This page contains discussions that have been archived from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.
I was recently told that my bot's minor edits should be marked as "mb" instead of just the "m". I only see the option to set the edit as minor and not as "minor bot edit". Am I missing something, or is this an outdated concept? Thanks!
§hep •
¡Talk to me!17:41, 31 May 2008 (UTC)
I don't know why it does this, but IE renders the small infoboxes a huge tables or very small ones and ignores the minimum width. The text and images are not usually much larger than the minimum width. You can see all kinds of differences
here. I don't know why the last table renders somewhat acceptably
last table, but it expands once you remove the ≤ from the last row. Other tables with that character act differently, however. I don't know how to fix this.
Potapych (
talk)
19:41, 31 May 2008 (UTC)
I asked
here and
here about the static HTML dumps being broken since April 2007, but I'm not getting any response. I'd post to the wikitech mailing list, but I can see other people asking the same question there, again with no response. Is there somewhere else I should bring this up? Thanks,
Bovlb (
talk)
23:26, 30 May 2008 (UTC)
Everyone who needs to know about it knows about it. In practice it's probably not going to get fixed until someone with access to run the dumps does some extensive troubleshooting, unfortunately. —
Simetrical (
talk •
contribs)
16:32, 1 June 2008 (UTC)
Thanks for the response. I'd be happy to assist in the troubleshooting, if I can be of any use. If it's just a question of scale (a reasonable guess, given that we're getting stuck on en), then maybe it would help to restrict the dump to namespace 0.
Bovlb (
talk)
17:48, 1 June 2008 (UTC)
Subsection headings heavier font than parent headings?
In my browser (Firefox 2.0.0.9, on an Asus Eee PC running Ubuntu), subsection headings have a heavier font than section headings they belong to, and it looks quite strange. In other words, subheadings look more prominent than their parent headings. What I'm referring to is constructions such as the following where my browser shows "Subsection" as heavier (although smaller) than "Section":
Not that I'm aware of. In my memory it has been like this for at least a year or so. Note that the larger heading (level 2) has a pagewide underline though. --
TheDJ (
talk •
contribs)
18:45, 31 May 2008 (UTC)
Thanks TheDJ. You mean you too find that "Subsection" has a heavier font than "Section"? If so, I suppose I've just not noticed it until now..--
87.252.38.216 (a,k,a, 87.252.35.199) (
talk)
19:01, 31 May 2008 (UTC)
h1 and h2 are for some obscure reason not bold in Monobook; h3 to h6 are. You can fix this with the following additions to
Special:MyPage/monobook.css (taken from my own monobook.css):
/* Non-bold headers look ugly on Ubuntu Firefox */h1,h2{font-weight:bold;}.editsection{font-weight:normal;}
Hi all. I keep finding that when I use IE7 on Windows Vista, clicking on "My Watchlist" causes the browser to crash. Has anyone else noticed such a thing? --
RFBailey (
talk)
22:35, 31 May 2008 (UTC) PS. Will try to stick to Safari in future....
I was making a change to the infobox on
Hong Kong, and the form is loading all 74KB worth of text everytime I want to preview it, which causes slowness and problems to scroll accurately to the end of the lead on the edit box. I have similar problems editing the lead of long articles.
Can we have a little button that lets me edit only the stuff that is before the first section header? I see that it's already implemented by using "§ion=0" on the URL, but I can't find any nicely placed button to access it easily with one click --
Enric Naval (
talk)
13:40, 1 June 2008 (UTC)
This feature request is simply to allow users to view article histories by time rather than number when they choose. So one could view a list of all the revisions to an article in the last day, week, month, ect. on one page.
How does one undismiss the notices at the top of the watchlist? I tried to dismiss the notice about the bot article-creation proposal, and it dismissed the notice about the elections too.
DuncanHill (
talk)
18:46, 2 June 2008 (UTC)
Try pasting the following JavaScript into your address bar:
That should undo the effects of your dismiss click. (You may need to use higher numbers than 17 if the watchlist was updated since I wrote this.) --
ais523 18:57, 2 June 2008 (
UTC)
Yeah, I don't think that both of them should have been clustered like that. The board stuff is in the sitenotice everywhere else; sticking it on the watchlist doesn't make much sense.
EVula//
talk //
☯ //19:06, 2 June 2008 (UTC)
The references on the
photovoltaics page are hidden. When you click on a ref it doesn't take you anywhere. I can see the benefit of hiding some things on a page (navigation templates etc.) but the dead refs are annoying. I don't know if these suggestions are workable but here they are: 1) The link should take you to the ref section even if it's closed. This gets you in the neighborhood at least. 2) Clicking on a ref should automatically open the hidden ref section and take you to it. That's it. Ciao for niao...
Mrshaba (
talk)
03:57, 3 June 2008 (UTC)
The anti-spam check measure added to the html source code on May 22 has resulted in a problem for BlackBerry users attempting to edit. Now, the edit screen (when using the BlackBerry's browser), produces just the anti-spam editform box and legend "Do not fill this in!". The rest of the screen is garbled/unusable.
Can the developers take a look at this so that us crackberry/wiki addicts aren't deprived of the ability to edit/reply to Talk pages using our BlackBerrys? JGHowestalk - 19:36, 1 June 2008 (UTC)
This issue is not affecting all blackberries, I just made an edit from one without an issue, as long as I didn't put anything in the "dont fill in this box" section. —
xaosfluxTalk22:01, 1 June 2008 (UTC)
What BlackBerry version are you using? Mine is the fairly common 8700 with the factory-preloaded 4.2 Browser. JGHowestalk - 13:08, 3 June 2008 (UTC)
How do I get a list of a particular user's contributions sorted by the number of characters they added in each contribution? Thanks! --
Pascal666 (
talk)
17:35, 2 June 2008 (UTC)
Strange behaviour with "what redirects links here"
I was checking the list of redirects that link to an article, using the filters that had been added recently to the mediawiki interface for Wikipedia, and I came across some strange behaviour. It seems that if you use a template on a redirect (for instance, the "unprintworthy redirect" template), and the template has a link in it, then if you use the filters to generate a list of "what redirects link here" for the article linked in the template, then all the redirects with the template on them show up in the list as well. Is this a feature or a bug? This also happens if a page contains wikilinks elsewhere on the page as well as the redirect markup. For examples, see
User:O'Malley II,
Talk:The Borribles: Across the Dark Metropolis and
User:Viper1928374/Sandbox. All three of these redirects were redirected without blanking or archiving properly first (edit them and you will see what I mean). Because they have "The Lord of the Rings" linked on the page somewhere, they are showing up in the list of
redirects pointing at 'The Lord of the Rings'. Again, is that a feature or a bug? (And a side note - can I 'fix' those pages, or is that not acceptable?).
Carcharoth (
talk)
21:48, 2 June 2008 (UTC)
I get a strange error on my main account
user:stefan since now, when clicking on a diff link, history or edit button I get a page called internal error with the following stack trace
Detected bug in an extension! Hook CentralAuthHooks::onGetUserPermissionsErrorsExpensive failed to return a value; should return true to continue hook processing or false to abort.
It is not a unified account and another unified account exists
and I did
this edit yesterday day, now impossible for me to revert it, looking at the logs it looks to have something to do with user so is probably SUL related but you never know. If any sysop sees this please revert the edit, not sure how to prove I'm that user since I can not do any edits, but most edits on this users userpage is done by the user stefan so it should be a good indication :-) email is confirmed for the user but I might not answer so soon.
StefanBot (
talk)
01:38, 3 June 2008 (UTC)
I'm glad I'm not the only one. I am also having the exact same issue. I cannot edit or view histories without getting that error. I have never edited my monobook file and pretty much all my account settings are left as normal. This just started yesterday. My user name is
User:APL. Any help would be appreciated.
72.10.110.107 (
talk)
13:10, 3 June 2008 (UTC)
Wouldn't it be great if we could have a page where we could see ALL the changes made to the document, including redirects, moves, and creation, from all the user who contributed? There could be a separate sub page, or function, where you have the Revision history page on one side, but only taking half (or whatever porportion) of this proposed page, and then lines to from the users who contributed (in bluelinks) to their changes! Please submit this idea to Bugzilla, since I don't have an account. Thank you!
68.148.164.166 (
talk)
05:03, 3 June 2008 (UTC)
Wouldn't it be great if we could search within our own contributions (or whatever, (or changes)) for say, all things we replaced with "{{main|". Please post this on bugzilla since I don't have an account, thanks!
68.148.164.166 (
talk)
19:17, 3 June 2008 (UTC)
I don't know what is going on, but I am a Wikipedia administrator trying to log into the site. I can log into my Wikimedia Commons account, but I can't log into my Wikipedia admin account for some reason. This is why I'm using this IP to post this message.
Every time I click the Log in/create account tab, I get a pop-up that says Internet Explorer can't open the internet site, and the operation is aborted. Is this a problem with Wikipedia? Because I'm having no trouble with Commons...
86.154.115.237 (
talk)
19:36, 3 June 2008 (UTC)
Yes that is definitely true. But I was able to get in through Commons. I have not tried the other Wikimedia projects to see if I could get in through them however. I have only tried Commons, where I had no problem, and English Wikipedia, where the login page generates some sort of error. --
Filll (
talk |
wpc)
19:40, 3 June 2008 (UTC)
English Wikipedia login problem with internet explorer
I just tried to log in on English Wikipedia from internet explorer and the login page fails to load properly, so that login fails. I was able to get into Commons and came here that way with the unified login however. This just started in the last hour or so.--
Filll (
talk |
wpc)
19:37, 3 June 2008 (UTC)
It's not fine here, the login page on the English Wikipedia is not working for me. I'm on product version 7.00.6000.16386. I suggest, Fill bringing it up at
WP:AN to try and direct more people to this discussion.(
D.M.N.)
86.20.54.64 (
talk)
19:47, 3 June 2008 (UTC)
"If there are any Javascripts running inside the body tag or inside a table which is directly a child of body, then most of the times this error is bound to happen. The solution is to move the script to the top or bottom of the body tag or even moving it after the body. The script can also be put inside a function and then calling it from window.onload. Another solution to this problem is to add defer=”defer” in the script tag."
[4]Prodegotalk19:58, 3 June 2008 (UTC)
For info, I've been approached for help by an inexperienced user who has been having the same problem. I would not know how to advise him to log on by another route. Helpdesk told me the problem had been reported here. Discussion is on my talkpage.
Itsmejudith (
talk)
20:13, 3 June 2008 (UTC)
The problem was an unclosed <span> in
MediaWiki:Loginend. This bad HTML triggered fatal JavaScript errors on the new anonnotice code while silently failing on the old code. With the fixed HTML, the new code works fine -- and should avoid triggering the "blank page except for fundraising link" error which we get lots of complaints about. --
brion (
talk)
20:35, 3 June 2008 (UTC)
So I've noticed that certain templates (specifically, stub templates) or other items sometimes shift over the start of all section headers below by about an eighth of an inch to the right (on IE7, at least). The former comes a bit on
AFC, and the latter I just noticed caused by a div box in
here. Does anyone know why it does this, or how the templates/boxes can be modified to not do this? The only solution I ever found on AFC was to comment out the template.
Someguy1221 (
talk)
22:13, 3 June 2008 (UTC)
A user
edited my talk page. So far, so good. The problem is that it has a "Newer Edit" link. When you click it, you reach
this empty diff page, which also has a newer edit link. But clicking it takes back to the same empty diff page. Continuing to click the newer edit link takes back to the empty diff page.
It can be reproduced by going to
User talk:ReyBrujo, clicking the history tab, and making a diff between the two consecutive edits by Useight from May 16 and June 3. Note that it is as if it were doing a diff against the same version of the page.
Since I guess the bug is interesting, and that it will disappear as soon as someone edits my talk page, I am fully protecting it (
WP:IAR) through cascading by transcluding it at
User:ReyBrujo/Sandbox. If it is not deemed interesting, it is fixed or cannot be reproduced, feel free to unprotect the sandbox. I am doing just a couple of edits per day, so there is low risk of anyone actually wanting to write anything in my page. Cheers! --
ReyBrujo (
talk)
02:55, 4 June 2008 (UTC)
Oh, you are right. As I said, I am rather inactive and don't check my watchlist anymore, or would have noticed that. Unprotecting my talk page since it can be reproduced anywhere. --
ReyBrujo (
talk)
03:08, 4 June 2008 (UTC)
Yep, I'm seeing it everywhere also. That's the only reason I wandered in here in the first place. I don't think it's actually doing any harm, it's just sort of annoying. --
Bongwarrior (
talk)
03:11, 4 June 2008 (UTC)
I am welcoming everyone warmly very much! I am a Polish Wikipedian. I would like to ask you for help very much in convincing my countrymen from Polish wikipedia so that they send the correct official name of the state of Nepal, following the example of you. Repeatedly I tried to implement updates, but they all the time for her aren't accepting. They are undermining the credibility of your information. Please compare both versions and pay attention to the official onomastics of the country:
english version,
polish version. Could you help it?
If grammatical mistakes are appearing in the text please to forgive me, because I am not knowing the English well. ; -)
I'm afraid that editors of the english wikipedia have absolutely no influence over foreign-lanaguage wikipedias. You will have to discuss this issue with other editors of the polish wikipedia.
Happy‑
melon15:51, 4 June 2008 (UTC)
Subpage moves
Resolved
According to the latest
Signpost technology report, subpages can now be moved during a page move. I have been testing this: when you do a move, you not get a checkbox for " Move all subpages, if applicable". When I try this with a user subpage with a subpage, I get:
A database query syntax error has occurred. This may indicate a bug in the software. The last attempted database query was:
(SQL query hidden)
from within function "Article::insertOn". MySQL returned error "1062: Duplicate entry '2-Gadget850/movetest1' for key 2 (10.0.0.235)".
This worked for me on my test wiki. I don't know why that error would occur. A shell user needs to look into why it's failing on Wikimedia projects. —
Simetrical (
talk •
contribs)
16:35, 1 June 2008 (UTC)
I like the
Wikipedia:Bounty board, but I see WikiProjects as not having the same support it affords to articles. If, for example, someone wanted to set
http://www.chipin.com/ on a Wikiproject, there would be no way to mark that up. Does the Foundation have the technical capability to fund WikiProjects as well as particular tasks?
76.240.230.195 (
talk)
17:07, 1 June 2008 (UTC)
The Foundation isn't funding particular tasks; individual editors are. And it's not unheard of for several editors to jointly fund a particular bounty. It isn't clear what the question here is - if it's whether the software can or will be modified to allow a chip-in widget to be embedded on a Wikipedia page, the answer is no - Wikipedia isn't Facebook. -- John Broughton(♫♫)20:05, 5 June 2008 (UTC)
Is it possible for an expression to return a user's last edit date & time ?
User:SoxBot V was indefinitely blocked today, sending us back to the dark ages as far as status updates go. Is there any expression that will return the date and time of a users last edit? Could this possibly be implemented? (Having this would allow a workaround for status indication that did not require a bot).
xenocidic (
talk ¿
listen )
23:21, 2 June 2008 (UTC)
The status bots were helpful, and since they have been shutdown by the dev's having another method to query this information would be helpful, but seems to beg for the perennially denied request for user variables... —
xaosfluxTalk23:53, 2 June 2008 (UTC)
Writing a simple
MediaWiki extension to provide this information via parser functions shouldn't be too difficult. Making it update reliably without totally ruining page caching is a little bit trickier, but not much. —
Ilmari Karonen (
talk)
03:08, 3 June 2008 (UTC)
With the fate of
User:SoxBot V, I have been constantly looking for an alternative. After looking at mediawiki.org, I found an extenstion that automatically does it:
mw:Extension:OnlineStatus. While some users may not like it, i believe it could be a benefit to Wikipedia, especially at
WP:HAU. Before filling out a bugzilla request though, I would like community consensus. What do people think of this extension?
Soxred 9322:33, 3 June 2008 (UTC)
One would set their Online/Offline toggle in their preferences, so one could log on, set this to online, getting ready to log out, set this to offline. There might be some code to add to a monobook file to do this automatically.
209.244.31.53 (
talk)
18:54, 4 June 2008 (UTC)
ID duplication
This is a problem I've run into before: pages with more than one stub marker or navbox (read: not one of each, at least two of either) generate XHTML errors because IDs cannot be called multiple times. The fix is to change the "id=" statement to "class=", which is simple enough, however Wikipedia has countless templates. Is there a solution to this other than trudging through the templates and changing everything manually?··TVOtalk05:08, 4 June 2008 (UTC)
This should not happen. Do you have some articles that have this problem ? The fix is not to change id= to class= btw. That is a bad way to solve this problem. --
TheDJ (
talk •
contribs)
08:27, 4 June 2008 (UTC)
Ouch, actually this does happen exactly like he said, because all stub templates (conforming to current standards) will contain a table wrapped inside a <div class="boilerplate metadata" id="stub">. The other two options would be
Run a bot to change all stub templates to use id="{{subst:PAGENAME}}" i.e. the name of the template e.g. id="Latvia-bio-stub" or whatever.
Change the software to prevent duplicate id attributes by numbering them sequentially after the first. Note that this is already done for section headings with duplicate names to keep the TOC working properly:
Latter option is probably better as it will fix more scenarios than have been identified, and it won't make the job queue go over 9,000. —
CharlotteWebb16:00, 4 June 2008 (UTC)
Probably closer to a million actually - we have 850,000 stubs assessed through
WP:1.0 quite apart from those that haven't been bannered. So the second option is considerably more elegant than the first and, as you say Charlotte, this problem occurs with more than just the stub templates.
Happy‑
melon20:14, 4 June 2008 (UTC)
Agreed, this is bad. Suggest a bot fixes it all. Having duplicate ids is not something that mediawiki should handle in my opinion. --
TheDJ (
talk •
contribs)
22:12, 4 June 2008 (UTC)
See
T6515. One problem (which is why Tidy doesn't fix this automatically) is that renumbering id's will basically break them: any styles or links to that ID will just break. People will have to stop abusing id's for things like stub templates in any case. —
Simetrical (
talk •
contribs)
00:00, 5 June 2008 (UTC)
Simetrical, I only suggested renumbering id's after the first... wouldn't any link intended to point to them would already be broken? —
CharlotteWebb20:29, 5 June 2008 (UTC)
Yeah, you're right, it causes no problems with links (which probably go to the first anyway). It breaks styles, however (#stub may be styled), and possibly some JS. —
Simetrical (
talk •
contribs)
20:32, 5 June 2008 (UTC)
Software tor handling
Just a quick note to let everybody know that the TorBlock extension has now been enabled on Wikimedia sites. All Tor blocks are now overridden by a Wikimedia-wide system. It will forbid unregistered users from editing through Tor. I realise that a lot of problems are caused by pagemove vandals using sockpuppets through tor, and this is why, in the extension, I have subjected all tor users to much stricter autoconfirmed requirements. Users editing through tor will now require 100 edits and a 90 days to become autoconfirmed. I hope that this addresses the objections to my extension (I've spoken to a few checkusers, stewards, and other developers, and they seemed to think this was a reasonable compromise). If not, you know where to find me. — Werdnatalk08:01, 4 June 2008 (UTC)
I like the idea, I wish it would hardblock the tor nodes, but I know some places like Chinese Wiki wouldnt work if that was done WMF-wide. MBisanztalk09:10, 4 June 2008 (UTC)
I haven't actually looked at the extension, but it might be possible to have a hardblock configuration variable.
Mr.Z-man23:22, 4 June 2008 (UTC)
Problem maximising Wikipedia images in IE7
Resolved
Hello, recently I've been unable to maximise about 75% of Wikipedia images. I get the error message
Most of the answers that I've found mention something about fiddling with Java code, something I don't have any knowledge of. I've tried running IE with no add-ons and the problem still persists. This isn't happening on any other website, and I'm not sure why it's just started up today!
There are bugs in Internet Explorer which trigger this behavior sometimes, when some (but not all) JavaScript operations are performed in a certain order when there's some bad HTML code (or sometimes maybe for no particular reason). Some recent changes to English Wikipedia-specific custom JavaScript which fix a separate bug can sometimes trigger this one, but that is fixed by a slight change to the HTML layout.
So, it's possible for the moment to encounter the second error with the new JavaScript and the old cached HTML:
If you use any browser but Internet Explorer, you won't encounter the bug.
If you visit pages updated since yesterday, you won't encounter the bug.
If you log in, you'll get all newly rendered pages and won't encounter the bug.
If you wait for up to 30 days, all the old cached HTML will fall out of the system and you won't encounter the bug.
We might poke a bit to see if we can tweak the JavaScript in a way that doesn't trigger either bug, but no guarantees. --
brion (
talk)
16:22, 4 June 2008 (UTC)
Thanks for the advice. I just installed Opera anyway, which seems a lot better so far. No bugs or crashes as of yet! Cheers
78.143.205.7 (
talk)
17:51, 4 June 2008 (UTC)
The trigger here was
an improperly-closed <span/> in
MediaWiki:Shareduploadwiki-desc. While technically that's valid XHTML, Internet Explorer gets confused by it, and its HTML parser freaks out when the JavaScript gets run. It's now fixed, but isn't a problem with the moved JS trigger anyway. Old cached pages with the bugs will still apply unless we can find a workaround at the JS. --
brion (
talk)
19:37, 4 June 2008 (UTC)
Ok, I found a tweak to the JavaScript code which doesn't trigger the "operation aborted" error even when the bad HTML is present. You _might_ have to clear your local cache if you've already got the previous version, but new visits shouldn't have any problems. --
brion (
talk)
20:22, 4 June 2008 (UTC)
underground nuclear explosions
In an attempt to convert from amateur to professional writer, I am in need of an important answer. A hypothetical cache of nuclear weapons, set off simultaneously a mile or so underground, what would the effects be? Massive earthquakes? Eruptions? Nuclear fallout of any kind? Keeping in mind the necessary pipelines, elevator shafts, and all the whatnot... would the surface get much of anything? Thanks in advance. —Preceding
unsigned comment added by
12.201.115.120 (
talk)
21:42, 4 June 2008 (UTC)
Ah, but it could be. What if the underground nuclear bomb is set off one mile away from the location of the WikiMedia servers? How much would the normal operation of Wikipedia be affected? ;-)
Waltham,
The Duke of01:38, 6 June 2008 (UTC)
Delete confirmation
This one might only be understood by admins who delete pages or those more familiar with MediaWiki than I am =D. When I delete a page, there's a page for confirmation in which I select the reason (A7, G4, G8 or whatever). Until about two weeks ago the delete reason was fed automatically from the CSD template. The content also didn't show up in the "other reason" box, which it now does. I don't know if this was changed by consensus, but is there a script or something that will stop it doing that? It makes CSD (currently with a +100 backlog) patrol more tedious because I have to delete the content first. Thanks,
PeterSymonds(talk)15:00, 5 June 2008 (UTC)
Can anyone figure out why
Orsilochus shows up incorrectly in
Category:Characters in the Odyssey? It appears first on the category page, as if it was defaultsorted as an asterisk, even though it's not, and I can't figure out why, and it's driving me absolutely nuts. The article appears sorted correctly in all other categories.
Ford MF (
talk)
22:39, 5 June 2008 (UTC)
It is because {{Characters in the Odyssey}} has the * sort key in it. I just manually added that category to the page, so know it sorts like the rest. I am guessing the sort key should be removed from the template though. -
AWeenieMan (
talk)
22:46, 5 June 2008 (UTC)
Ahh. Now I see what was going on. Someone wanted the template to be in the category at the top, but their way of doing it added a sort key to all the articles. Anyhow, I fixed the template, so going forward there shouldn't be problems. -
AWeenieMan (
talk)
22:50, 5 June 2008 (UTC)
I would be great if we could click on the version instead of having to have to click on the 2 radio buttons just to get to the later version. Please post this on bugzilla, because I don't have an account, thanks!
68.148.164.166 (
talk)
02:22, 6 June 2008 (UTC)
If you click the (last) link next to the revision you want to diff, you will be able to see the next change.
Nakon02:44, 6 June 2008 (UTC)
It looks like the
m:WikiMiniAtlas has been disabled or is otherwise broken in the English Wikipedia. It no longer appears on any coordinate pages at all. There's no mention of its removal on the {{coord}} template talk page, the
m:WikiMiniAtlas page, or the
wiki project. Anyone know what's up? There's also a discussion on this at the
help desk --
ShinmaWa(
talk)22:20, 30 May 2008 (UTC)
I really don't think either of those changes should be causing the problem: mine is to a completely unrelated part of the code, while Brion's shouldn't have any effect unless you're using the secure server. (And no, the globe icon doesn't appear on the secure server, either.) Besides, I think we can usually assume Brion knows what he's doing. :-) The problem might be in
m:MediaWiki:Wikiminiatlas.js too, although that page doesn't seem to have changed since April. —
Ilmari Karonen (
talk)
23:31, 30 May 2008 (UTC)
I would guess that it may be related to the domain change of the toolserver, but I really have no idea. ---
RockMFR00:17, 31 May 2008 (UTC)
Gee, one week of vacation and suddenly something exciting is happening. Thanks for adding the new toolserver.org URL. That should be it. --
Dschwen16:46, 2 June 2008 (UTC)
Is this related to the fact that suddenly all coordinate URL's are now rendering as gibberish in my browser (Firefox)?
Nibios (
talk)
15:13, 6 June 2008 (UTC)
What happened to the fraction "buttons" (1/2, 1/3, 2/3, etc.) below the article text edit window? They were very handy to insert the symbols into an article. Should I take their removal as a Wikipedia
WP:MOS "rule" they should not be used in articles? —
X96lee15 (
talk)
16:58, 2 June 2008 (UTC)
Looking at the history for
MediaWiki:Edittools, "Please don't use Unicode superscripts, subscripts, and fractions in article text - it creates accessibility problems. Use the <sup> and <sub> tags instead, or TeX for formulas." I don't see anything in
Wikipedia:Accessibility on this, so I'm not sure what the problem is. --——
Gadget850 (Ed)talk - 17:12, 2 June 2008 (UTC)
And by doing this the editor has made it impossible to enter for example fractions of an inch, and suggests installing software (i.e. :LaTeX) to allow you to continue to edit. So much for the "Encylopedia Anybody Can Edit" - well anybody if they are willing and able to install software items on their computer and edit off line. The arrogance of some users is amazing.
Nigel Ish (
talk)
17:43, 2 June 2008 (UTC)
With superscript and subscript one can type 1<sup>1</sup>/<sub>2</sub>" and get 11/2", but it is far from ideal - too many clicks.
DuncanHill (
talk)
17:54, 2 June 2008 (UTC)
That is a very good question. We have a lot of good templates, but the organization stinks. If I have a job that needs a template I've never used before, I use
Special:PrefixIndex and look in template space for a logical name— this really works a lot of times. Other times, I use a template because I saw it used somewhere else. --——
Gadget850 (Ed)talk - 20:44, 2 June 2008 (UTC)
Also, the template documentation page itself for {{
frac}} shows problems in my brower (IE7), with truncated text. I stopped using frac for simple fractions (like ½) long ago for this reason. —
Andrwsc (
talk·contribs)
20:41, 2 June 2008 (UTC)
Random832 asks, "What's wrong with 1+1⁄2?″ The apparent grounds for rremoving the fraction "buttons" is "Please don't use Unicode superscripts, subscripts, and fractions in article text - it creates accessibility problems". 1½ (Unicode) is a lot more accessible than 1+1⁄2 (using the frac template) because:
Decent screen reader programs can handle Unicode, which has been the dominant character encoding since about a year after the launch of
Windows XP.
+1⁄2 (using the frac template) generates the following HTML (which I got by using my browser's "View source" facility on this page):
which produces an adequate visual representation of the fraction ½ but which no screen reader software is ever likely to interpret as a fraction.
Philcha (
talk)
21:17, 2 June 2008 (UTC)
I wouldn't be so sure about that, though of course real evidence can only be obtained by testing various screen readers, which I haven't done. Still, stripping away the HTML markup, that boils down to "+1⁄2", where the "⁄" is a Unicode
fraction slash. I seems at least reasonable that a screen reader should read that as "one half", or at least "one slash two". Certainly it's less ambiguous that just plain "1/2" using an ordinary slash. (Oh, and on the other hand, "¼", "½" and "¾" are long-established
ISO Latin 1 characters, so it seems at least misleading, even if technically correct, to label them "Unicode fractions".) —
Ilmari Karonen (
talk)
23:39, 2 June 2008 (UTC)
The only screen reader I am slightly familiar with is
JAWS, and it seems quite popular. Checking their support page,
[8] it supports all of the Unicode characters, including fractions. Not an exhaustive study. --——
Gadget850 (Ed)talk - 23:55, 2 June 2008 (UTC)
½, ¼, and ¾ work if you don't want to scroll down for those three. --
NE2 00:08, 3 June 2008 (UTC) (Ampersands escaped to allow the entities to display as intended. —
Ilmari Karonen (
talk)
01:44, 3 June 2008 (UTC))
Sorry to be the cause of so much trouble...the fractions are fine really, and I shouldn't have removed them. My problem is with the Unicode superscript characters.
HTML superscript: E=mc2
Unicode superscript: E=mc²
As you can see the HTML superscript is much easier to read. Also, if you're trying to search for a number in a superscript, you can't if the Unicode superscript character was used unless you copy-and-paste it into the search box. Searching is easier if the <sup> tag is used instead because you can search for "2" instead of "²". —
Remember the dot(
talk)00:51, 3 June 2008 (UTC)
Yes it's true that
JAWS, the screen reader I use, supports Unicode characters, and
has done so for a while now. The problem is that the pronunciation for almost all unicode characters outside Latin 1 and the users
Windows code page is undefined because of memory limitations. The only Unicode characters that have defined pronunciations by default are some letters of the Arabic alphabet and arrow symbols. I use the default settings and only ½, ¼, and ¾ speak properly for me with JAWS 8 - not the most recent version but IMO a representative sample. Graham8709:52, 3 June 2008 (UTC)
Superscript 1, 2, and 3 are in ISO Latin-1 (8859-1) and available as ¹ ² ³. In Firefox I find expressions like E=mc² or "g = 9.8 m/s²" more legible when they are used; HTML superscripts come out too large and cause the line spacing to be irregular. And all the more so when I use
lynx, which I sometimes do. --
207.176.159.90 (
talk)
22:26, 6 June 2008 (UTC)
T16356, WONTFIX. It's not a bug, it's completely intended. The feature would be nonfunctional if the throttle applied per page rather than per action. Restricting the action to sysop only would be the appropriate thing to do here if it's a problem. —
Simetrical (
talk •
contribs)
00:10, 5 June 2008 (UTC)
Er, did anyone think about writing it such that the throttle is updated after the page moves are done? My thoughts were that if you've got at least one pagemove left in your allowance you're allowed to move the page, but then it adds however many pages to your count so the next time you try and move something it finds you're way over. Basically, you'll always be allowed to move all the subpages, but in many cases that will put you over the throttle limit so you can't move anything else. Is there a reason that wasnt' considered in the bug thread?
Happy‑
melon10:17, 5 June 2008 (UTC)
Why should that be treated differently, though? I explained my thoughts on the purpose of throttles in the commit message for
r35897: "Rate limits should be applied per user action, not based on how large the effect of each action is. Note that moving a page with its talk page only counts as one move for rate limits; this is the same principle. The only point of rate limits (as far as I can think of) is to prevent unauthorized automated scripts from creating a mess in 30 seconds that it will take 10 hours to clean up by hand. Since a move with subpages is no harder to clean up (i.e., revert) than a move without subpages, they should count the same for rate limits." —
Simetrical (
talk •
contribs)
20:35, 5 June 2008 (UTC)
Well, is there a button to delete the hundreds of redirects left behind by page-move vandalism in a single click, too? —
Cryptic21:31, 5 June 2008 (UTC)
Admittedly, no, not yet, as I mentioned in that revision comment. It might be a good idea to add an additional permission for this, and restrict it to only sysops on enwiki for the time being. —
Simetrical (
talk •
contribs)
13:41, 6 June 2008 (UTC)
Nie tłomačtie metek dla kratek do vypełniania na ięzyk narodovy
Problem description
Preferring a language different from English causes the labels of text input controls to be translated, which in turn causes the user thus solicited to fill the fields in the preferred language. As a result, the information entered there is unavailable to users whose preferred language is different, which disrupts the collaborative character of Wikipedia.
Recommendation
Please make these labels an exemption from localisation. See my talk page for an extensive discussion of the subject.
That would make it very difficult to log in and work with a site whose primary language is not one you speak, so that's not a good idea. --
brion (
talk)
00:10, 6 June 2008 (UTC)
Yecril's position is that, because the input fields for edit comments and section headers (see above) is in the user's localized language, he is obligated to enter that information in his language. Hence, he has insisted on going around making edits with comments in Polish, and adding Talk sections (like this) with headers in Polish, despite the fact that he is obviously capable of writing in English.
To me, this obstinate behavior defies common sense (and common courtesy) on an English-language wiki.
However, as far as the UI is concerned, it does seem like a reasonable (if minor) improvement to label input fields with the site's language if a response in the site's language is expected—if the user knows enough English to enter an English edit comment, they should know enough English to understand "Edit summary". (How is the user supposed to "work" on the site otherwise?) Other UI elements that are read-only should of course be localized.
All in all, does this request apply to Wikipedia or to Wikimedia or to MediaWiki? I am not good at discerning the layers involved. I would like to file a feature request and I am not sure where to do it. --
Yecril (
talk)
11:19, 6 June 2008 (UTC)
Is anyone else finding the tooltips with every wikilink excessive and annoying, or am I the only conservative fool? Whereever I'm placing the cursor a tooltip is blocking the text I'm trying to read. Is there a way to shut them off? Arman(
Talk)07:23, 6 June 2008 (UTC)
The article on MARK STEYN display improperly but I cannot understand why that is. The "AWARD" section should follow that about the Canadian Human Rights. It seems correct in the edit window but, close the edit, and the article page shows the two sections running together. --
Interactbiz (Norm, Vancouver Canada) (
talk)
20:05, 6 June 2008 (UTC)
GeoHack isn't working. That is, if you click the latitude and longitude at the upper right corner of
Venice for instance, it shows a page of garbage instead of links to maps. The same problem occurs with other links to GeoHack such as
The Bronx and
Athens, and on both Flock and Explorer browsers.
Art LaPella (
talk)
22:03, 6 June 2008 (UTC)
That was bizarre. There was nothing wrong at the template, but effectively null editing it fixed the issue. Whatever. ---
RockMFR02:42, 7 June 2008 (UTC)
Wikipedia logo font
What is the name of the font used in the Wikipedia logo?
It's
Hoefler Text. (roman, smallcaps, italic, 13px, are all mentioned, at
meta:Logo and
this and
this and
this and
this, but I'm not sure which
style each element is). It's apparently available if you have an Apple computer (and possibly adobe programs?), or pay Hoefler & Frere-Jones US$299 for a single cpu licence. --
Quiddity (
talk)
03:47, 7 June 2008 (UTC)
Tool for finding diffs
Are there any tools for finding the diff URL for a certain piece of text? It's really a lot of work to search through the history manually if there are many edits.
(And on a tangent, I wish we could just overhaul the entire talk page system into something more like a web forum - with full wikisyntax and communal-editable spaces in each thread, of course - but with thread formatting, usernames, timestamps, meta-links, archiving, and so on taken care of automatically.) —
Omegatron (
talk)
02:54, 5 June 2008 (UTC)
That looks like a really cool tool. Bookmarked! Regarding improvements to the talk page, it will most likely not happen because the whole point of a wiki is to be as flexible as possible, so it's how it is right now.
GaryKing(
talk)07:07, 5 June 2008 (UTC)
In the works does not guarantee implementation, however. It hasn't been touched for four months. Also, I'm sure there will be some disagreements over whether it should be implemented or not.
GaryKing(
talk)17:09, 5 June 2008 (UTC)
Yeah, I've used WikiBlame. It's not that useful for this kind of thing. The user has made more than 1000 edits to the talk page, and it's about 10x as much work to find each individual diff and copy and paste it. I have the username and timestamp, but not the URL. —
Omegatron (
talk)
00:50, 6 June 2008 (UTC)
Yes, it's an external tool exclusively for Windows, unless you know of another operating system that can read .exe's. It is really useful and I recommend it. --
penubag (
talk)
01:45, 6 June 2008 (UTC)
I'm in Linux, though I might be able to run it with WINE. Is it kind of the same functionality as WikiBlame, though? —
Omegatron (
talk)
01:48, 6 June 2008 (UTC)
I haven't used wikiblame yet but what wpW5 does is searche for when a diff was first inputed and tells you who and when it happened. For example, I tell the tool to look on the article, Dog and then I input a sting of text from the article (or was) that I want the tool to find when it first appeared. It then tells me User:Example wrote that sentence on this date. --
penubag (
talk)
01:54, 6 June 2008 (UTC)
Yeah, that's what WikiBlame does, though the interface is kind of confusing. This tool goes through the edit history diff by diff to find it?
Does anyone know if there's a way to generate diffs directly from datestamps? It seems the diff URL is just a sequential number for every single edit on any page. —
Omegatron (
talk)
01:57, 6 June 2008 (UTC)
Ah, that's helpful. Could probably write a javascript for that.
I was originally digging through the talk page's history, but discovered that it was easier to go to the user's contributions, list 500, and then Ctrl-F search for the time of the edit in question. —
Omegatron (
talk)
17:16, 7 June 2008 (UTC)
Sections generated by templates, and the section "edit" link
You create sections by using the "=" character, thus "==== Section ====" gives
Section
I'm not sure how this would work in your template. You can add an edit-like template with {{}}. What are you trying to do? --——
Gadget850 (Ed)talk - 16:22, 6 June 2008 (UTC)
Well if a "section edit" link did appear, it would point to a section of the transcluded page (which in this case doesn't exist unless the parameter is filled in). There isn't any way to have a "section edit" link on a template properly link to a section of the page where the template is being used, unless you spoof the edit link and add an extra parameter for the section number (something like /index.php?title={{FULLPAGENAME}}&action=edit§ion={{{2}}}, the last part being necessary because there's no way for a template to know how many times it has been used on a page, or how many sections are above it) if that's what you meant. —
CharlotteWebb16:30, 6 June 2008 (UTC)
(edit conflict) I am trying to make a template that will create a new section if a certain variables are inputted, but that will create nothing if other variables are inputted. The name of the new section, if it is created, would be based on the inputted variable. I hope this makes sense. Thanks.--
Pharos (
talk) 16:37, 6 June 2008 (UTC)--
Pharos (
talk)
16:37, 6 June 2008 (UTC)
OK, I've given up on the first idea. Is there any way to create a new section if certain variables are inputted, but that create nothing if other variables are inputted (supposing that the code was on the page itself, not a template)? So far, this seems to be impossible too.--
Pharos (
talk)
02:28, 8 June 2008 (UTC)
i can not access to this
talk page. i think a fascist Spanish hacker has blocked my access. Please, could some checkuser check it and see what is happening there with my IP?
i don't know where to leave this message so if anyone know what i have to do a little help i would thank it very much. —Preceding
unsigned comment added by
Sclua (
talk •
contribs)
17:31, 7 June 2008 (UTC)
I mediated a case involving you and this article a while back, and when I was dealing with it there was no mention of topic-banning you or anything - obviously this is worse than a topic ban (and is not permitted by WP policy). I doubt it's a hacker (remember
WP:AGF) so it was probably a problem with your internet. We (Wikipedia) can't do anything so you could try again, or leave it......
Dendodge ..
TalkHelp17:39, 7 June 2008 (UTC)
Also suggesting that 'Spanish fascists' are responsible is
disruptive, and makes your question much less legitimate. Please try to state your problems clearly and objectively, rather then blindly accusing 'fascists' for your problem. Thanks, Prodegotalk22:52, 7 June 2008 (UTC)
Problem solved
i would like to thank the anonymous expert for solving my problem. If he leaves a messsage like "hello" on my page talk i will call him if i have this problem again, if not, i will try to come back here (i am saying "try" because i was unabled to access to these Village pump sections too, only post. Many thanks. (i will try to moderate my vocabulary). --
Sclua (
talk)
11:25, 8 June 2008 (UTC)
Has anyone else been having problems with their watchlist in the last day or two? Recently, sometimes when I've refreshed my watchlist, my browser locks up completely and crashes: this happened to me several times yesterday, and just once now. Is this a problem with me, or are other people having this? It only seems to be locking up when refreshing my watchlist. Thanks.
Acalamari16:24, 4 June 2008 (UTC)
Heh, well, I already do have Firefox, but it simply looks terrible on my monitor, and therefore, I don't use it. I'm actually confused by the watchlist crashing, as it's never happened before until yesterday: I'm trying to work out what's going on.
Acalamari17:27, 4 June 2008 (UTC)
No thanks: I'd rather find out what the problem is rather than just change browsers. After all, the problem may not be limited to me, and it's better to find out what's wrong.
Acalamari18:05, 4 June 2008 (UTC)
I was referring more to the numbers for maximum days and edits. My watchlist is over 9,000 but I have it set to only go back one day, by default, so it never crashes. Also I don't use IE. How much does it load before crashing, or can you tell? Also if you have any kind of javascripts or "gadgets" that affect the appearance of the watchlist (the "(unwatch)" button comes to mind) IE's ability to handle them might be suspect. —
CharlotteWebb17:59, 4 June 2008 (UTC)
It doesn't load anything: the entire screen just goes white, and I'm told that it's "not responding". In addition, I don't use any gadgets that affect editing or the watchlist, so they're not the problem either.
Acalamari18:05, 4 June 2008 (UTC)
I should note, however, that it doesn't crash every time I refresh my watchlist: only occasionally, but frequently enough that it's extremely annoying getting logged out and having to log back in again.
Acalamari18:09, 4 June 2008 (UTC)
No, I can log in perfectly. It's just that, when updating the watchlist sometimes, it doesn't open anything: the browser locks up, and most of the screen goes white, and the bar at the top of the screen says that it's "not responding". After that, I have to use "control, alt, delete" to exit the browser.
Acalamari18:19, 4 June 2008 (UTC)
I've actually had this happen in IE when attempting to refresh
WP:AN or
WP:ANI. My solution was to simply click on "Project page" at the top of the page, next to "discussion". In essence, it's what you'd click to get from the talk page to the project page; the fact that you're already at the project page is immaterial, as the browser treats it as a new request and a new page. This works for the watchlist as well, except that you click "special page" to reload.
UltraExactZZClaims~
Evidence17:20, 9 June 2008 (UTC)
ANNOYING PROBLEM WITH EDITING
Help me! Whenever i click edit on a page or Edit this Page, i get a download box asking for me to download 'index.php PHP file from wikipedia.org'. I have found that adding &internaledit=true to the URL of the edit page (e.g.
http://en.wikipedia.org/?title=TITLE&action=edit&internaledit=true ). But now i have to do that all the time so i can go to the edit page. This is very annoying. Can you please help promptly.....
Juggernaut0102(talk)09:47, 8 June 2008 (UTC)
This image page suffers from a strange problem. There
is no history but the page clearly exists. There is no associated file. I can't edit the page to tag it for deletion because I always get an edit conflict, apparently with nobody. The length of the page is (allegedly) 0 but when I view the page source, it is not empty. It probably isn't worth bugging the devs at this point, we might be able to find a resolution to this problem here.
MER-C09:01, 9 June 2008 (UTC)
I've noticed an interesting bug on the RfA Analysis counter. If looking at a current (as of 2008) Rfa, e.g.
Epbr123's 2nd, the end date shows 2008. However, if looking at
my own RfA (I'm not an admin) for lack of trying to find another example, shows the end date as end of 1999. Can someone explain this? This does not seem to happened to many others.
Simply south (
talk)
21:52, 9 June 2008 (UTC)
If you have seen this before, i moved this from WikiProject edit counters
Recursive page moves (
rev:33565) is a great tool for vandals
A recent wiki enhancement,
rev:33565, allows
autoconfirmed users to move not only a page, but every
subpage under it. Not surprisingly, this has become a fantastic tool for vandals like
Fuzzmetlacker (
talk·contribs·logs) and
AV-THE-3RD (
talk·contribs·logs) and
Matt the barber (
talk·contribs·logs). I imagine we could recursively revert the moves (although I haven't tried recursive
move-over-redirect) but there is no way to recursively delete the leftover redirects. Luckily, today's attacks stopped (
WP:ANI#User:Fuzzmetlacker) but zillions of other pages - mostly user talk pages - are perfect vandal targets and just a few recursive moves take a very long time to completely undo. A few ideas:
Turn off the feature entirely here.
Allow only administrators to use the feature.
Allow only bureaucrats to use the feature (it is most useful for renaming users and all of their subpages).
Implement recursive deletes.
Any other ideas? Opinions? I personally vote for 2 or even 3. I've never had a need for recursive moves. 4 - which isn't exclusive to the rest - has some appeal to me for vanishing users, etc. Hopefully we can expedite this to the developer folks before a full scale move war breaks out. (BTW, feel free to move this to a more appropriate venue like
WP:VPP). —
Wknight94 (
talk)
23:17, 6 June 2008 (UTC)
I would agree with restricting it. The code is already built into the user renaming extension, bureaucrats have had that ability for a while. But besides renaming users, there really aren't a whole lot of uses for it here as we don't use subpages too heavily (I'm sure its great on Wikibooks though). Brion is pretty adamant about not allowing the automatic suppression of redirect on page moves through the normal interface (admins and bots already have the right to do so, but it can only be done though the editing API, which isn't enabled here yet) so I doubt automatic deletion of the redirects, which is basically the same thing but less efficient, will be allowed. Due to the limited usage, I would support limiting it to admins though.
Mr.Z-man23:25, 6 June 2008 (UTC)
I agree with restricting it only to admins+. It does work over redirect (
when I did it for PhilKnight). Except today most admins deleted the main talk page first so there was no way to use the feature. When all the admins get used to the feature though all that will be left is deleting and/or protecting. Now only if d-batch and p-batch worked when looking at someone's contributions. --
ÐeadΣyeДrrow (
Talk -
Contribs)
01:41, 7 June 2008 (UTC)
It's ratelimited to the same number of actions per minute as normal moves. I.e., a vandal can only move, what is it? Six top-level pages plus however many hundreds of subpages they have per minute. This is an exceedingly poorly thought-out feature that's imposing hours of work for administrators to clean up one click by a vandal, whose legitimate uses are not remotely frequent enough to justify it. —
Cryptic00:52, 7 June 2008 (UTC)
Werdna told me that each page was counted, so an autoconfirmed can move say one top page, and five subpages per limit. →
AzaToth01:10, 7 June 2008 (UTC)
Werdna did make the change exactly as he said, but I reverted it, agreeing with Brion that it didn't make sense. It would have basically made the feature non-functional for non-sysops/bots/bureaucrats on all wikis; if that were desired, it would be best done through a separate permission. Apparently Werdna didn't notice that his commit had been reverted. —
Simetrical (
talk •
contribs)
19:00, 8 June 2008 (UTC)
Is that new? One of the vandals I mentioned above did 100 page moves from 19:25 to 19:26 today. I alone spent the better part of a half-hour reverting what it took the three of them about four minutes to do - and they were being merciful in stopping. —
Wknight94 (
talk)
01:22, 7 June 2008 (UTC)
Same here, precious time that I could have been using better. I mean I understand the good it can do, but I think the possibility of abuse and use by vandals far outweighs the positives. I say make available to admins+.
Tiptoetytalk02:11, 7 June 2008 (UTC)
How about an alternate possibility? Is it possible to simply prevent non-admins from moving user-space pages not belonging to them? That would get rid of a lot of these shock value moves. --
B (
talk)
04:11, 7 June 2008 (UTC)
I'm absolutely in favor of that, B. There is no reason for any user other than an admin to be moving someone else's userpage or its subpages.
NawlinWiki (
talk)
04:14, 7 June 2008 (UTC)
But what if someone finds someone else's abandoned sandbox page (via Google or whatnot) and wants to move it into article space? Rare yes, but could happen. Besides, then the vandals would go elsewhere and recursively move
WT:TV-NC, etc. The problematic issue is still recursive moves in general. —
Wknight94 (
talk)
04:26, 7 June 2008 (UTC)
I agree that recursive moves should also be limited. But I think the need to move someone else's user page is rare enough that it shouldn't be a problem to ask an admin to do it.
NawlinWiki (
talk)
04:51, 7 June 2008 (UTC)
I just spent the last hour or so move-protecting some of the bigger targets, so hopefully any vandal moves are limited to just a few hundred subpages at a time. *rolls eyes* ---
RockMFR05:18, 7 June 2008 (UTC)
I've written a program to find a list of pages outside the article namespace which have 10 or more subpages, and it reckons there's about 8,000. If this vandal knew what he was doing he could have caused serious damage by moving
User:UBX (3984 subpages and not protected until a few minutes ago). We need to either restrict the use of this tool or get an adminbot to move protect thousands of pages. Hut 8.510:43, 7 June 2008 (UTC)
Sorry, there was a bug in the program. The actual figure of pages with more than 10 subpages is nearly 12000. Hut 8.512:13, 7 June 2008 (UTC)
From looking at the recent autoconfirmed reform, it looks like the devs like polls, if only because it's easy to see how widely supported/attended they were. Is it time to adjourn to
WP:VPP with a strawpoll along the lines of "the recursive pagemove feature should be restricted to administrators. Support/oppose"??
Happy‑
melon13:27, 7 June 2008 (UTC)
I'm all for it. Unless you think the devs would be impressed with the damage that's already been done and the potential for more. —
Wknight94 (
talk)
17:25, 7 June 2008 (UTC)
The feature is limited to 100 subpages. Though, I'm not sure if that's just for non-admins or for everyone. --
MZMcBride (
talk)
18:23, 7 June 2008 (UTC)
The default limit of 100 pages is for 1) performance, 2) log spam. It applies to admins as much as anyone. In the future, this feature might be refined to do all the operations in a single query, and/or some method of dealing with the huge number of log entries (bot flagging?) might be used. In this case the default limit could be greatly raised. —
Simetrical (
talk •
contribs)
19:00, 8 June 2008 (UTC)
If anyone is still wondering recursive is limited to 100 page moves for admins. 1 for the mainpage 99 for subpages. Just take a look at my contribs. My apologies to the server and any patrollers. --
ÐeadΣyeДrrow (
Talk -
Contribs)
01:05, 8 June 2008 (UTC)
In
r36038 I added a new right that can be used to restrict moving subpages. It seems very sensible to disable subpage moving for the English Wikipedia at least, maybe for all Wikipedias. I wrote the feature for Wikibooks. —
Simetrical (
talk •
contribs)
19:00, 8 June 2008 (UTC)
(Incidentally, "recursive move" is a fairly silly name for it, since there's nothing recursive about the process. "Moving with subpages" is the name used in the software and makes considerably more sense. You wouldn't call mving a directory "recursive" because it happens to move all subdirectories as well, would you? Things like cp -r actually do recursively traverse the directory tree and so deserve the name. —
Simetrical (
talk •
contribs)
19:03, 8 June 2008 (UTC))
Ok, for the moment the subpage moves are restricted to sysops.
As a general issue, I'd consider a nice queuing system for relatively disruptive actions such as page moves, where users who aren't quite trusted to do everything immediately can initiate a delayed action, which can be reviewed and canceled by other users before it goes through. (Something like this has been kicking around as an idea for deletion for years, but we've just never gotten round to it.)
This should be a decent compromise between keeping basic maintenance functions open to all contributors, while reducing the damage that can be done with it by giving an opportunity to review and accept or cancel before, instead of after, the hard-to-clean-up action is finalized. --
brion (
talk)
21:06, 9 June 2008 (UTC)
Ah of course, the deletion queue - I remember seeing that on a todo/wishlist somewhere on svn or mediawiki. It does sound like a good idea... in the meantime, though, thanks for doing the shell change.
Happy‑
melon15:47, 10 June 2008 (UTC)
A related question: do these uncollapse by default during print view, and if not, can that be made possible? --
MASEM02:24, 10 June 2008 (UTC)
It doesn't look like it does. You could substitute the template, and the edit the class to one which does show—though that's seriously yucky. Maybe there's a friendlier template. I poked around briefly, but didn't see one. —
EncMstr (
talk)
02:29, 10 June 2008 (UTC)
It would be great if we could figure this out; there's some rather detailed infoboxes that could use collapsing in sections but should really autoprint full --
MASEM02:42, 10 June 2008 (UTC)
Additional query: since this is meant to be for hiding debates, the note when you expand it says "The following is an archived debate. Please do not modify it." Is there a way to change that? (I know you can change the "Reason"). --
ZimZalaBimtalk12:28, 10 June 2008 (UTC)
In this diff
[10] the only change I introduced into the editing box was the addition of a time and date on a comment, yet the result was the loss of a significant amount of text, and of the class and importance parameters from the Wikiproject template. What happened?
DuncanHill (
talk)
14:45, 10 June 2008 (UTC)
Looks like you accidentally edited an old revision when you went back to find the timestamp. ---
RockMFR15:16, 10 June 2008 (UTC)
Hmm, yes I see what you mean - usually I just view the diff in pop-ups, must have clicked on something without meaning to!
DuncanHill (
talk)
15:20, 10 June 2008 (UTC)
What gives? The div tags are showing up at the bottom of every image page:
<div class="linkstoimage" id="linkstoimage">The following pages on the English Wikipedia link to this file (pages on other projects are not listed):</div>
Got broken again by an attempt to add plural support (totally unnecessary, probably, since it should work correctly in wikitext form). I've reverted the change. --
brion (
talk)
01:58, 11 June 2008 (UTC)
I can confirm the first and fourth formulae contain a symbol  () which doesn't render in LaTeX, apparently. x42bn6TalkMess03:10, 8 June 2008 (UTC)
The first and fourth expressions have a bogus byte "", 0x81. This is invalid UTF-8 (I think, haven't calculated it out) and makes no sense to have in LaTeX formulas in any event. I don't know how it got there, but try manually retyping the bad formulas instead of copy-pasting, and it should go away. Like this:
It's valid UTF-8 encoding, it's just not for a useful character -- it's in a control character block, but isn't assigned to any... control... :) --
brion (
talk)
16:44, 9 June 2008 (UTC)
Not that it's really relevant to the question, but 0x81 isn't valid UTF-8. Bytes 0x80 and up don't have meanings on their own; they have to be used in combinations of two or more to form extended characters. You're probably thinking of Latin-1.
rspeer /
ɹəədsɹ 08:12, 10 June 2008 (UTC)
Oh, never mind that, I'm at the wrong level of abstraction. Since Wikipedia presumably contains only valid UTF-8, the formulae probably actually contain the bytes 0xc2 0x81, which encode U+0081, the useless control character you were referring to.
rspeer /
ɹəədsɹ 08:20, 10 June 2008 (UTC)
This may be a dumb question, but is there an existing facility to list the edits one single editor has made to one single article? Ideally, this would be a special page ('cause it would return clickable diffs), next a toolserver page, last-best an API call (which I could use though). Thanks!
Franamax (
talk)
07:07, 10 June 2008 (UTC)
Each bulleted entry should, in almost every case, have exactly one navigable (blue) link
The link being disambiguated should be the first word or phrase in each entry
I'd like to create a template, that only appears in edit-mode, at the top of any page tagged with a disambig template (over 100,000 articles) e.g.
A Wonderful Life. Something like the
MediaWiki:Talkpagetext box that appears when you edit talkpages. It would contain 2 or 3 lines cautioning against the most frequent
MOS:DAB mistakes, such as "Each bulleted entry should, in almost every case, have exactly one navigable (blue) link", and a link to the style guideline.
Technically yes. With javascript, it should be possible to detect the presence of the disambig template on a page (using the disambig id, which hopefully is only used in that template). If it is detected, the javascript could append an editintro parameter to the edit link that would translude a message on the edit page. The difficult part would be gaining consensus for this change :) ---
RockMFR21:32, 10 June 2008 (UTC)
Very nice. Nitpick: Should 2nd bullet read "The link being disambiguated should be ..." since, in the case where an entry has a redlink and a bluelink, the redlink goes at the beginning? --
AndrewHowse (
talk)
00:51, 11 June 2008 (UTC)
Sure, example changed. Feel free to update it yourselves, it's mainly for demonstration purposes at this point. I've also updated RockMFR's sandbox, so his implementation displays the same thing. --
Quiddity (
talk)
01:30, 11 June 2008 (UTC)
In the popular meta-script
automod I see a number of paramaters, such as "addafter" and "addbefore", that can be passed to index.php when calling action=edit. However, I am unable to find documentation for them or other such paramaters at MediaWiki, English Wikipedia, or in the raw source code. (They exist, as I have used them in a new script I am working on.) Does anyone know where I can find such information? Thanks. --
Thetrick (
talk)
00:42, 12 June 2008 (UTC)
That's because they're not MediaWiki parameters, they're parameters used by that specific script. Look at the function auto_mod, it reads those parameters out of the query string and takes various actions depending on them.
Anomie⚔00:55, 12 June 2008 (UTC)
Thank you for clarifying that. For some reason the breakpoints I had set in that function aren't being hit, so I haven't paid attention to it before. I take it that the page loads, and then the function manipulates the text in the edit box? --
Thetrick (
talk)
02:17, 12 June 2008 (UTC)
Watchlist auto-refreshing...
This has only recently started to happen -- maybe in the past two days or so. Basically, the way it has worked is that the watchlist page (and any other WP page, not sure if it's affected elsewhere) would stay at the last it was last on, like most pages. There's some pages on the web that will automaticallyr refresh whenever you switch pages (which can be annoying as it also tends to put you back up top), but here it seems that it's only doing it if I move two pages deep (like click "older edit" twice) or more. Anyone else notice this, and is there a way to fix it...or perhaps it's a bug?
♫ Melodia Chaconne ♫ (
talk)
16:10, 9 June 2008 (UTC)
Well assuming you don't have some javascript set up to automatically refresh a browser window or tab which contains your watchlist (Brion would probably kill you for that), you probably mean that you are clicking the "back" button in your browser (to go back to your watchlist) and finding a newer list of edits than when you last saw it. That would mean that your browser is for whatever reason no longer caching the watchlist page (maybe caching is disabled now, or maybe you changed browsers), or that your browser did cache the page but it has been removed from your cache as "too old" or to clear space to cache something else. So you might start by adjusting the page-caching settings in your browser. —
CharlotteWebb17:19, 9 June 2008 (UTC)
That's just it, I haven't changed any settings, nor has it ever happened like this before. The only difference is the updating of two plugins (I use Firefox) but nothing beyond that has changed. I can't imagine either of those plugins affecting it. I figured something in WP's software changed recently since it seems to only be affecting things here that I've noticed.
♫ Melodia Chaconne ♫ (
talk)
19:08, 9 June 2008 (UTC)
And now I can determine it's happening at more than one computer (and this one has no FF plugins), so it's almost surely something that's changed on WP or the wiki software itself. Perhaps I should submit to bugzilla? I dunno how to word things really, but I'd like to know if anyone else has noticed this...
♫ Melodia Chaconne ♫ (
talk)
11:37, 10 June 2008 (UTC)
Is the watchlist just sitting open in its own window or tab, and you just notice that it's changed at some point? Or are you navigating away from it by clicking links within that window/tab, and later returning to the watchlist in that window/tab you've been looking at other pages in, and now it's different from when it was last up?
If it's just sitting there in a window/tab, it should never change. I see no automatic refresh code snuck into the site JS, and you have no custom user JS, so there is nothing that should cause it to change while it's sitting open.
If you navigate away within the window/tab and then return to the watchlist by clicking a "watchlist" link, following a bookmark, etc, then it will certainly be loaded anew, with a fresh, current list of changed pages.
If you navigate away and then return to the watchlist by clicking the "back" button, then the behavior may depend on the browser, and the browser's caching settings, and whether you're logged in through the secure.wikimedia.org SSL interface. In most cases it will probably give you the version you had open previously... unless it's fallen out of the browser's cache... or the browser decides it shouldn't cache it. (The page is marked as not cacheable and must be reloaded every time it's visited, but 'back/forward' button navigation often is a little different.) --
brion (
talk)
22:40, 10 June 2008 (UTC)
It's from navigating away (but still on WP, using "older edit" links), and then back. It seems to happen randomly now, if at all the past few hours...could have been already fixed or who knows. But as I said, it's not a setting I changed since it happened on two seperate computers (in two completely seperate locations).
♫ Melodia Chaconne ♫ (
talk)
00:25, 11 June 2008 (UTC)
As Brion said, behavior when hitting back is not reliable. If the browser still has a cached copy, it will use it. If it doesn't, for instance if it's been cleared to make room for other pages (quite possible after a long browsing session), it will re-request the page. You can encourage it to cache pages for longer by increasing the cache size; in Firefox 3 this is Preferences -> Advanced -> Network -> Offline Storage, set the cache size to whatever you like. The larger you set it, of course, the more disk space it will use up. —
Simetrical (
talk •
contribs)
01:06, 12 June 2008 (UTC)
RFC 2616 has a nice little rant about browsers trying to use cache control settings for the browser history:
13.13 History Lists
User agents often have history mechanisms, such as "Back" buttons and
history lists, which can be used to redisplay an entity retrieved
earlier in a session.
History mechanisms and caches are different. In particular history
mechanisms SHOULD NOT try to show a semantically transparent view of
the current state of a resource. Rather, a history mechanism is meant
to show exactly what the user saw at the time when the resource was
retrieved.
By default, an expiration time does not apply to history mechanisms.
If the entity is still in storage, a history mechanism SHOULD display
it even if the entity has expired, unless the user has specifically
configured the agent to refresh expired history documents.
This is not to be construed to prohibit the history mechanism from
telling the user that a view might be stale.
Note: if history list mechanisms unnecessarily prevent users from
viewing stale resources, this will tend to force service authors
to avoid using HTTP expiration controls and cache controls when
they would otherwise like to. Service authors may consider it
important that users not be presented with error messages or
warning messages when they use navigation controls (such as BACK)
to view previously fetched resources. Even though sometimes such
resources ought not to cached, or ought to expire quickly, user
interface considerations may force service authors to resort to
other means of preventing caching (e.g. "once-only" URLs) in order
not to suffer the effects of improperly functioning history
mechanisms.
As Brion and I both clearly stated, this is unpredictable behavior that may change at any time for any reason or no reason at all. Your browser is not guaranteed to maintain a copy of the watchlist in cache that it uses when you hit back. It may or may not do so, depending on any criteria its authors thought were reasonable. Any change in your browsing habits or your computer that may have occurred at any time recently could have caused the issue, even if your settings remained the same. If, for instance, recently there happen by coincidence to have been relatively involved edits on your watchlist, requiring you to browse away from your watchlist for an extended period of time, that might cause the issue. If you happen to have been interrupted more often lately, that might cause it. For all I know, your browser might notice that you're getting low on disk space and decide to be more aggressive in purging caches.
I suggested a settings change that might fix the issue, even if that setting has never been changed before and the current setting used to be working fine. It might not fix the issue. In any event, this is a client-side decision that's totally out of Wikipedia's control. If you want to complain to anyone, complain to Mozilla. —
Simetrical (
talk •
contribs)
21:48, 12 June 2008 (UTC)
It means that you have also been logged out of ALL the other sites you have unified to your account. However, WikiSpecies is not included, because it is not on SUL as of yet, IIRC.
Soxred 9303:11, 12 June 2008 (UTC)
I've just checked, and my SUL works on wikispecies. I had to sign in their manually, though, which I don't have to do anywhere else.
Algebraist09:34, 12 June 2008 (UTC)
Single User Login != CentralAuth global session. Both systems were activated around the same time hence the confusion. SUL is active on wikispecies, apparently though, global login does not encompass wikispecies yet at this moment. See also
bug 14407. --
TheDJ (
talk •
contribs)
11:21, 12 June 2008 (UTC)
Yeah, on bug 14407, Tim Starling said ""The login system is such that it requires users to wait at the login screen while they are logged in to all relevant domains. Due to security issues, we can't log users in to *.wikimedia.org (where Wikispecies is located), so we have to log users in to the third-level domains individually. I added Meta and Commons to the list due to their high traffic transferring from other wikis. There are 32 wikis on *.wikimedia.org which are excluded, plus www.mediawiki.org. There's a clear incentive to draw the line somewhere, and that's where I've drawn it. ""--
penubag (
talk)
20:21, 12 June 2008 (UTC)
I have started a thread at
Template talk:Convert#expression error after two recent incidents when users reported seeing {{convert}} appear as a large red expression error. I have not seen any mention of this problem off of the
Main Page, so it seems to be a bit of mystery. More explanation and several screenshots are at the page. Help from the technically inclined would be welcome. -
BanyanTree11:00, 12 June 2008 (UTC)
Hi! I'm
w:hu:User:Gothika from Hungarian Wikipedia. When I try to make my global account I find that somebody is using my name (Gothika) here in English Wikipedia, but that somebody was blocked for vandalism (see
User talk:Gothika). It is possible to delete that registration ? --
85.66.62.70 (
talk)
12:06, 12 June 2008 (UTC)
The
list of baryons has been nominated for Featured List once, but failed on various account. I've fixed most of them, and the only one remaining is from
Crzycheetah, who experiences
weird scroll bar problems, caused by some use of {{SubatomicParticle}}/{{PhysicsParticle}} templates. The specific templates causing problems were identified to be those in references T. Aaltonen et al. (2007a) and T. Aaltonen et al. (2007b), and the {{SubatomicParticle|charmed Xi+}} in the Particle Classification section.
I can't find anything different in the use of these templates in those instances than in the rest of the page. I also tried looking in the template codes and I can't see anything there either. Could someone take a look at the
List of baryons and {{SubatomicParticle}}/{{PhysicsParticle}} pages to see if he/she can find a problem?
I just had a bot tag an image as orphaned yet the image is in the article. In fact, all the images in the article are not showing the article as linked on the respective image page. See
Charlotte, North Carolina.
Morphh(talk)13:49, 12 June 2008 (UTC)
Not as far as I know. That also sounds like it would be "dangerous" if possible, because then certain articles would probably "censor" some sections from the TOC (like a Controversy) section. {{TOClimit}} exists, but it isn't allow for as much pick-and-choose.
GaryKing(
talk)15:26, 12 June 2008 (UTC)
Hello. I am trying to improve the
Castlevania page by adding
references, but cannot find the source page. All I find is {{reflist|1}}. Anyone know how to find it? Sorry if this is the wrong page to ask my questions. Thank you. ---
Fallen_RealityT/
C05:02, 13 June 2008 (UTC)
The references are in-line with the main article, bounded by <ref></ref> tags. For example, there should be one in
this section. The reflist creates the references from those tags. See
Citing sources for more information. --
MASEM05:09, 13 June 2008 (UTC)
Strange problem: I'm 'half-blocked'
A few days ago, I started getting a message telling me that my IP or IP range was blocked whenever I went to a page-editing screen whilst not logged in. (Not that I ever edit without logging in first anyway, but I sometimes go to edit screens whilst not logged in so that I can see a page's code).
I can still edit when I am logged in. What's going on? Have I been 'caught in the crossfire' between some other user and the admins? --
The Machine (
talk)
00:45, 8 June 2008 (UTC)
Apparently, your IP is blocked. But that's a feature to MediaWiki, unless autobloc kis enabled, logged-in users can still edit.
Soxred 9301:46, 8 June 2008 (UTC)
Actually this will prevent other people's auto-blocks from affecting you, and also prevent you from forgetting to log in before editing. Might be a blessing in disguise. —
CharlotteWebb11:00, 8 June 2008 (UTC)
It seems like someone's softblocked your IP, or a range containing it. Most commonly this is because there's vandals on the same IP range as you, but legitimate contributors (such as you) there too. The only inconvenience to you should be that you're forced to log in to edit, so the software can tell that you're one of the legitimate users (generally speaking, creating accounts from inside such ranges is prevented, and vandal accounts which are inside the range are blocked, so only legitimate users inside the range have unblocked accounts to log in with. This sort of softblocking is especially common on school IPs (where there are often many vandals and a few legitimate users editing from the same IP), but can happen elsewhere. --
ais523 16:14, 8 June 2008 (
UTC)
Nothing bad will happen if you just edit logged-in as usual. If you are very concerned about the IP block and have some time to spend you may be able to find out by writing to unblock-en-l what administrator issued it. This doesn't guarantee the block would be lifted, because it could be there for a good reason.
EdJohnston (
talk)
15:45, 10 June 2008 (UTC)
if you visit this page, go full-screen (F11), page down to the bottom (so you can see the yellow 'Nervous system: Sensory systems / sense' table and external links), then put your mouse up top to bring FF3's address bar down, the external links and yellow box dissapear.
can anyone tell wether that's a bug with the js at the bottom of the page, wp, or ff?
It seems to work fine in Firefox 2-point-whatever-the-most-recent-stable-is. This sounds like an issue with Firefox 3 - yay for beta testing working all that out.
Hersfoldnon-admin(
t/
a/
c)00:53, 13 June 2008 (UTC)
cheers. someone must have allready reported the bug; upgrading to latest FF3 fixes it. Yay for not having to fill out a bug-report :-) --
Dak (
talk)
Warning
I followed this link
User:Dustihowe/monobook.js, to be greeted by a page of code and a dire warning. I am a coding illiterate, so I have no idea whether visiting the page has exposed me to any risk. Reassurance needed please.
jimfbleak (
talk)
06:14, 13 June 2008 (UTC)
Relax: Viewing the page isn't risky. It would have to be executed, perhaps by including that page in another page which is loaded. Two examples of that are at the bottom of that page with importScript. —
EncMstr (
talk)
06:26, 13 June 2008 (UTC)
I've clarified the warning message a bit. Now that I'm thinking about it... this warning probably needs to be in other places, too. ---
RockMFR18:07, 13 June 2008 (UTC)
Links to logs on File Upload pages
I just stumbled across the fact that on all file upload pages, the links near the top to the deletion log and the Commons log for that image link to the logs for the page "Image:$1" instead of "Image:(FileName)". --
Spring Rubber (
talk)
07:17, 13 June 2008 (UTC)
Someone had removed content, and only part of the damage was repaired. I see someone has fixed it now.-
gadfium09:15, 13 June 2008 (UTC)
Unified watchlist
Now that we have unified login, we can access watchlists on other Wikimedia sites without logging in from our browsers. Has anyone written a javascript to scrape the watchlists off other wikis and display them all on the same page? If not, can we? —
Omegatron (
talk)
17:18, 7 June 2008 (UTC)
Great idea! I'd like an other project new messages banner too:
While they don't have unified watchlists, Wikia's similar system of unified login does tell you when you have new talk page messages on any other of their wikis. In other words, we know it's possible. --
Ned Scott03:38, 13 June 2008 (UTC)
Making various types of spam/vandalism that little bit more difficult
Regarding the anontalk vandal, who spams a URL into articles as plaintext, apparently using a bot, and is not caught by the URL autoblocker. If we required a captcha confirmation for adding certain (manually) blacklisted strings to Wikipedia articles, in the same way that we currently do for non-autoconfirmed users adding URLs, the spammer/vandal would need to expend a tiny bit of human effort for each edit. Even when confirmed, these edits should be logged in a dedicated log, to allow for rapid cleanup. If we did this for not only the article content, but also the edit comments, this would perhaps also help discourage several other kinds of bulk vandalism, eg Grawp.
Possibly there should even be both a greylist (which fires the CAPTCHA generator) and a blacklist (which would prevent anyone but admins from adding the string to any article). --
The Anome (
talk)
13:18, 14 June 2008 (UTC)
I'm pretty sure the
ConfirmEdit extension, which we have installed, already supports this; the relevant configuration parameter is $wgCaptchaRegexes. Just need to bug the server admins to add some new entries to it. (There's also
$wgSpamRegex, which is part of MediaWiki core, and can be used to disallow edits matching certain regexps entirely.) —
Ilmari Karonen (
talk)
21:02, 14 June 2008 (UTC)
I have a simple question: Why does this link work:
Talk:Metalcore#8 and this not:
Talk:Metalcore#48? The first one redirects me exactly to the section 8; but the other one just to beginning of the talkpage. Why? Thank you very much for help. :)
LYKANTROP 15:59, 12 June 2008 (UTC)
It was appended due to a bug. If you want the exact reason for the bug, look at the link below to the revision in which it was fixed, but don't expect to understand it if you don't know something about programming. —
Simetrical (
talk •
contribs)
18:06, 13 June 2008 (UTC)
Yes, and all similar errors that probably occur on other pages. We're only up to 1.43.0-wmf.13 (8eaf4e5); when we're up to r36248 or higher, it should be fixed the next time the page is edited. —
Simetrical (
talk •
contribs)
18:06, 13 June 2008 (UTC)
Ok, firstly, I love the fact we can now see log entries on our watchlists- I think that is an excellent feature. However, I sometimes find that my watchlist will be swamped by many log entries which I am not too interested in (a large batch of images I tagged with no source a few days before, for instance) but I will want to see the one or two edits between them. So, I am asking- is there a way to remove logs from my watchlist temporarily (like you can with your own edits, minor edits and/or bot edits) and if there isn't, would it be too hard to create one?
J Milburn (
talk)
13:50, 14 June 2008 (UTC)
I would prefer just the previous view without the log entries. I don't really care if some redirect has been deleted, or if some page I am watching has been protected... --
ReyBrujo (
talk)
16:54, 14 June 2008 (UTC)
There are options to filter watchlist entries by namespace. There is also "Hide bot edits"; "Hide my edits"; "Hide minor edits". Surely some option to "Hide log entries" would be possible?
Carcharoth (
talk)
20:23, 14 June 2008 (UTC)
AlexSm, thankyou, I am checking that now, but I still feel that what Carcharoth says could very easily be done. Of course, I'm no techy.
J Milburn (
talk)
20:38, 15 June 2008 (UTC)
Remaining logged in even after clearing just about everything
Resolved
For increased security, I have Firefox set up to clear out private data every time I close it, specifically deleting browsing history, search history, downloads, saved form information, cache, saved passwords (which don't exist), and authenticated sessions. So long as I didn't check the "Remember me" box, which I don't any more, I would have to log in every time I pulled up Firefox, just the way I like it. However, recently (within the last few days), I've been logged in every time I start up the browser. When I clear out private data manually, I'm still logged in. I keep a close eye on my computer, but even so, I'd rather not be accidentally logged into my admin account when someone else happens to be on here. Both of my accounts are globalized (that is,
User:Hersfold and
User:Hersfold non-admin which I'm currently using are both set up to be global accounts, however this is the only project I've logged into the latter for obvious reasons). Was there some recent configuration change that caused a change in how accounts are logged in? I've had my admin account globalized for some time now and it wasn't having this issue until very recently. As a warning, I may not respond to any suggestions left in a timely manner, as I only have occasional access at the moment, but I'll check the archives if responses are left. Thanks.
Hersfoldnon-admin(
t/
a/
c)00:50, 13 June 2008 (UTC)
I can confirm this behavior. When logging in with the "remember me" checkbox unchecked, closing the browser does not remove the centralauth_Session cookie (which is set for 24 hours after login). In the meantime, make sure you click "log out" to force your account to be logged out. ---
RockMFR03:18, 13 June 2008 (UTC)
He's technically correct, though: you're not clearing your cookies, as you indeed state above (the list of cleared private data you give does not include them). MediaWiki uses cookies to keep you logged in, so if you do clear your cookies, it's technically impossible for you to stay logged in. That said, the persistence of the centralauth_Session cookie does seem to be a bug, and I've filed it as
bugzilla:14564. —
Ilmari Karonen (
talk)
20:22, 16 June 2008 (UTC)
The Preferences says "keep me logged in" if I check the box. I don't check the box. Until unified login, I was logged out every time. Now I'm not unless I explicitly log out. It's a Wikipedia problem, not a user problem.
Corvus cornixtalk20:27, 16 June 2008 (UTC)
Works for me. I log in, check "keep me logged in", and if I click "Clear Private Data" (or not, it doesn't matter) in my browser (or close my browser with "clear private data when I close firefox" checked), I'm logged out of WP. Granted, I had to go into the private data settings and check cookies (which are not cleared by default). You can test this by clearing your data, then showing your cookies. If none remain, you'll have to re-login. If any cookies remain, then it's a browser problem, not WP. --
Kbdank7120:51, 16 June 2008 (UTC)
Uhhuh. Wikipedia should not require users to have to clear private data when there is a box in Preferences which is supposed to handle this. And why did it suddenly become a browser problem with no browser update, but there was a change to the way Wikipedia handles logins?
Corvus cornixtalk22:12, 16 June 2008 (UTC)
I'm primarily an image reviewer...I have repeatedly run across individuals who are otherwise great contributors, but who apparently have no understanding of intellectual property law and/or Wikipedia's
non-free content policy, and seemingly have no desire to improve in this regard. I'd like to propose that admins have an ability to block a user from uploading, without affecting that user's ability to contribute in other areas. I'm interested in thoughts regarding this.
Kellyhi!20:12, 14 June 2008 (UTC)
It certainly would be a good alternative to simply blocking problem uploaders. I think the best way to go about this technically is to make the upload right a
user right of it's own, set it autoconfirm, then give admins the ability to remove the user right.
1 != 220:14, 14 June 2008 (UTC)
It would only deal with future problem users. The current ones would have to have the right switched off, and those discussions will be interesting, especially if people take differing interpretations of what is abusive and how many "second chances" to give.
Carcharoth (
talk)
20:31, 14 June 2008 (UTC)
If users are unable to contribute productively and legally, they shouldn't be contributing at all. From the technical side, Until is of course correct that the right could be similar to rollback, however, you'll need to demonstrate community support for such an idea, as it wouldn't be uncontroversial.
I will say that from a social standpoint, for years we have tried to not create social classes, and slowly we seem to be headed more and more in that direction, for better or for worse. It may be better to just get on with it and create a 'trusted' group, something similar to a semi-admin, instead of continually creating new usergroups for every userright. But, as I said, there's never been a clear indication that those in the community want such a thing – in fact, it has always appeared that it was something they didn't want. But with this community, who can tell? ; - ) --
MZMcBride (
talk)
20:35, 14 June 2008 (UTC)
Well currently the only "class" of user that we can prevent from upload are users blocked from any editing whatsoever. This would allow for a lesser action instead of being forced to block. I don't think this introduces a new "class", more it allows for more specific revocation of privileges that all "classes" normally get, and that all "classes" can normally lose.
1 != 220:43, 14 June 2008 (UTC)
Well in real life "trust" is not a package deal. I have friends whom I trust not to lie to me, and I have friends whom I can trust not to steal from me. For the most part they aren't the same people. —
CharlotteWebb11:32, 15 June 2008 (UTC)
Another great thing about making upload 'autoconfirm' is that it would eliminate a lot of the problems caused by people creating accounts solely to upload penis pictures for vandalism. :)
Kellyhi!20:42, 14 June 2008 (UTC)
Uploading is already autoconfirm, it is just autoconfirmed in the same group for page creation and editing semi-protected page. The people that upload those pictures use
sleeper accounts.
1 != 220:43, 14 June 2008 (UTC)
I don't have a link, but I remember it happening. Was a few months ago, I think. Commons adopted it, then us shortly after, IIRC. --
Ned Scott06:02, 15 June 2008 (UTC)
No link, but if you create a new account it will not allow uploading till you are autoconfirmed.
1 != 220:55, 14 June 2008 (UTC)
(ec *2)I'm generally opposed to upload blocking based on the same logic MZMcBride, if they won't follow the rules.. block'em.. That said, I did setup a site JS kludge on commons ages ago for upload blocking (edit the bad user's user JS to give them a cookie that made the submit button grey on the upload form) .. there it was useful because people would mass upload without ever checking their talk page because they would go back to their home wiki. No one seemed to use it on commons. Since no one used it on commons, I doubt it would be all that useful here.
Another interesting proposal was to require a confirmed email address for upload: it got a lot of support on the mailing lists, but never was implemented. The idea was that we could be more aggressive about deletions but avoid incorrectly deleting good stuff because we'd be able to contact users to clarify their claims.
If we were to limit upload to autoconfirmed users, what I'd prefer to do is allow non-autoconfirmed users to upload works which they claim to have created. This could easily be done with site JS. (who cares if it's easy to evade, we want a idiot filter, not high security)... If you really created the work yourself the level of copyright understanding required to make a good upload is pretty minimal. If someone falsely claims to have created something just to evade the limit it is generally easy to spot those cases and delete them (oh you made this frame of StarWars? really...).
Limiting all uploads to autoconfirmed really would curtail the release of genuinely free content... I think we should rather have to delete 100 stupid copyvios in order to get one more image actually freely released. --
Gmaxwell (
talk)
20:54, 14 June 2008 (UTC)
Gmaxwell does have a good idea, though. {{Self}} and {{PD-self}} are no problem, and vios are easy to spot, it's the non-free and more complicated public domain ones that are hard to spot. I suppose someone could game this by uploading as user-created and changing it later, but that would be easy to spot too. An interim step is better than nothing.
Kellyhi!00:29, 15 June 2008 (UTC)
<unindent> I want to restate the problem - we have many users who do great research, write great prose (even featured articles), are even administrators - but who don't understand copyright. I want a way to keep those users while limiting damage.
Kellyhi!21:01, 14 June 2008 (UTC)
Right, but you don't have to understand copyright any more when releasing a self made image then you do when writing prose. A big part of why we are restrictive about copyright (even rejecting things we could legally take, or at least "get away with") is because we're trying to encourage the creation and distribution of freely licensed stuff. Pretty cruddy that a prose-writer can contribute without creating an account, but a photographer has to wait several days. --
Gmaxwell (
talk)
21:05, 14 June 2008 (UTC)
I wonder what sort of good content writer, when receiving something like {{Uw-upload4}} wouldn't stop and say "Jeez, I need to learn more about copyright before I upload more images" MBisanztalk21:07, 14 June 2008 (UTC)
You'd be surprised. I don't want to name names (<cough>
SlimVirgin</cough>). Joking aside, there are others for whom a total block would simply be impossible but an upload block would be appropriate.
Kellyhi!21:10, 14 June 2008 (UTC)
Erm, I really doubt that. If you tried to get an admin to remove the upload right from another admin (though I doubt that would be even technically possible), I imagine it would be seen as a slap in the face. --
MZMcBride (
talk)
21:16, 14 June 2008 (UTC)
Hmmm, I was thinking strictly in the context of editors, not admins...I guess if an admin is violating site policy it has to go to ArbCom, even if they are doing so strictly as an editor (I guess). To my knowledge, RfA has absolutely no screening for copyright knowledge before adminship, unlike Commons. But what, say, if
Giano was violating copyrights on uploads (totally hypothetical example) - would a block be possible?
Kellyhi!21:22, 14 June 2008 (UTC)
There's technical ability and there's social ability. My point is that for long-time contributors, removing a userright from them will be seen as impolite and will be interpreted poorly by them, I think. --
MZMcBride (
talk)
21:24, 14 June 2008 (UTC)
(ec)Nope, block would last maybe 20 minutes on a good day in that case. I suspect the number of prototypes of good content editors on whom an image upload block would stick, would be a tiny number. If one does good content, they almost always have enough admin friends, to make sure no sanction sticks, and if they don't, then no admin would bother splitting hairs between a total block and an upload block. MBisanztalk21:26, 14 June 2008 (UTC)
Wouldn't those problems go away with time (as old contributors fade away and new contributors come online)? And users who have had upload rights removed as users would be much less likely to be successful at a bid for adminship. Also, the "admin" thing is irrelevant here, I think - after all, who uploads images in their capacity as an admin? (Aside from some Main Page housekeeping stuff.) 99.9% of the time, anyone who uploads images does so as an editor, not an admin. I thought the roles were supposed to be separate.
Kellyhi!21:30, 14 June 2008 (UTC)
Of course, admins who are less concerned about copyright would be able to restore the "upload" right to themselves and anyone else who has had it removed. —
CharlotteWebb11:32, 15 June 2008 (UTC)
I think some sort of semi-voluntary upload restriction would be better (and easier) than a MediaWiki based forced restriction. Give the user an option - stop uploading images, or only upload after receiving approval for each image, or be blocked. Basically a topic ban from image uploads. If necessary, a list of users on upload restrictions could be set up for monitoring.
Mr.Z-man21:49, 14 June 2008 (UTC)
It's a selective use of my power to block. Since it affects the user less harshly than a full block, which I'd also be entitled to do for the same offense, I don't see why I shouldn't be entitled to enact the weaker sanction instead of the stronger one. A fortiori. Go ask the people I banned if they would rather be blocked instead.
Fut.Perf.☼07:28, 15 June 2008 (UTC)
I see a lot of admins saying if the user can't follow the image rules then just block them. Well, I don't want to block them, I would far rather I could solve the problem without preventing the user from contributing elsewhere. If the situation calls for a block the button will still be there, but I for one would like the alternative of taking lesser action.
Repeat abuse of the upload option is not always an issue of bad faith, often it is simple cluelessness in that one area.
1 != 214:37, 16 June 2008 (UTC)
I feel the same as Until(1 ==2). Very often there's no bad intentions from problematic uploaders. It would also allow one to stop an image revert war temporarily without a full block. --
Ned Scott00:59, 17 June 2008 (UTC)
We've traditionally been leery of partial blocks like this; I'd tend to prefer a "delay" or "hold for review" system rather than hard blocking for "don't quite trust this guy but don't want to kick him out" or "not sure this noob isn't a vandalbot yet". I'm hoping we can get some things developed along these lines in the next few months. --
brion (
talk)
16:21, 17 June 2008 (UTC)
It seems to happen with any combination of digits and letters a-f as long as there are at least 3 colon-separated groups of 1-4 characters, as long as it doesn't conflict with an existing page or interwiki prefix. If I had to guess, it's probably (mis?)interpreting it as some sort of
IPv6 address; the same thing happens if you enter an
IPv4 address that doesn't have an article, such as
192.0.2.0.
Anomie⚔02:44, 17 June 2008 (UTC)
Yeah, it's getting picked up in the IPv6 check. I'm not sure why the search is coded that way at all... Why would someone use the search box to find an IP's contribs? ---
RockMFR03:38, 17 June 2008 (UTC)
I actually use that feature quite frequently, but I agree, for the majority of users its not very useful.
Mr.Z-man03:59, 17 June 2008 (UTC)
# Entering an IP address goes to the contributions pageif(($title->getNamespace()==NS_USER&&User::isIP($title->getText()))||User::isIP(trim($searchterm))){returnSpecialPage::getTitleFor('Contributions',$title->getDBkey());}
This is obviously a weak test for IPv6, but a more
thorough test might not actually help much. Perhaps it could just suggest in a link at the top "Did you mean Special:Contributions/1.2.3.4?" or such. --
Splarka (
rant)
07:13, 17 June 2008 (UTC)
It should probably be fixed to use the existing regexps defined in
IP.php. The IPv6 regexp there isn't perfect either, but better than the one currently used in User.php. —
Ilmari Karonen (
talk)
00:56, 18 June 2008 (UTC)
30 day login limit
I've just now noticed that your login cookie expires after thirty days. This is not something I like, as I will often see an error on a page (usually grammar-related) and correct it quickly, giving me no time to log in. Plus, I've never liked being forcibly logged out of anything. Seems I should have the right to decide how long I want to be logged in on any computer.
So, the question becomes, Is there any way for me to tell the system "I want to stay logged in as long as possible (from this computer). Screw the (virtually nonexistent) security risk"? If somebody breaks into my computer, I've a lot more to worry about than my WP account being blocked.
Something that has been a problem for a long time ... and I finally got fed up with it on the Wiktionary and fixed it. The [edit] links on the right-hand side of the page "wander" if there are multiple images and/or RHS float boxes. It can be very difficult to find the right one for section editing. Images also appear over the horizontal rules in L2 headers. See
Øresund for an example, and compare
wikt:saxophone, where the [edit] links stay on the correct header line, shifted left of the image, and the L2 horizontal rule does not run into the side of (under) the image.
/* make headers include contained floats, so they don't wander around (yes, is magic, it tells the browser the tag is a container) */
h1, h2, h3, h4, h5, h6 {
overflow: hidden;
width: auto;
}
See discussion at
T3629. Your solution
was tried, and was reverted because it caused errors for RTL display in some browsers, and caused the headers to vanish entirely on IE5 for Mac. If those issues were addressed with appropriate workarounds, it could be committed to the software in modified form.
Explicitly specifying width (even though the default) seems to be part of the magic with some browsers. The code can be hidden from IE5 etc (the s/w already does various bits of that). Is RTL an issue if it was done here, not in the WM .css? (And this is why I wanted to get this looked at ;-)
Robert Ullmann (
talk)
16:49, 17 June 2008 (UTC)
External links search slightly broken?
I was tracking down some spam links and searched external links for the spammed site
"home.tiac.net/~wroger/", getting no results (link is to search results, not site). I then searched for
"*.tiac.net/~roger/" and got a lot of unexpected results. The search seems to be ignoring everything after the domain, but only when the * is used. Is this a known issue?
Delicious carbuncle (
talk)
16:44, 17 June 2008 (UTC)
You don't get a result for the first because the matching link doesn't have that final /.
You don't get narrowed results for the second because the way the index works we can't efficiently do wildcards for both the domain *and* the path; if there are a large number of potential matches, the query ends up being very slow (eg, searching through ten million links to *.wikipedia.org for the three that match a particular path). Ideally it should be possible to do that more efficiently, but atm I think the DB isn't very smart about that double-wildcard case, so we just drop the further wildcards. --
brion (
talk)
21:07, 17 June 2008 (UTC)
I wasn't expecting a result for the first case (although you're right about the slash) - it was really the second that puzzled me. It seemed a bit odd. Thanks for your response.
Delicious carbuncle (
talk)
21:30, 17 June 2008 (UTC)
Request for character
Hi, at music articles we used to use "b" for the flat sign (as in "B-flat"), then we began using ♭, but it's too big and has too much space around it. Now we're using {{music|b}}, which produces ♭ (as in B♭), which is good, but cumbersome. Can we simply replace the ♭ character with the {{music|b}} one (♭)? It looks much more professional and I think we should have an actual character for it rather than the workaround we're using now. By the way, copying and pasting the professional-looking one (♭) without using the "workaround" template produces the less-professional one (♭), so that's not an option.
Badagnani (
talk)
20:02, 16 June 2008 (UTC)
All the {{
music}} template does is wrap the ♭ sign in some HTML <span> tags with
CSS styling. You could, if you wanted, write it out as "<span class="music-symbol" style="font-family: Arial Unicode MS,Lucida Sans Unicode;">♭</span>", but I think the template is, all in all, more convenient. (BTW, for me, using Firefox on Ubuntu Linux, the tags have absolutely no effect: the symbol looks identical (and just fine) with or without them. This seems to be a Windows-specific (and possibly IE-specific) issue.) —
Ilmari Karonen (
talk)
20:56, 16 June 2008 (UTC)
Thank you. The thing is, the easiest way for editors (using the symbol itself) looks unprofessional, and the more cumbersome way (using a template) looks very professional. May I ask, again, if the output of the template can be created as a discrete character that can be entered without the use of the template? Or is it the case that the two display the same way on your computer? For me, the one with the template is smaller, with less space around it, and much more similar to the way such a symbol would appear in professional publications.
Badagnani (
talk)
22:15, 16 June 2008 (UTC)
This is not possible. Both are the same "discrete" character. The only thing that differs, is that with the template, a different
font is selected on the computer of the user. In the font "Ariel" the character is probably a lot "smaller" than in the default font used to display wikipedia text. --
TheDJ (
talk •
contribs)
00:02, 17 June 2008 (UTC)
I assume it would also not be possible to make the symbol itself a template? Kinda like the way ~~~~ acts as a template? —
trlkly10:58, 17 June 2008 (UTC)
Theoretically, the software could probably be adapted to do that, but I very much doubt the Developers would write such a hack. What we can do, is make {{♭}} produce the same result as {{
music|b}}, but I doubt that will save you much time when you are working on this stuff. --
TheDJ (
talk •
contribs)
13:07, 17 June 2008 (UTC)
We've got a template for the symbol: it's called {{
music}}:-P. In seriousness, would you really want that CSS being expanded like signatures do every time you type ♭?? On another note, until IE gets fixed (so probably never), "using the symbol itself" is not the easiest way for editors using IE. The {{
music}} template makes it display correctly in rendered articles, but in the edit window, all of the ♭ characters here appear as irritating little squares: I can only tell that the code {{♭}} is supposed to display the flat symbol from context. In a complicated technical passage, IE editors have to keep referring back to the rendered text just to see which squares are which (most of the unicode musical notation, including the sharp sign ♯, does not display). By contrast, the code {{music|b}} is highly intuitive (and fairly professional-looking) in the edit window, is not that much of a hardship to type, and produces a highly professional output; we should be encouraging its widespread use. Perhaps we could add ♭ → {{music|b}} as an AWB typo-fix - I can't think of any possible instance where this could be a false positive except in
flat (music).
Happy‑
melon13:07, 18 June 2008 (UTC)
Note the the first one is in itallics, and it's space is a non-breaking whitespace. if you check the resulting HTML, you'll see that the first one has dropped some </span>s or gained some <spans>, resulting in the 'space=non-breaking' style carrying across the whole line... also, cursor-over the itallic 'B' and you'll see it shares the title that only the 'citation needed' template should have (same reason, the spans are too long).
anyhoo, my investagative bug-pinning-down ends there 'cos i know nothing of how templates work. anyone who knows anything about them, can you figure out wether it's a bug with the template or with wikimedia please.
I've reviewed, and applied, the fix suggested there. It seems to work in all the cases I've looked at so far. Please let me know if it causes any other problems. --
The Anome (
talk)
11:44, 18 June 2008 (UTC)
I've been using the
MediaWiki API for a robot I'm programming, and all is going well -- except that I can't edit pages! The API doesn't recognize action=edit, apparently. The completion of the API was announced several months ago, but per
mw:Talk:API:Edit - Create&Edit pages, it seems that the editing functionality hasn't yet been implemented on en-wp (and other projects?).
I don't mind editing my code to use the traditional index.php?...action=submit, but it seems this is an issue that really needs addressing. Should I go bug the developers via
wmf:Contact us, or is there a good reason for this lack of API implementation? At the very least, I think we should figure out the exact scope of the issue and update the documentation accordingly.
Thanks for the link -- I saw that the MW discussions were very old so I assumed the issue was stagnated, but I guess it's being addressed after all. I suppose that makes this thread unnecessary. —
xDanielxT/C\R08:33, 18 June 2008 (UTC)
"Expires" headers?
I just noticed
this March 2008 posting at O'Reilly Radar, saying "Every time someone revisits [the
the Wikipedia front page, their] browser has to make thirteen HTTP requests to the Wikipedia server to check if these images are still usable, even though these images haven’t changed in over seven months on average. A better way to handle this would be for Wikipedia to put a version number in the image’s URL and change the version number whenever the image changes. Doing this would allow them to tell the browser to cache the image for a year or more (using a far future Expires or Cache-Control header)."
I realize that the page in question is for the Wikipedia project as a whole, not for en.Wikipedia.org, but is this analysis correct with regard to the Wikipedia logo that appears on every en.Wikipedia.org page? -- John Broughton(♫♫)16:30, 18 June 2008 (UTC)
Images are already cached (for at least 30 days I think). Once that expires, the browser quiries the server, then gets told the image hasn't changed so the browser doesn't download the image again. Such queries are cheap and usually done within one keep-alive session. —
Edokter •
Talk • 16:38, 18 June 2008 (UTC)
(ec) There's a tradeoff here: appending version numbers to image URLs would avoid the need for the browser to check for updates to each image, but it would also require
purging the server page cache whenever any image on the page changes. That said, since images do tend to change rather slowly, this might be worth considering. It shouldn't be hard to implement for local images, but making it work properly for images loaded from Commons would be a bit more tricky. —
Ilmari Karonen (
talk)
16:41, 18 June 2008 (UTC)
Also, regarding the O'Reilly article, I think their estimate that "one web server can handle 100 of these requests/second" may be way too low (though I don't have any actual statistics that would let me give even an order-of-magnitude estimate of the real maximum rate). Wikipedia image requests are (AFAIK) handled by
Squid proxies backed by dedicated
lighttpd servers in order to minimize the (real and CPU) time consumed by each request. —
Ilmari Karonen (
talk)
16:47, 18 June 2008 (UTC)
If I were looking for energy inefficiency in Wikimedia's setup, I'd point to the application servers, not the frontend caches. But note that Wikipedia is driving growth in desktop computer usage worldwide, and that uses far more energy than the servers. --
Tim Starling (
talk)
17:20, 18 June 2008 (UTC)