles.css" />
{|class="wikitable sortable static-row-numbers plainrowheaders srn-white-background " border=1 style="text-align:right;"
|-class="static-row-header " style="text-align:center;vertical-align:bottom;"
|-
! Country
! ...
|-
| ...
|}
|}
It's an attempt to nest a table inside another, but the outer table has (a) two captions (the maximum is one) and (b) too much crammed into the second caption. --
Redrose64 🌹 (
talk)
17:50, 22 November 2022 (UTC)
Right now if you go to
Special:Log via "View logs for this page" on page history, e.g.
[30], the Target field is empty, so if you modify the form to filter the results and click Show, it shows logs for all pages, unless you fill in Target yourself. And if you do fill in, now the special page remembers the input in that field; this appears to be dependent on whether the URL query includes &wpFormIdentifier=logeventslist. I'm pretty sure it didn't use to be this way and it's annoying. Is this a bug or a (misguided yet) deliberate decision?
Nardog (
talk)
16:23, 23 November 2022 (UTC)
Addendum: And worse yet: when I transcluded the sections to my
subpage, the tables' column headers's borders were cut off.
Qwerty284651 (
talk) 21:46, 22 November 2022 (UTC)
@
Izno:, I see you intervened on the rankings page. The thing is, the column headers should stay as they are, not be converted to table captions.
Qwerty284651 (
talk)
21:52, 22 November 2022 (UTC)
Okay, but that still doesn't explain the fact as to why the tables using the "display:inline-table" style param aren't displaying inline but below each other.
Qwerty284651 (
talk)
22:04, 22 November 2022 (UTC)
I see that they are functioning as expected in Vector; the keys are inline with the tables.
What I think you want is for each of the sections' tables to inline with each other. That doesn't work because you have headings between each, which are block display.
Izno (
talk)
22:12, 22 November 2022 (UTC)
This appears to be resolved now. The OP has structured the sections using the {{col-begin}} family of templates, which appears to produce the desired result. –
Jonesey95 (
talk)
22:21, 22 November 2022 (UTC)
I removed the subsection headings and col-begin, leaving only display:inline table, thinking the headers were the issue, as you pointed out, but they are still one below each other. Display:inline should work like it does in this
old version of the page. {{col-begin}} is the easy way out. Was hoping the style param would get the job done, so it's easier to edit the tables in VE, while also keeping both tables in each section inline.
Qwerty284651 (
talk)
22:26, 22 November 2022 (UTC)
Found a workaround. Should have know the legend templates between the tables were the cause of the problem. I relocated them to the bottom of the section. Works as intended. Thanks.
Qwerty284651 (
talk)
00:09, 23 November 2022 (UTC)
It all works great on desktop, but on mobile the table's caption is confined within the first column (skewing up the table, in the process) instead of spanning across the entire table.
Qwerty284651 (
talk)
02:34, 23 November 2022 (UTC)
But why does it only expand within the first column and is not sort of like colspan-ed across the entire table's width? It looks bad on mobile. And, if that's the default operation for captions, can it be fixed so it's stretched across the whole table?
Qwerty284651 (
talk)
02:45, 23 November 2022 (UTC)
You're going to need to provide an image of what you see. The behavior you claim to exist is not the behavior I am seeing, and I know what I'm seeing is also correct behavior. :)
Izno (
talk)
02:55, 23 November 2022 (UTC)
Yes, I see it too in Firefox now that I look at the correct resolution.
This appears to be how browsers react to CSS which sets the display of the <caption> to display: block (which is what Minerva and/or MobileFrontend is doing below 720px) when a <table> is display: table-inline.
I don't really understand why the <caption> has that CSS being applied to it since it always "display: block" in the ways that matter for mobile. That's probably a reasonable bug report.
Izno (
talk)
04:30, 23 November 2022 (UTC)
Well, I looked at it. If the table is display block (IDK why Minerva sets tables to display block though I've known it's how it is) and the caption is not display: block and also has more content than the table algorithm identifies, then the caption becomes wider than the table. I'm not sure if that's sufficient cause to make it display: block.
Izno (
talk)
04:39, 23 November 2022 (UTC)
No, probably not, but I'm seeing a different experience than either of you (one page of PDF with some content is how Firefox handles it). cc @
JPxG. @
SSastry (WMF), are these bad links in some way or is this a bug? (Forward as appropriate.) --
Izno (
talk)
22:46, 21 November 2022 (UTC)
I believe the REST base link is only simulating that, so if we can fix the template we'll fix this too. Guessing this is some kind of upstream Chromium error?
Jdlrobson (
talk)
02:45, 23 November 2022 (UTC)
Does the REST API use something like puppeteer to open a Chrome browser at the target page? That might explain why it is also having the same bug.
RPI2026F1 (
talk)
04:41, 23 November 2022 (UTC)
Why promise 100, 250, 500 when there may not be that many?
When someone clicks on the View history tab to see Revision history there are clickable numbers that are presented as (newest | oldest) View (newer 50 | older 50) (20 | 50 | 100 | 250 | 500) even when there are fewer than 500, or 250 or perhaps even 100. Is this false advertising? Isn't such a presentation, suggesting that an article that now has say 67 edits, including the initial article, has more edits than advertised? Is this a bug or a misfeature?
Nuts240 (
talk)
02:40, 24 November 2022 (UTC)
This is a fairly cheap optimization that doesn't pose much of a user mismatch: A static number is cheap on the database and a non-static number is not.
Izno (
talk)
03:23, 24 November 2022 (UTC)
Also,
false advertising would require us to actually be advertising something.
OK, not false advertising, and not even too misleading, but . . . can an hourly running bot store a count range, so that for "say 67 edits" the '250' and '500' numbers aren't displayed? Can this be put in a "suggested improvement" list?
Nuts240 (
talk)
19:53, 24 November 2022 (UTC)
That would require the same changes as your suggestion, and to boot would be less efficient. This is not likely to be a task that will be worked on.
Izno (
talk)
20:01, 24 November 2022 (UTC)
No. Read "500" as "up to 500". It's a constantly changing database so any promise, such as next 67 edits, cannot be accurate because pages might be added or deleted at the time you push the button.
Johnuniq (
talk)
23:33, 24 November 2022 (UTC)
Could anyone help me understand why the image Bgc3_logo.svg continues to appear on the page
bgC3 after being removed from the code? I have purged the page cache every way I know of. (It is an outdated logo.)
67.180.143.89 (
talk)
16:37, 26 November 2022 (UTC)
The infobox template pulls it automatically if in Wikidata. It needs to be updated (or deleted) there.
MB16:47, 26 November 2022 (UTC)
Thanks, but deleting the empty short description is only a workaround. Could someone edit the annotated link template so that it adds nothing from an article if its {{Short desc}} template is empty? --
Espoo (
talk)
15:22, 26 November 2022 (UTC)
@
Espoo The whole point of the template is to add a short description, I added one to the article but if it isn't a good description then the best way to deal with this is to just wikilink it ([[Caricatures of Charles Darwin and his evolutionary theory in 19th-century England]]) Since there is no real point bypassing the core functionality of a template in order to just use the template.
Terasail[✉️]15:26, 26 November 2022 (UTC)
OK, thanks. I guess we'll just have to live with this kind of confusing description sometimes existing for many months or years (like
here since 2018) after being accidentally added when people change many wikilinks to annotated link templates without checking the results. --
Espoo (
talk)
15:37, 26 November 2022 (UTC)
This has nothing to do with the English Wikipedia; following the directions at the top of that page also appear to have resolved the problem. —
xaosfluxTalk17:49, 27 November 2022 (UTC)
Character only exists in lowercase; blue link leads to nonexistent page
At
Wikipedia:Redirects for discussion/Log/2022 November 24 I linked to
ΐ, which is a blue link to a page that doesn't exist. The character has no capital equivalent so combining characters are used, and the link has the iota followed by two combining accents (U+0399 U+0308 U+0301), which is a redirect to
Greek diacritics#Diaeresis. Click on the link and it leads to a title (U+03AA U+0301) where the separate iota and diaeresis are combined as "Greek capital letter iota with dialytika".
Ϊ́ (U+0399 U+0308 U+0301, which is the redirect's title but actually links to U+03AA U+0301) is a red link. The redirect existed since 2005 and was at the lowercase title until 27 October when it was moved by
User:Maintenance script. Is this a known bug? Also can this redirect (and others affected by this change
[31]) be moved?
Peter James (
talk)
20:23, 27 November 2022 (UTC)
Further on that, there is not a "redirect" (anymore than there is one from
k to
K) so no on-wiki process or request is going to be able to change that. —
xaosfluxTalk20:46, 27 November 2022 (UTC)
@
Masoud.h1368 well to answer your technical question, yes anyone should be able to make it. (That doesn't mean it is a good idea). If you get an error when attempting, please let us know more details. —
xaosfluxTalk22:45, 27 November 2022 (UTC)
Incorrect article count on the Wikipedia landing page
Something seems to have broken on
https://www.wikipedia.org/. For some reason the article count for the English Wikipedia is displayed as 657,000+, it should be 6,579,000+. Does anyone know who is responsible for maintaining this page?
163.1.15.238 (
talk)
15:54, 25 November 2022 (UTC)
Mobile Wikipedia front page has wrong article count
The mobile view home page of Wikipedia (Wikipedia.org) seems to be giving the wrong article count for the English version - 657k articles instead of 6.57 million. Not sure whether this is the right place to report it but it's been like it for at least a few days.
YFB¿00:22, 28 November 2022 (UTC)
There must be an error in the calculation of days in the software infobox.
Please see this article.
Notepad++. As of today November 28, the stable release date is November 8, so it should be '20 days ago', but it describes it as '12 days ago.'
The problem is due to caching (see
Wikipedia:Purge#Server cache) and is unavoidable. A few minutes ago, the infobox at
Notepad++ said "Stable release 8.4.7[1] 8 November 2022; 12 days ago". I fixed that by editing the article and saving it without making any change and with no edit summary. It now says "20 days ago" because the wikitext was freshly interpreted to generate the HTML web page that we see. Try doing that at the Korean page and you will see the number update.
Johnuniq (
talk)
05:40, 28 November 2022 (UTC)
What
Johnuniq describes is a
WP:NULLEDIT - whilst this will certainly work, a
WP:PURGE on each of the affected pages should be all that is necessary. Note that whether you nulledit or you purge, the effect is only correct until midnight (UTC), it won't keep on adjusting the "days ago" figure going forwards. --
Redrose64 🌹 (
talk)
10:09, 28 November 2022 (UTC)
Tech News: 2022-48
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you.
Translations are available.
Recent changes
A new preference, “Enable limited width mode”, has been added to the
Vector 2022 skin. The preference is also available as a toggle on every page if your monitor is 1600 pixels or wider. It allows for increasing the width of the page for logged-out and logged-in users.
[32]
Changes later this week
The
new version of MediaWiki will be on test wikis and MediaWiki.org from 29 November. It will be on non-Wikipedia wikis and some Wikipedias from 30 November. It will be on all wikis from 1 December (
calendar).
Some wikis will be in read-only for a few minutes because of a switch of their main database. It will be performed on 29 November at 07:00 UTC (
targeted wikis) and on 1 December at 07:00 UTC (
targeted wikis).
Mathematical formulas shown in SVG image format will no longer have PNG fall-backs for browsers that don't support them. This is part of work to modernise the generation system. Showing only PNG versions was the default option until in February 2018.
[33][34][35]
In
Shepseskaf (current
WP:TFA), {{Infobox pharaoh}} uses U+2640♀FEMALE SIGN and U+2642♂MALE SIGN. In my mobile view, both characters have a whitespace added, appearing like "(♀ )".
These edits did not resolve the issue. Bug does not reproduce when checking through desktop-pagebottom m-link [en.m.wikipedia.org/wiki/Shepseskaf]. Something to learn for me?
DePiep (
talk)
05:53, 29 November 2022 (UTC)
@
DePiep: I don't see a problem in either revision. Perhaps it's an issue with your setup? I'm using Windows 10, Firefox 107, MonoBook - which OS, browser and skin are you using?
To be clear: it only happens on my mobile: iOS 12.5, via safari not wiki-app; I don't know which skin (default=Vector 2010?). It does not happen on desktop, and not when looking via desktop-mobile-link (at very bottom of article page).
Having read TheDJ below: quite possible. So I propose to drop it, not worth much time.
Hello,
Is this the right place to ask if anyone knows whether the
Rater script is functioning? I can’t load the script (I did bypass the browser cache). Thanks, — chat O'Brien 20:13, 29 November 2022 (UTC)
I've added two talk page discussions in the past 10 minutes and the first try to post results in a dialog box that evilly informs me I've been blocked from editing. Clearly I haven't, so what's the bug?
Elizium23 (
talk)
13:33, 29 November 2022 (UTC)
Chiming in to say that I too kept experiencing this issue earlier. The most perplexing part of it was when I'd use Twinkle to tag a page with a CSD tag. In doing so, Twinkle carries out 3 actions and 1 of them would sometimes fail. I'd have it where a single action of the three would fail and the next one would immediately succeed. I got the "You've been blocked from editing" message probably about 10 times within a 45 minute window (7:45am to 8:30am ET), and was able to continue editing immediately after. I got busy and stopped editing so I'm not sure if it's still an issue.
Hey man im josh (
talk)
14:52, 29 November 2022 (UTC)
CU goggles on: Elizium is not editing from a blocked range at all, so this shouldn't be happening to them (though I guess I'm not sure whether trying to take an action and a block preventing it would create a CU log entry). Hey man im josh is on a range subject to an anon-only block. Neither should be seeing this kind of message, far as I can tell.
GeneralNotability (
talk)
15:05, 29 November 2022 (UTC)
Earlier this morning I Experienced a bug where I was briefly blocked from editing because I was somehow on an IP that was a WMF owned Proxy (I looked at the Whois and it actually said it was from the WMF). I checked my IP to make sure it hadn't changed and also my VPN and "VPN" (Opera's Proxy) to make sure I hadn't had them on by accident. I tried again and it was fine. ―
Blaze WolfTalkBlaze Wolf#654515:10, 29 November 2022 (UTC)
I do but not on the PC I'm currently editing on. I had the WHOIS for both the specific IP and the range (it's an IPv6) open on my PC back at my house. I've closed both of them but I can bring them back up (thanks history!) so I'll be able to give it to you in a few hours. Also you can blame the Proxy bot for blocking it. ―
Blaze WolfTalkBlaze Wolf#654515:17, 29 November 2022 (UTC)
phab:T324018 - There was some kind of server misconfiguration which meant some edits were seen as coming from internal WMF server IP addresses instead of the editor's actual IP. A steward blocked these IPs to prevent vandalism, but initially made it a hardblock (affecting logged-in users) before correcting to anon block only. In any case the servers with the problem have been removed from use, so the problem should be fixed now.
the wub"?!"15:34, 29 November 2022 (UTC)
ST47ProxyBot has problems. Please see
[39]. Apart from the serial blocking, this is not a P2P proxy, the IP belongs to the New York Public Library, which is actually an AS (ASN AS35999). Other IPs from the same AS were also subjected previously to inexplicable blocks. I believe I had left a note to the bot operator a year or 2 ago. Apparently, the AS was misreported as a proxy by some mysterious API. The particular interface is still misbehaving.
65.254.10.26 (
talk)
02:04, 30 November 2022 (UTC)
The 2017WTE's find function finds the |2019 (in its regex mode) with \|2019. What do I need to type after that to pick up the next line?
WhatamIdoing (
talk)
00:11, 29 November 2022 (UTC)
That 'find' string doesn't find anything in the 2017WTE. You can
try it at this link. Find and replace is at the bottom of the hamburger menu near the big blue button. Click the (.*) icon (and make sure that case sensitivity/the Aa button is off) to enter regex mode.
WhatamIdoing (
talk)
00:31, 29 November 2022 (UTC)
That link just takes me to
List of MDPI academic journals There is no hamburger; there is no big blue button; there is no (.*) icon. I gather that 2017WTE isn't the generic wikitext editor. That editor (Toolbar > Advanced menu > Quizzing glass icon, finds for me. Check the 'Treat search string as a regular expression' checkbox. If the regex doesn't work in 2017WTE, that suggests that regex in 2017WTE doesn't work as it should.
(Trappist, since ?veaction=editsource didn't work for you but the 2010WTE toolbar is still visible to you, I assume you have the visual editor completely disabled in your prefs.)
WhatamIdoing (
talk)
01:34, 29 November 2022 (UTC)
[\r\n]+ matches 1 or more occurrence of carriage return or newline. I don't know whether a saved wikitext page can actually have a carriage return instead of a newline. You can just say \n to match a single newline. See more at
WP:REGEX.
PrimeHunter (
talk)
13:28, 29 November 2022 (UTC)
That's what I thought, but \|2019 picked up the first line, and \|2019\n didn't pick up anything. I did enough poking around to discover that there are different flavors of regex, but I don't know whether that's what's going on, or if I need to file a bug against the 2017WTE.
WhatamIdoing (
talk)
17:26, 29 November 2022 (UTC)
Not true. The 2010WTE can insert signatures in the mainspace, which the 2017WTE won't let you do. The 2010WTE's citation-filling toolbar is IMO slightly worse overall, but it lets you add a ref name in the dialog, which the 2017WTE doesn't. The 2017WTE also doesn't allow you to set redirects or add categories (except for typing the wikitext by hand, of course).
WhatamIdoing (
talk)
20:01, 29 November 2022 (UTC)
I don't think Nardog meant it's the only thing. It appears from
phab:T151671 that the 2017 wikitext editor can simply not do any regex on multiple lines and you have to use another method, e.g disable it or log out to get the 2010 editor.
PrimeHunter (
talk)
20:09, 29 November 2022 (UTC)
I'm confused. If one is to believe the indenting, Editor
Nardog replied to your post wondering about a regex to find/include a newline. That reply says nothing about the 'other differences' between the 2010 and 2017 wikitext editors. But in your reply to Editor Nardog, you have enumerated a whole list of those differences – none of which, so far as I can determine, relate to regex searches. How do those differences apply to the issue of 2017wte regex searches?
If, as it appears from comments in this discussion, 2017wte can't locate newlines using regex, then 2017wte is indeed broken and should be fixed and Editor Nardog did not misspeak.
One other thing you might try is (\| *2019\s+\|) – assumes that in 2017wte regex, newlines are considered to be whitespace.
Nardog said "This is one thing the 2010 editor can do that the 2017 editor can't". It sounds like WhatamIdoing incorrectly interpreted it as a claim that it's the only thing, and then gave random unrelated examples of other things the 2010 editor can do that the 2017 editor can't. (\| *2019\s+\|) doesn't work either. regex in the 2017 wikitext editor is only able to work on one line at a time, apparently because it's difficult to implement otherwise with the way the editor is coded. It will not work to try to guess at a syntax. There is intentionally no solution.
PrimeHunter (
talk)
20:54, 29 November 2022 (UTC)
Yes, I thought he meant it was the only thing. They're close to feature parity, but not quite the same.
It's also pretty easy to disable that functionality, by making coord_wrapper in the Module:Coordinates return an empty string. But having geodata is kinda nice in that in the future this should also allow you to make use of the Nearby functionality in Kartographer, that should become available before the end of the year. —
TheDJ (
talk •
contribs)
14:18, 1 December 2022 (UTC)
Infobox assistance
At the moment, classical composers with infoboxes use the "notable works" parameter to put their "list of composition"—e.g. the articles for
Beethoven and
Bach—as per the documentation which says Link to "List of works" subarticle here. Do not list individual pieces. This creates the awkward implication that everything they wrote was "notable".
If what you want to is change the label in the rendered infobox, all that you have to do is change the label in {{
infobox person}}. That change isn't difficult but, before you make that change, you should establish a consensus to make the change because {{infobox person}} is used in some
450,000+ articles.
I still think that consensus is required from the classical music community because |works= (according to {{
infobox person}} documentation) overrides |notable works= and |notable_works=.
@
Plantaest no, the software does not support that. There is a
request for a tool to output the CIDR ranges for an ASN to make this easier - but nothing in the works to directly block by ASN. —
xaosfluxTalk01:19, 2 December 2022 (UTC)
I've blocked all IP ranges of Selectel and Clouvider according to two list from Hurricane Electric:
I was aware, so I put the
Main Page in and it didn't do anything so I assumed it was broken. In retrospect it makes sense the main page wouldn't have those properties. Although it would be nice if the #property parser function showed a warning when the property did not exist in the linked item.
RPI2026F1 (
talk)
14:03, 2 December 2022 (UTC)
Some templates use #property and if it started outputting error messages rather than nothing it could break these templates
Terasail[✉️]14:10, 2 December 2022 (UTC)
@
RPI2026F1 I disagree about the warnings/error messages, they would be a massive nuisance and would get in the way. If you want to create a wikidata based template currently, you can just take an existing template and fill in the parameters with #property statements, the standard "display this if it has input" template code already in use deals with everything else. If property statements started outputting errors every single one of the #property parser functions would need to be contained within error handling code (e.g. a #iferror switch) to turn any errors into blank strings.
163.1.15.238 (
talk)
14:18, 2 December 2022 (UTC)
I've attempted to set up auto-archiving on
Talk:Michelle Kwan, but it's not working: for whatever reason, it's not attracting the attention of
Lowercase sigmabot III.
I've never had problems setting up archiving before. I've cleaned up/added missing signatures; made sure the page is saveable (i.e., no blacklisted-domains); and pretty much everything else listed on
Causes of no archiving; no dice.
Now that I think of it, is there any reason that bots aren't exempt from the spam blacklist? I would assume it's either a technical limitation or simply something that isn't feasible to do in case the bot account is compromised. ―
Blaze WolfTalkBlaze Wolf#654523:39, 1 December 2022 (UTC)
Admins aren't exempt either. It would be impractical to allow some users to save links which cannot be saved by others, e.g. when moving content around or reverting a blanking.
PrimeHunter (
talk)
00:23, 2 December 2022 (UTC)
I acutlaly discovered this a while back when I was attempting to copy
Slime Rancher to my sandbox, only to be stopped because one of the links was on the blacklist. ―
Blaze WolfTalkBlaze Wolf#654523:45, 1 December 2022 (UTC)
I wasn't pinging; I was simply acknowledging and thanking them. I generally assume editors who edit in a topic will see responses, including the acknowledgement.
TJRC (
talk)
00:50, 2 December 2022 (UTC)
Edit comments can include wikilinks, which are rendered as links in diffs. Is there anyway to see the raw comment, before links are rendered? I'm sure I could figure out a database query to dig it out, but I'm looking for a way through the normal web interface.
PS, what I actually want is Possible RS addition? - prowrestling.net for the fragment, i.e. before the spaces were mapped to underscores. --
RoySmith(talk)14:37, 3 December 2022 (UTC)
How can I convert a date from a year-first all number form (20230102) to a normal date textual form using month names (January 2, 2023) in Lua?
93.172.247.76 (
talk)
10:12, 4 December 2022 (UTC)
Cast to string, insert dashes between characters 4 and 5, and 6 and 7, then use os.date or mw.language:formatDate.
Izno (
talk)
10:17, 4 December 2022 (UTC)
I don't know how to represent this in Lua, but {{#time:F j, Y|20230102}} → January 2, 2023 is easier to code. Replace the timestamp with a variable if required. —CX Zoom[he/him](
let's talk • {
C•
X})14:43, 4 December 2022 (UTC)
I wasn't sure if that really Just Worked without insertion of the dashes, though it's nice to know it does.
Izno (
talk)
19:39, 4 December 2022 (UTC)
ClueBot III
Hello, VPT,
I just noticed that ClueBot III has been randomly archiving my talk page and moving around archived talk page material, generally causing havoc. Just look at
this!. I never set up this archiving and I can't find any code on my talk page that I can delete to stop this. ClueBot III's talk page gets redirected to one that is inactive and I don't know how to stop this. It will take a long time to undo all of the chaos to my talk page archives but right now I just want the bot to leave my talk page and talk page archives alone. I feel like someone set this up as a prank and I need it to stop. Thanks for any help you can offer. LizRead!Talk!19:20, 4 December 2022 (UTC)
The origin of the problem was
[44] where
Einahr posted archive instructions in a named section. After the section was archived, it was interpreted as instructions to archive the archive.
PrimeHunter (
talk)
21:49, 4 December 2022 (UTC)
Sorry if I'm in the wrong place, I don't know what the problem is or where to report it.
List of commelinid families uses this template several times, and it works perfectly for me, cropping the images on my laptop (Monobook and Firefox, if it matters) and my iPhone 12. (I can reproduce one of the pictures here if you like.) At
WP:Featured list candidates/List of commelinid families/archive1,
User:Eewilson is reporting that it doesn't work at all on her iPhone 8. Is this a known bug? Is there a different template I can use that might fix the problem? - Dank (
push to talk)
00:43, 4 December 2022 (UTC)
Yeah, tricks like this often work differently or not at all in different formats like print, mobile, and especially apps. Its one of the reasons why it was never adopted more widely. —
TheDJ (
talk •
contribs)
07:52, 5 December 2022 (UTC)
I'm trying to make a treemap using the MediaWiki graph extension. I copied a JSON specification from
here and modified the data, but there are errors when it tries to render on Wikipedia; however, it works fine in the
Vega editor. The code is
here. What do I need to do to fix this? Qzekrom (she/her •
talk)00:00, 5 December 2022 (UTC)
The version we run of Vega is old(er). It may also have only limited support for graphs imposed on it.
Izno (
talk)
00:04, 5 December 2022 (UTC)
We are using Vega 2.5.2. The style you are trying to use, stratify is not in that version. It would have to be simplified quite agressively, to the point it will not look remotely the same.
Snævar (
talk)
00:17, 5 December 2022 (UTC)
According to
the documentation: "the stratify transform generates a hierarchical (tree) data structure from input data objects, based on key fields that match parent and children nodes. Internally, this transform generates a set of tree node objects that can then be processed by tree layout methods such as tree, treemap, pack, and partition." So basically, this version of Vega can't generate treemaps at all. Qzekrom (she/her •
talk)06:33, 5 December 2022 (UTC)
When I try to save edits (in edit mode not edit source mode) I get an error message "Something went wrong: Error contacting the Parsoid/RESTBase server (HTTP 404): (no message)" Any ideas?
Jonathanlynn (
talk)
16:44, 2 December 2022 (UTC)
@
Jonathanlynn It's unfortunately a known issue, VisualEditor can't save edits when it was open for more than 24 hours while editing an old revision or after switching from wikitext:
phab:T235822. If you still have that tab open, you should be able to copy-paste from it into another tab with the editor open (make sure it does not restore your auto-saved edit – if it does, close the editor, discard your changes, and open it again), and save that way. Make sure to review your changes when doing this, since you won't get an edit conflict screen, so you might accidentally undo someone else's changes to the page.
Matma Rextalk16:56, 5 December 2022 (UTC)
Many thanks for the explanation: I'll know next time. I'm afraid I can't copy/paste in this case as it was a series of citations/refs.
Jonathanlynn (
talk)
17:05, 5 December 2022 (UTC)
This award is highlighting software tools that have been nominated by contributors to the Wikimedia projects. The ceremony will be a nice moment to show appreciation to our tool developers and maybe discover new tools!
Could someone pls have a look at
Talk:Square Enix Montreal and the article history error, resulting from
Bearcat removing a featured topic (which seems to be a valid featured topic, but populating a non-existent category)? When I reinstate the Featured topic to remove the article history error, Bearcat reverts to remove the non-existent category, which is not something I know how to fix (nor do I know from where it comes). The talk page is now populating
Category:Article history templates with errors, and best I can tell, the featured topic is still valid, so I don't know what to do next.
SandyGeorgia (
Talk)
22:51, 2 December 2022 (UTC)
I haven't fixed it per se as the issue is that
Category:Wikipedia featured topics Square Enix Montreal good content was deleted in the short time the |ftname= parameters were changed to "Onoma". The category was not moved alongside the topic, so the category became empty and was deleted as a consquence. I requested the person who deleted the category to restore it, and I reset the parameter to its original value on the affected articles. As was typing out this message, the category has been orderly restored, so indeed the issue is now fixed.
IceWelder [
✉]
22:58, 2 December 2022 (UTC)
The problem is that redlinked categories are always forbidden. That said, the actual error appears to be that at one time the category did exist, but a couple of times over the past several weeks editors have tried to change "ftname=Square Enix Montreal" to "ftname=Onoma" during a recent back-and-forth dispute about the name of the article — which caused a cascading error because (a) the resulting "Onoma" category did not exist and had to be removed, and (b) the original "Square Enix Montreal" category got emptied out, and therefore got speedy deleted as an empty category, so that it didn't exist either when I originally tried to revert the ftname= change. So there was an error here, but it wasn't by me — my job when I'm cleaning up redlinked categories is "make non-existent category go away and move on", not "spend half an hour investigating whether there's some extended history of fuckery behind the category's failure to exist" — but the problem has thankfully now been fixed, and hopefully won't recur again.
Bearcat (
talk)
23:10, 2 December 2022 (UTC)
I appreciate the tireless and often thankless work of experienced editors in this area. As you know, per
WP:REDNOT, Either the category should be created, or the non-existent category link should be removed or changed to one that already exists. Removing the category link is often the right solution, but sometimes it may be better to create or move the category itself instead. Even when the category link should be removed, the edit which added it may make other changes which we wish to keep, rather than reverting a baby along with its bathwater.
Certes (
talk)
11:08, 3 December 2022 (UTC)
Well, in order to create the category, one first has to know that a category is warranted and legitimate in the first place (lots of redlinked categories absolutely aren't), and even then one has to know where in the category structure it would actually belong. In this instance, for example, creating the category would require me to already have preexisting knowledge that "Wikipedia featured topics [something something] good content" is a category tree that even exists in the first place, and where to actually file such a thing even if I did intend to create it myself — but since I don't typically work on the "featured topics" project at all, there's no real reason why I could really have been expected to already know either of those things.
As
Special:WantedCategories invariably lists several hundred redlinked categories to be cleaned up at any one time, it would take days and days to deal with the list if I actually had to invest extended amounts of time into researching the history of each individual redlink — and then by the time the list was actually completed, a new run would already have been produced with several hundred more redlinked categories to clean up, meaning I would never have any time left to do anything else. And, in addition,
Special:WantedCategories has a list-size cap on it, meaning that if it hits that cap it stops detecting additional redlinked categories at all — so just leaving some categories there as unresolved entries isn't an option either, because every time I skip over a category as "I'm just not dealing with this one at all" I push the list closer to that cap.
So I have to be able to fix every individual redlink in the absolute quickest in-and-out fashion possible, which means that I simply can't be expected to invest extended amounts of time into each individual entry — if I don't already know that the category is appropriate and what its appropriate parent categories are right off the top of my head, in order to create it right away without having to invest any additional research into it, then I can't create the category, but I can't just leave it on the list as an unresolved problem either, meaning that removal is the only other option left.
Really, the principal issue here is that "Wikipedia featured topics" categories should probably have the {{Possibly empty category}} template on them, so that if and when something like this happens they don't get speedy-deleted before the problem gets repaired. Of course, it would also be nice if there were a way to make
Special:WantedCategories separate mainspace categories from projectspace categories, so that they could be dealt with separately instead of getting in each other's way, but I grant that may not be possible.
Bearcat (
talk)
16:50, 3 December 2022 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you.
Translations are available.
Recent changes
The Wikisources use a tool called ProofreadPage. ProofreadPage uses OpenSeadragon which is an open source tool. The OpenSeadragon JavaScript API has been significantly re-written to support dynamically loading images. The functionality provided by the older version of the API should still work but it is no longer supported. User scripts and gadgets should migrate over to the newer version of the API. The functionality provided by the newer version of the API is
documented on MediaWiki.
[45][46]
Changes later this week
The
new version of MediaWiki will be on test wikis and MediaWiki.org from 6 December. It will be on non-Wikipedia wikis and some Wikipedias from 7 December. It will be on all wikis from 8 December (
calendar).
It's probably not a cite.php problem; I'm probably not seeing something which should be obvious to me, but see
this edit and then look at the problem it fixed in the previous article version.
Wtmitchell(talk) (earlier Boracay Bill)
23:32, 5 December 2022 (UTC)
@
Wtmitchell, one of those quote marks was "curly" rather than straight, conflicting with the actual name in the main reference definition. Very hard to see without expanding the magnification.
StarryGrandma (
talk)
00:00, 6 December 2022 (UTC)
‹The
templateCategory link is being
considered for merging.›Category:Contents describes itself as "the top level of Wikipedia's category system (which is why it has no parent category)", but the lack of a parent category means that it just permanently sits on
Wikipedia:Database reports/Uncategorized categories as a perennial speed bump that has to be worked around and can never be cleared. So I wanted to ask, is there any way that it can be flagged as an exception to that report, since it doesn't need to be on a cleanup list if it doesn't require any cleanup? I know there must be, because
Main Page used to get regularly detected as an uncategorized article, but some solution was found to stop that from happening -- so if that category's lack of a parent category is intentional, then can something be done to keep it off the tool?
Bearcat (
talk)
18:18, 3 December 2022 (UTC)
@
Bearcat: sure, that's possible. Does @
BD2412's suggestion work to address the problem? Otherwise we can update the report to ignore that category I guess. (In the future feel free to make these kinds of requests on
WT:DBR.)
Legoktm (
talk)
02:02, 6 December 2022 (UTC)
I assume that would probably work, but I don't know for sure: for example, is it a hidden category, and is the report programmed to account for hidden categories or not? That's for people with more technical expertise than I've got to determine — but if it would work, I don't see an issue.
Bearcat (
talk)
18:14, 6 December 2022 (UTC)
Despite using RedWarn, this edit shows the
WP:ROLLBACK tag
Could you show me where you found it? I think the tool has been deprecated and is broken and no longer maintained. ―
Blaze WolfTalkBlaze Wolf#654515:20, 7 December 2022 (UTC)
I used it frequently in the year 2017 (and maybe in 2018), it used to work fine for me. —usernamekiran
(talk)19:34, 7 December 2022 (UTC)
Private wiki with database access
Yesterday I spoke with representatives of the
Ukrainian Book Institute. They dream of creating their own wiki, in which part of the information would be automatically added to the articles from their database, which has been filling by publishers and verifying by the institute's employees. Such a base has already been created and is functioning. The database is accessed through an application programming interface (API).
It could be an interesting project and a reliable source for Wikipedia articles about writers, editors, illustrators, their works, as well as about books, exhibitions, literary awards, etc.
Are there programmers in our community or representatives of companies who are able to implement such a project, or create at least a prototype? What are the difficulties in implementing the project? --
Perohanych (
talk)
13:12, 7 December 2022 (UTC)
Why am I seeing a "review revision" box at the bottom of the page?
I am seeing a "re-review revision" box at the bottom of
Communication. The box refers to a 13:37, 22 Feb 2021 logged action by
User:Samsara, has an "Unaccept revision" button, and an input field to enter an edit summary in. It appears below the list of categories, and above the "last edited" and licensing information at the bottom of the rendered page. This box is visible in rev.
1125980619 of
Communication. Here's a capture, in case it disappears later:
Snippet of page source Html from bottom of the page
There was a problem at {{Pages with authority control identifiers}} with a Lua module, with the result that earlier today a user commented out a piece of the template coding pending resolution -- but their change has had the entirely unintended effect that maintenance categories such as ‹The
templateCategory link is being
considered for merging.›Category:Articles with VIAF identifiers, which are supposed to be hidden as they're internal tracking categories for maintenance purposes rather than end-user content categories, are starting to display as visible categories on some articles. It was clearly that change that did it, because I did a test edit it to see what happened if I removed the hidden comment tags, and it made the maintenance categories disappear off the article I was looking at.
Accordingly, since the hidden maintenance categories had to be rehidden and thus the broken module couldn't be commented out of the template, can somebody look into how to fix the Lua module so that the "Lua error in Module:Pages_with_authority_control_identifiers at line 31: bad argument #1 to 'pairs' (table expected, got nil)." error message goes away? Thanks.
Bearcat (
talk)
16:11, 8 December 2022 (UTC)
It's a bit more complicated than that. I have changed all the tables to use named indices and the functions which produce the links have all been changed. Not difficult though. — Martin (
MSGJ ·
talk)
17:19, 8 December 2022 (UTC)
Template-generated categories that don't exist, redux
Several issues with templates autogenerating nonexistent categories again...
‹The
templateCategory link is being
considered for merging.›Category:Historic districts in United States San Francisco County is being autogenerated by the use of "United States San Francisco County" as a locator map in the {{Infobox NRHP}} template on two articles. But it can neither be created (it's an obviously ridiculous naming format that no other Wikipedia category uses at all) nor just left there as a redlink, and has to be removed, but I don't know how to do that without messing with the locator map. And even more importantly,
WP:TEMPLATECAT explicitly forbids using templates to autogenerate categories based on variables, precisely because that can lead to the generation of nonsense categories that don't and rightly shouldn't exist if any sort of mistake happens in the template coding. So this particular category certainly needs to go away, but the infobox template also needs to lose its ability to autogenerate categories based on what maps happen to be used in locmapin=, so that categories like this can't happen again in the future.
The tracking categories generated by {{Single chart}} were recently renamed by CFD discussion from "Singlechart usages" to "Single chart usages" -- but because changes to templates sometimes delay in actually propagating through the system,
Special:WantedCategories is now polluted by about 100 categories of the old "Singlechart" format, many of which are still populated by hundreds or thousands of articles -- and even though the categories were renamed three days ago, the categories are only very slowly depopulating (as witness the fact that so many of them still have hundreds or thousands of articles after three days.) Is there a bot that can be set loose to null the old categories out of the way faster so that these aren't lingering for weeks as a speed bump that the categorization project has to work around? And is there any way that this can be prevented in the future so that things like this don't happen again?
There's also a recurring problem with roller coaster categories; when an existing roller coaster closes for renovations, people editwar over whether the status should be denoted as "closed" or "under construction", but since "under construction" in the infobox autogenerates a "Roller coasters planned to open in [whatever year is specified in year=]" category, that leads to recurring redlinks at WantedCategories of the "Roller coasters planned to open in [20-50 years ago]" variety. Again, this comes down to the fact that templates aren't supposed to be autogenerating categories based on variable plug-ins at all, so one way or another the infobox needs to lose the ability to generate categories that don't and shouldn't exist, but again it's beyond my ken to fix — but one way or another, WantedCategories needs to stop seeing these at all.
I've now gone through the NRHP orphans and added categories to them where applicable. In no case was the category I added the same one as the category previously generated by the template. (usually I added "Historic districts on the National Register of Historic places in foo", whereas the template generated just "Historic districts in foo".
* Pppery *it has begun...20:43, 8 December 2022 (UTC)
I believe I've fixed the issue with {{infobox settlement}} populating categories with commas in
Special:Diff/1126293153. For #3, at one point we had ProcBot approved to do this, but that task has been disabled for months. Rather than creating redlinked categories and requiring them to be manually fixed, I think the right behavior for JJMC89 bot III should be when moving a category to first try to move all pages to the new category, and leave a redirect when moving the category page itself if it fails to move at least one page. (I was going to suggest that bot automatically null editing the pages it can't find anything else to do with, but that would repeatedly null edit the same page every 15 minutes). Does that make sense,
JJMC89?
* Pppery *it has begun...15:39, 8 December 2022 (UTC)
Due to issues like this,
WP:CFD/W (and therefore the bot) is not supposed to be used when template editing is needed. If the category needs to be split among multiple destination categories, requires template editing, or requires editing the documentation subpage of templates, or any other special circumstances that require manual review, list it at
Wikipedia:Categories for discussion/Working/Manual rather than here. When I realized
Timrollpickering was incorrectly using CFD/W, I
disabled the set of moves and manually stopped the bot from processing. Timrollpickering decided to ignore the directions (and me) again and
reenabled it. It is up to the admins working at CFD to abide by the stated restrictions on the bot's use. —
JJMC89 (
T·C)
17:31, 8 December 2022 (UTC)
The correct fix for #4 is to add the categories in the wikitext directly, then remove them from the template. This can mostly be done one by one with AWB for a given set of categories as coded in the infobox. You'd have to work through it by category.
There's no other way to fix that issue fundamentally. As for auto-catting in NRHP, I would suggest that it's easier to fix things if you know what cat they were supposed to be in beforehand, rather than removing the category first....
Izno (
talk)
19:05, 8 December 2022 (UTC)
Here's a less trivial search
for steel coasters, some 500 with a template cat and no local categorization. It's a project, probably best coordinated at the page of interest, that should be done with AWB or similar.
Izno (
talk)
19:17, 8 December 2022 (UTC)
@
CX Zoom If you're happy to do it in two steps, then something like: use PetScan to find the union of categories A plus B, and save it as a pagepile. Then load category A + B into PetScan as union, the pagepile into "other sources", and set the combination to "categories NOT pagepile". I think that should probably do it? You'll get all the pages and then remove the ones known to be in both.
Andrew Gray (
talk)
22:47, 8 December 2022 (UTC)
Another alternative with Petscan is to type one category into "Categories" and the other into "Negative categories", saving the result. Swap the two and repeat. The two lists combined are what you need. For most applications, you'd want to treat the two lists differently anyway.
Certes (
talk)
22:53, 8 December 2022 (UTC)
Category:Articles with MusicBrainz work identifiers
Why am I suddenly seeing this category everywhere? Is this a temporary thing, or is it only visible to registered users? It strikes me as a category ordinarily meant to be hidden or for maintenance. I'm just curious because it definitely doesn't seem
WP:DEFINING, but I've been seeing it for at least a few days now. Zeke, the Mad Horrorist(Speak quickly)(Follow my trail)16:53, 9 December 2022 (UTC)
I made a simple example about making userscript with Vue and Codex:
User:Plantaest/TestVue.js. For testing, need to install the script to your common.js and go to the
Special:BlankPage/Vue page, you will see a simple counter program with two buttons. I think it's easier to build the user interface in Vue/Codex than in OOUI.
Plantaest (
talk)
22:16, 9 December 2022 (UTC)
Weird view for Non-free media information and use rationale on mobile view
Today I found the summary for non-free images becomes weird when I see the page on my phone, the title of each field aligned to right side, but the content of field aligned to left. For example, when you look at
File:Aerojet Rocketdyne Holdings logo.svg, you'll see:
Description
This is the logo for Aerojet Rocketdyne Holdings, inc.
I suggestion to write a TemplateStyles for fix, just reverse their text alignments for mobile phone screen. Do you have any other idea to improve? --
Great Brightstar (
talk)
17:06, 9 December 2022 (UTC)
The correct solution is to fix
MediaWiki:Filepage.css and then remove the inline styles, at least until such time as the underlying template(s) use TemplateStyles. I haven't
focused on fixing these because of the existence of that MediaWiki page (and also I didn't know about it until a few months ago). There may be many Templates affected by the change, including ones that don't live on Wikipedia but instead on Commons.
Izno (
talk)
19:15, 9 December 2022 (UTC)
Commons has some 3.7k uses of the class licensetpl or a relative and
we have some 300. If we look just for the specific class,
120 here and
210 there. The alignment might be treated in one of the other apparently common classes associated with that item (there seem to be a few), which you can consider searching for yourself with those searches as a baseline for use.
Izno (
talk)
23:15, 9 December 2022 (UTC)
Quarry of my page moves that renames a page with a certain suffix
About 2 months ago, I attempted to marry all orphaned talk page archives with its parent page. In cases, where the parent page has gotten an "/Archive 1" and more of its own, I named the oldest ones as "/Archive A" & "/Archive B", as {{talk header}} perfectly supports this system. Sadly, most other scripts and templates (including {{automatic archive navigator}} etc.) do not support this system. Thus, I want to run a quarry to find all the pages that are eventually renamed into "/Archive A" & "/Archive B" suffixes, to remarry them using the most widely used system. Can someone please write this quarry, as I myself don't know how to? Thanks! —CX Zoom[he/him](
let's talk • {
C•
X})17:19, 3 December 2022 (UTC)
Here are the 30 pages called Talk:Something/Archive A, B, etc. They may not all have been renamed, but there should be few enough to check manually. There are several /Archive As, a few /Archive Bs and a suspicious number of /Archive Is which might be a typo for 1 (or a Roman numeral).
Certes (
talk)
19:59, 3 December 2022 (UTC)
No, I don't want redirects in results. I just moved it to the standard title, so it became a redirect. So, maybe it was the only such archive then. —CX Zoom[he/him](
let's talk • {
C•
X})15:13, 6 December 2022 (UTC)
I've noticed that the "subscribe" link is not present on XfD threads for files, categories and redirects, eg
Wikipedia:Redirects for discussion/Log/2022 December 9. These pages lump multiple discussions on to one page. Subscribe would be outstandingly useful there. Is there a way to force a subscribe? It is clearly possible to have it on non-talk pages. It's here at the VP for instance and
Wikipedia:Files for discussion has some level 2 headings subscribeable but not others. However, the threads of interest are at level 3 or level 4.
SpinningSpark08:54, 10 December 2022 (UTC)
Hi, this is a really stupid and minor issue, but regarding the section title, the Feedback sentence that is automatically added to talk pages contains an extra period, whereas the message posted to a user's talk page does not contain an extra period. Or maybe it's the other way around, the user talk page message is missing the period. I'm sorry if this isn't the correct place to take but I don't really know where else to take it. If you want to see what I'm talking about, pop on over to
Talk:Praga Bohema and
User talk:X750#I have sent you a note about a page you started.
X750. Spin a yarn?Articles I've screwed over?00:34, 11 December 2022 (UTC)
Hi everyone, I'm trying to deorphan all those talk page archives on Wikipedia, but I've bumped into
Talk:44/Archive 1, which I'm unable to understand what to do with. If you read the content, you'd think that this is from
Talk:+44 (band), but the latter's page history doesn't indicate that. The archive was created by Werdnabot. If you look at
Werdnabot's contribs on 15:37, 18 February 2007, it actually came from
Talk:AD 44. But that itself was created by HagermanBot with the edit summary "172.142.30.213 didn't sign...".
Special:Contribs/172.142.30.213 indicate that it had actually edited
Talk:+44 (band). So, we are now back a full circle. There had been so many page moves, history merges and history splits, that nothing makes sense now. So, I need help to take it to its correct location. Thanks! —CX Zoom[he/him](
let's talk • {
C•
X})15:29, 11 December 2022 (UTC)
A plus sign in URLs means space, and filtering out initial spaces is not unreasonable. I expect a bot has become confused and turned the plus into a space before dropping it. Alternatively, it might have evaluated it as an arithmetic expression equal to 44. Either way, it's an archive of Talk:+44.
Certes (
talk)
15:44, 11 December 2022 (UTC)
Hi, a few months ago
Wikipedia:Village pump (technical)/Archive 201#Watchlist complexity I posted about slowdowns, possibly due to Watchlist use. I'm still being plagued by UI lag, Chrome hangs often when I'm using Wikipedia, and only when I'm using Wikipedia. I don't have issues if I quit Wikipedia, when I am watching YouTube, or normal browsing, and I have a separate account for work that doesn't have any slowdowns. But Wikipedia alone is definitely taxing on my poor system now. I've been through a few updates (I tried to stave off this slowdown by uninstalling updates, but Windows insisted on reinstalling them daily.)
Elizium23 (
talk)
23:56, 8 December 2022 (UTC)
A page has many things that can influence how it renders. As you are the ONLY person reporting this, usernamekiran is trying to help you eliminate possible causes by changing the environment you use in order to pinpoint which influence is triggering the behaviour. Have you tried disabling the filters/setting them back to default ? Can you try loading
/info/en/?search=Special:Watchlist?safemode=1 (this loads the page without gadgets enabled, which. might be getting in the way). —
TheDJ (
talk •
contribs)
10:42, 9 December 2022 (UTC)
If it goes away, then we could proceed to test things that are different between the two browsers (e.g. browser extensions you have installed) to try to determine the cause. If it doesn't, then we could test things that are the same (e.g. Wikipedia gadgets you have enabled). Perhaps we'd find the cause, and then maybe fix it, or at least find the only feature you'd have to give up on.
Matma Rextalk19:09, 11 December 2022 (UTC)
Since TheDJ's suggestion to run in safe mode, I have done so with my watchlist. But I am unsure how this will isolate any problems, because to work with the watchlist I need to constantly open new tabs of diffs and reverts and discussions, and none of those tabs can be in safe mode.
I have disabled a few non-essential scripts from my common.js, in case those were clogging up the works - I wouldn't be surprised.
Elizium23 (
talk)
18:36, 11 December 2022 (UTC)
I've never had this kind of problem before: I am trying to edit
climate change adaptation with the visual editor but the page just freezes up. It gives no error message but it just doesn't go through to the edit page. I thought it was just a short temporary glitch but it was first reported to me on 8 December by another editor. The source editor works fine for this article. The visual editor also works fine on other articles, just not on this one. Could it be that some hidden code was added perhaps around 8 December which has caused this IT problem? The last time when the visual editor still worked for this page was
here. I have now removed the external link that had been added in that edit, thinking perhaps it caused the IT glitch but so far, nothing has changed - the visual editor still freezes up for this article.
EMsmile (
talk)
13:15, 12 December 2022 (UTC)
Fixed There was an incorrect anchor link on the page, which cause a JS crash. The destination page also no longer contained that anchor, so I removed it for now. —
TheDJ (
talk •
contribs)
13:34, 12 December 2022 (UTC)
I edited two talk page templates to allow use as editnotices, then realized a problem
{{Talk fringe}} and {{Black project}} will now use {{editnotice}} instead of {{tmbox}} when placed in an editnotice (see
diff and
diff). However, I just realized that when it is viewed as an editnotice while editing the article, the user is technically not viewing a page starting with Template:Editnotices/, meaning they will still see a tmbox. I have no way to test these editnotices. Is there anything that can be done about this?
137a (
talk)
15:20, 12 December 2022 (UTC)
The reason the option was removed is because the renderer is going to be removed due to being old, AIUI.
In the case of copy-pasting math formulas, I should think you can just take a screenshot with a snipping program of some sort to support the use case of copy-pasting random *image* of a formula. If someone wants the LaTeX for whatever reason, they can edit the section to get the wikitext directly.
Windows has native snipping with Win+Shift+S. I assume Mac has a similar option, and Linux is Linux and anyone using it should be able to hack some solution together.
Izno (
talk)
22:26, 4 December 2022 (UTC)
Well, yes, a screenshot saves time, you need to make it, then paste it into Word, choose the size, you're kidding, aren't you. Why not just return the PNG, as long as it exists and does not bother anyone, copied what you need, inserted it into a text document and you're done, everything is fast and simple.
217.118.90.169 (
talk)
11:28, 8 December 2022 (UTC)
The point behind the change is that the PNG will not exist into the future due 1) to needed software upgrades and 2) it's a generally inferior image for web uses.
Izno (
talk)
17:56, 8 December 2022 (UTC)
Well, it's terrible, as if by removing the PNG, something has changed, to be honest, I didn't notice anything, EXCEPT that I can't do anything normally now, it's some kind of mockery that you're going to change, it's not clear. And about the quality, well, I can't say anything bad, it was normal when I inserted the formula into a text document, the quality did not change in any way, as it was clearly visible, it remained that way.
I can't understand, PNG has always been here, well, if you don't want, you can switch to another format for displaying formulas, but no, we will remove the convenient format and leave those that cannot be used.
91.148.224.101 (
talk)
20:01, 8 December 2022 (UTC)
Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you.
Translations are available.
The character = cannot be used in new usernames, to make usernames work better with templates. Existing usernames are not affected.
[48]
Changes later this week
The
new version of MediaWiki will be on test wikis and MediaWiki.org from 13 December. It will be on non-Wikipedia wikis and some Wikipedias from 14 December. It will be on all wikis from 15 December (
calendar).
I have noticed that redirect categories in the redirect category shell show up under a header called "Page issues" on the mobile website. But a redirect category doesn't seem to be an actual issue, so is the naming a bug or does the "issue" have a different meaning?
WorldExplorer120322 (
talk)
23:37, 12 December 2022 (UTC)
It’s an informational article messagebox, but all messageboxes are treated the same. This is a balance game between providing the user with “too accurate not understandable to the layman”-labels vs “very rarely, not completely accurate”-labels. —
TheDJ (
talk •
contribs)
23:50, 12 December 2022 (UTC)
background might be overridden by a more specific style from a class. For instance, this sets the background of the row, but the cell elements within the row already set their own color. (Note that || is used here as a cell separator, but if you want your rows not be be actual header cells, you need to start a cell on a fresh line to change the cell type [it this new ? don't remember this being like that before])
Another note, this table uses color to communicate information about a tournament being defunct or not. This is an accessibility problem for people who are (color) blind. You should not use color to communicate information unless you also use text to communicate the same information. —
TheDJ (
talk •
contribs)
10:47, 12 December 2022 (UTC)
it this new ? don't remember this being like that before No, that's ancient behavior. Table row headers have always needed to be on a separate wikitext row.
Izno (
talk)
17:17, 12 December 2022 (UTC)
I have that option checked so when I float my mouse (desktop view) over this link at the ship's name or its hull designator, all that I get is a simple tooltip. I expect to get the same result as I get when I float my mouse over the ship's USS prefix – the page preview tooltip. I don't. Is that me and my browser (win10 chrome current) or is this a known issue.
[[USS Will Rogers|USS ''Will Rogers'' (SSBN-659)]]
and MediaWiki provides this html:
<ahref="/wiki/USS_Will_Rogers_(SSBN-659)"title="USS Will Rogers (SSBN-659)">USS <i>Will Rogers</i> <spanclass="nowrap">(SSBN-659)</span></a>
(not at all clear to me why there is a combination of and <span class="nowrap"> but that is beyond the scope of this question)
It would appear that Enable page previews chokes on html markup inside the <a>...</a> tag.
I see the same behavior in Firefox current Win 10. It appears to be due to the nowrap. Testing the link without it or with only non-breaking spaces appears to function as expected. The reason a nowrap is there (but probably shouldn't be) is that parentheses sometimes separate from the text to which they are connected. It looks like it's the italics that cause the nowrap-nohover behavior to extend to the earlier words. And based on adding an empty span below, it appears to me that your conclusion that HTML markup causes the feature to choke is a reasonable one. It also chokes on italic markup only.
I suspect that the reason for the <span class="nowrap">(SSBN-659)</span> is not to keep the hull designator on the same line as the ship prefix and name but to prevent a line break at the hyphen in the hull designator. US hull designators have one and sometimes two hyphens.
Attempting to undo an edit, or range of edits, that the system was not able to resolve used to result in a message something like "unable to undo because of intervening edits". This seems to have recently changed to "The edit could not be undone because it does not exist or was deleted." This is very confusing when the edit does still plainly exist.
SpinningSpark10:47, 13 December 2022 (UTC)
Sure. Go to
User:Spinningspark/Sandbox. View
this diff which is a sequence of two edits. Press undo. You will get the same result if you try to undo only
the second edit in the sequence. If you look at
the diff of the final edit to the section (which is the one preventing the undo) it is clear that the first two edits are still there. That is, they do still exist and have not been deleted. This is especially confusing because there is (or used to be) a different message for attempted undos that had really already been undone.
SpinningSpark12:04, 13 December 2022 (UTC)
The usual way to align a table column is style="text-align: left/center/right;" in every cell in the column. That's a lot of repetition for long tables, and sometimes forgotten in some cells. See
List of submarines of World War II for an example with 3000+ style=. I'm looking for a better way (in general, not for a specific table). A table can be build with
Help:Table#Row template but that's often more trouble than it's worth. If a whole table is in a template then
Wikipedia:TemplateStyles can be used. For example,
Template:COVID-19 testing by country uses
Template:COVID-19 testing by country/styles.css. Creating a template and associated style page and moving article content to the template even if it's only used in one article is usually not a good solution.
For testing I made
User:PrimeHunter/cell align/styles.css with a bunch of named classes to enable column alignment without creating new pages. The below example has <templatestyles src="User:PrimeHunter/cell align/styles.css"/> and starts the table with class="wikitable right2 center4 right10" to right-align column 2, center-align column 4, right-align column 10.
Caption text
Columm 1
Columm 2
Columm 3
Columm 4
Columm 5
Columm 6
Columm 7
Columm 8
Columm 9
Columm 10
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
The last row starts with colspan="2" to show that it moves the alignment one cell to the right for the remaining cells in the row. That's nearly always bad. I have limited knowledge of CSS, templatestyles and consequences of using them.
Would it be OK to create such a templatestyles page in template space and promote wide use or are there bad consequences?
Can it do something smarter than defining a class for every combination of column number and alignment?
Can it do something to handle colspan? The table can override the styling with style=... in each relevant cell but that's cumbersome.
Would it be worth considering to place such classes in
MediaWiki:Common.css so tables don't need <templatestyles src="..." /> to use the feature?
A similar approach is already used for automatic numbering of table rows using {{Static row numbers}}, which just adds templatestyles with {{Static row numbers/styles.css}}. I don't think there have been any problems. As long as the class names are especially distinct it seems a good way of customising tables beyond the Wikimedia defaults. — Jts1882 |
talk15:50, 13 December 2022 (UTC)
Fine (Although those names are very generic, you should prefix them with a template id, to avoid colliding with other templates)
Not really
Nope... CSS has no idea about colsspan, it only counts elements. This is exactly why we don't widely use classes like this yet, as you'll inevitably run into this problem and then ppl will complain that the template isn't working etc etc.
No, we are trying to get rid of as much of the contents of Common.css as possible
Not really. I've occasionally asked browser developers to create logical col/row pseudo selectors for tables that deal with colspan/rowspan, but chances of that being a thing any time soon are pretty low. There is a column combinator selector being considered, but it seems to have very little traction among browser developers.
I have been thinking about this question. I would not support generic namings like this. I might support giving the classes real names like "date" and such and adding it to each cell of interest rather than doing it at the table level. Class names like "date" and etc. CSS might look like td.date { text-align: right } for what I mean.
Item 4 is an absolute no go.
I think item 3 makes the suggested way of doing things bad as well.
I don't think we fundamentally need anything that would align things to the left anyway. Once you have a system like this you're just modifying what's centered and what's right.
Item 1 hits the normal caveat listed in
mw:Extension:TemplateStyles regarding whether styles should be able to "break out" of their containing template, but we 1) have issues with that anyway with templates like the aforementioned and basic templates like {{div col}} which are much more used, and 2) the stated reason for that original warning was for functions in VisualEditor that haven't come to pass and likely won't and which will need to be worked around anyway. I really need to get around to starting a discussion regarding that line, as it's holding up a bot request regarding plainrowheaders.
Izno (
talk)
19:02, 13 December 2022 (UTC)
One of the reasons I haven't really jumped on dealing with the table case is because it kind of opens the flood gates to arbitrary styling of just about anything on the page. I'm not really sure how good one should feel about that.
Izno (
talk)
19:04, 13 December 2022 (UTC)
"More" and "TW" menu on Vector 2022 skin intentionally do not open on hover
See
T277322 and
T325035 for details. Basically, in Vector 2022, the "More" and "TW" (Twinkle) menus do not pop open on hover, as they do in Vector. I have grown accustomed to this behavior, and I was disappointed to see that it didn't work in Vector 2022. When this was noticed and submitted as a bug, the bug report was declined as "works as intended", i.e. the menus should never have been popping up on hover in the first place.
Nevertheless, I like this behavior. I use these menus on at least half of the pages I edit these days, so saving clicks is helpful to me. Does anyone know of a Javascript hack that I could use to replicate this behavior in Vector 2022? –
Jonesey95 (
talk)
17:33, 13 December 2022 (UTC)
This is a blank page, which is quite strange to me, since I went to this page yesterday (and there's not even a deletion log there: it just says there's no such page and never was. The page title is this: UserːCyberbot I/Current AfD's.
So I type it in manually, and go to
User:Cyberbot I/Current AfD's. This page definitely exists. What is going on? I copy the old URL and look through the codepoints of the characters... and I notice something. The colon is different! Compare.
@
Matma Rex:@
PrimeHunter: Thanks. This isn't really an issue that prevents me from doing anything, but it is very strange (I only ever accessed those oldids by clicking from the revision history). I suppose this will just be one for the ages; hopefully, if someone else has this issue a few years down the line, they can read this and become educated ;^) jp×g23:38, 13 December 2022 (UTC)
It's not too simple when you're using it to e.g. switch to the input method for Cyrillic or Arabic script while typing on a standard English-language keyboard. It's just that the only input method available for English is the IPA one with the funny colon. Besides, it's been this way for like 6 or 10 years.
Matma Rextalk05:05, 14 December 2022 (UTC)
Template and Wikipedia namespaces not showing up in Android contributions
As the title suggests, when I open the Wikipedia app on Android those two namespaces are not visible when I tap on my own contributions. This is odd, seeing as a couple months ago I could. I am able to provide a screenshot if necessary. Using version 2.7.50426-r-2022-12-08. No custom CSS or anything, and I have "All namespaces" ticked on the namespace filter too. Is it supposed to be like this? Thanks.
X750. Spin a yarn?Articles I've screwed over?06:55, 14 December 2022 (UTC)
I'm seeing this in one of my log files (for ChristieBot): fatal: detected dubious ownership in repository at '/mnt/nfs/labstore-secondary-tools-project/pywikibot/public_html/core_stable'/To add an exception for this directory, call:/git config --global --add safe.directory /mnt/nfs/labstore-secondary-tools-project/pywikibot/public_html/core_stable. Everything seems to be running fine, but I'm posting here in case someone knows what, if anything, should be done about this.
Mike Christie (
talk -
contribs -
library)
12:48, 14 December 2022 (UTC)
@
Qwerfjkl not that I know of, buuuuuuuuut - since you have over 500 lines of customization between
User:Qwerfjkl/common.js and
meta:User:Qwerfjkl/global.js -- who knows what could be in collision. Same advice we give everyone: turn off ALL of your scripts and troubleshoot by turning them on in groups or one by one until you find a possible collision. Also note that with the specific ones you are mentioning above seem to belonging to a user that appears to be globally inactivate for a couple years and there's likely a lot to unravel here. —
xaosfluxTalk02:05, 10 December 2022 (UTC)
I was going to suggest a specific ticket on Phab, but yeah, if you've got 500 lines of customization, you're on your own IMO.
Izno (
talk)
02:16, 10 December 2022 (UTC)
@
Izno, it's more like 200, as I explicitly prevent my global.js running on enwiki, and I have commented out around half the lines of my common.js. I'll try loading the scripts individually and check for conflicts, I was just curious if there was some change in mw js, like the recent one with multiple importScript calls. —
Qwerfjkltalk10:47, 10 December 2022 (UTC)
Never mind. No sooner did I post this than I suddenly found the documentation subpage that had been eluding me, found the category there and fixed it myself. Nothing to see here, carry on.
Bearcat (
talk)
17:14, 17 December 2022 (UTC)
Urgently need a list of recently-introduced bracket-within-bracket changing errors
I noticed this morning that AWB is incorrectly trying to "fix" brackets within brackets as genfixes. For example in
Symbolist painting, AWB tried to "fix":
[[File:Somni_Joan_Brull.jpg|thumb|''Ensueño'' (1897) by [[Joan Brull]], National Museum of Art of Catalonia, Barcelona]]
to:
[[File:Somni_Joan_Brull.jpg|thumb|''Ensueño'' (1897) by]]Joan Brull]], National Museum of Art of Catalonia, Barcelona]]
I have made thousands of AWB edits in the past few weeks to fix new disambiguation links and the like, and I am sure others have used the tool as well.
I have
filed a bug report at AWB to correct this for the future, but I need a quick list of instances where a correct bracket-within-bracket arrangement has been changed to an incorrect one, so I can go back and fix any of these that can be found.
BD2412T15:50, 17 December 2022 (UTC)
So far as I can tell, only 18 were edited by me, and none of those edits introduced the error, which I would be very self-conscious about. Even the ones that have been AWB-edited appear not to have had errors introduced by AWB (which it turns out was only attempting to make a fix to balance a bracket error already occurring earlier on the page).
BD2412T19:24, 17 December 2022 (UTC)