This page contains discussions that have been archived from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.
The PDF comes up for me, but it says at the top "WARNING: Article could not be rendered - ouputting plain text. Potential causes of the problem are: (a) a bug in the pdf-writer software (b) problematic Mediawiki markup (c) table is too wide". Is this the same issue you're getting? /
ƒETCHCOMMS/22:19, 7 January 2011 (UTC)
If I mark a word and drag and drop it into the search box, I get SearchExample instead of Example. Does this happen also to others? --
Leyo16:38, 7 January 2011 (UTC)
I have been frustrated by that on Firefox as well. Is there any workaround, or is this an inherent Firefox issue? I never get the problem with other search bars on FF. /
ƒETCHCOMMS/22:17, 7 January 2011 (UTC)
"Due to high database server lag, changes newer than 245 seconds may not appear in this list." Whoa, what brought on a lag of that magnitude? Ten Pound Hammer,
his otters and a clue-bat • (
Otters want attention)22:07, 8 January 2011 (UTC)
Don't know if it's related, but page retrieval is taking too long: Firefox (3.6.13) often times out, and when it does bring something back, it's usually failed to bring in all the CSS classes, so the page appearance is screwed. Examples include: infobox frameless and showing at upper left instead of upper right; navbox showing without content (as if collapsed and unable to expand); the "This page was last modified on ..." box, normally full-width extreme bottom showing squeezed to the width of the left margin (the strip with the Wikipedia globe at top); browser default font in use. Images also take longer to load. Page saves usually take two or three attempts to go through. This has been happening for about 24 hours. --
Redrose64 (
talk)
14:43, 9 January 2011 (UTC)
Did some change happen overnight in Wikipedia? I use Firefox. When I edit, the Reftools Cite button that allowed me to choose a template has disappeared. Nothing changed on my preferences.
Maile66 (
talk)
14:46, 9 January 2011 (UTC)
I've got the same problem - I also use Firefox. I've tried unchecking and then checking the refTools option in My preferences but this hasn't helped.--obi2canibetalkcontr15:01, 9 January 2011 (UTC)
Currently, MediaWiki
doesn't render wikilinks in site/user CSS/JS pages. Some time ago I created a small script for converting the link syntax [[xx:yy|zz]] to true links on javascript and css pages.
I would like to know if is there any interest in this kind of script being added either to Common.js or a to gadget. Here is the code:
//Workaround for [[bugzilla:10410]]//Convert link syntax [[xx:yy|zz]] to true links on javascript and css pagesfunctioncreateLinks(t){//[[Links like this]]t=t.replace(/\[\[\s*([^\|\]]+?)\s*\]\]/ig,'[[<a href="/info/en/?search=$1">$1</a>]]');//[[Links like this|with an alternative text]]t=t.replace(/\[\[\s*([^\|\]]+?)\s*\|\s*([^\]]+?)\s*\]\]/ig,'[[<a href="/info/en/?search=$1">$2</a>]]');//[[Category:Links|with an index for sorting]]varreCat=newRegExp('\\[\\[<a href="\\/wiki\\/(?:Category|'+wgFormattedNamespaces'14'+'):([^"]+)">([^<]+)<\\/a>\\]\\]','gi');t=t.replace(reCat,'[[<a href="/info/en/?search=Category:$1">'+wgFormattedNamespaces'14'+':$1</a>|$2]]');//Links to MediaWiki site (Workaround for [[bugzilla:22407]])t=t.replace(/href="\/wiki\/mw:/ig,'href="http://www.mediawiki.org/wiki/');returnt;}if((wgNamespaceNumber==2||wgNamespaceNumber==8)&&wgPageName.match(/\.(js|css)$/)&&wgAction=='view'){varpre=document.getElementById('bodyContent').getElementsByTagName('pre')[0];if(pre){varc=getElementsByClassName(pre,'span','coMULTI');for(i=0;i<c.length;i++){ci].innerHTML=createLinks(ci].innerHTML);}c=getElementsByClassName(pre,'span','co1');for(i=0;i<c.length;i++){ci].innerHTML=createLinks(ci].innerHTML);}}}
What do you think?
PS: Maybe someone with better knowledge of jQuery would like to simplify the script before using it.
Helder18:06, 9 January 2011 (UTC)
Seems like a very limited usecase. We shouldn't add JS to Common.js for millions of users for features of little use to the 'common' user. —
TheDJ (
talk •
contribs)
18:41, 9 January 2011 (UTC)
targeting exactly the users who are potentially benefited by that feature? I mean, wouldn't be sufficient to put that if before an importScript, so that only the potentially benefited users will download the script?
Helder19:29, 9 January 2011 (UTC)
If you look at my contributions from just before midnight UTC on January 4, I couldn't type edit summaries. There were other things going on and I don't know why I had so much trouble correcting the template when advising an IP of a contribution that was not productive (edit conflicts were a big problem even though I was the only one editing and I thought I clicked on "preview" rather than "save"). The only other hint (probably making this a Computing reference desk question) is that when I tried to turn the unproductive contribution blue to delete it, the cursor became a dot with an up arrow above it and a down arrow below it. I also couldn't copy and paste because that's what would happen.
Turn it blue? What are you talking about? As before, this is for discussion Wikipedia technical things, not for solving your IT problems. If you think someone needs to be told, tell them yourself - it has nothing to do with Wikipedia. The cursor you describe sounds like the middle button scroll function, so you were probably using the mouse wrongly.
OrangeDog (
τ •
ε)
11:55, 6 January 2011 (UTC)
I wasn't doing anything wrong. The mouse was broken. I was trying to get some advice on what might be wrong. And the problem, whatever it did, DID affect Wikipedia use; therefore it is a legitimate Wikipedia technical problem.
And when I say "turn it blue", that should be obvious. If I want to delete a section of text from Wikipedia, I turn it blue. The mouse is supposed to do that. It did not.18:21, 6 January 2011 (UTC)
Vchimpanzee· talk·contributions·
The usual term for 'turn it blue' which people will recognize is 'highlight'. To produce the cursor you describe (a dot with an arrow above and below), usually one clicks the
scroll wheel on the mouse. This puts the cursor into a scrolling mode, where if you move the cursor to the top or bottom of your screen it will scroll the page of text up or down. This scrolling mode can be turned off with another click of the scroll wheel. Testing under Internet Explorer 8 on Windows Vista, the scroll mode is also turned off by any other mouse click, but I can't confirm what might happen under other Windows versions.
I would speculate that (as this was a public computer subject to less-than-tender use) the scroll-wheel-click is sticking, and it may sometimes be difficult or impossible to turn off the scrolling mode. If the button is stuck and the mouse can't return to its normal mode, you won't be able to click or highlight normally. The problem you were experiencing almost certainly had nothing to do with Wikipedia, but was rather a hardware problem on the public computer.
TenOfAllTrades(
talk)
18:40, 6 January 2011 (UTC)
Right, but we have a symptom of a problem which is actually affecting the ability to use Wikipedia. At least in a search for what might be causing such a problem, this section might come up.
Yes, but it seems to be a problem that's easily fixed — replace the defective hardware. There isn't anything that needs to be done on Wikipedia's end.
TenOfAllTrades(
talk)
02:32, 7 January 2011 (UTC)
I would suggest that you were unable to select the edit summary and search boxes, if your mouse was stuck in autoscroll mode. RichFarmbrough,
19:20, 7 January 2011 (UTC).
If someone pours coffee on their keyboard they would also find it difficult to edit Wikipedia. That doesn't mean that it should concern us in any way.
OrangeDog (
τ •
ε)
17:57, 10 January 2011 (UTC)
Nowiki-slash for fewer if-logic worries
I have found, now, what other people have known about the nowiki-slash tag ("<nowiki/>"): it preserves spaces or "#" (or ":" or "*" or ";") to deter the newline bug. I finally joined the resistance movement of "Wikipedians worrying about performance, now, for carefree wikiwork later" (anti-
WP:PERF), and discovered many people have already solved massive performance problems in the
MediaWiki software. Now, any if-logic can be written as super-robust, without worries about color hex codes (such as: #f8eaba) becoming auto-indented, numbered lines:
I recently also created more utility templates,
Template:Xifeq,
Template:Xifnoteq and
Template:Xifexpr as quick if-logic functions to bypass whatever glitches in #ifeq & #ifexpr. So, if any other bugs are introduced into future changes to the parser functions, then those if-logic templates can be adjusted to still be reliable (hopefully).
Meanwhile, the "<nowiki/>" tag can CANNOT be used everywhere, because it is highly unlikely the MediaWiki software will screw it up as well. I use had used it to separate nested braces "{{{{{}}}}}" as in "{{val|{{{x|{{{x2|54123}}}<nowiki/>}}}<nowiki/>}}" without affecting the results, but calculations were ruined by nowiki tags. Also, it can truncate wikilinks: [[common sense|common-sense]]<nowiki/>less, as "
common-senseless". Anyway, nowiki-slash is just one of many semi-great ideas discovered by people who have been worrying about performance and finding great ways to make things much easier. More ideas later. -
Wikid7707:04, 6 January 2011 (UTC), revised 14:34, 6 January 2011 (UTC)
Thanks. I never would have expected nowiki tags generate internal hidden characters, which do affect the results:
"{{padleft:|1|<nowiki/> <nowiki/>}}" → " "
"{{padleft:|1|<nowiki/>abc<nowiki/>}}" → "a"
"{{padleft:|1|<b/> <b/>}}" → "<"
"{{#expr: 3*<nowiki/>10<nowiki/>}}" → "Expression error: Unrecognized punctuation character ""."
"{{#expr: 3*<nowiki></nowiki>10}}" → "Expression error: Unrecognized punctuation character ""."
Just when I thought they wouldn't screw up nowiki tags, they had already done so, long ago: embedding invalid characters in calculations. Kids don't do this: Don't ever write a twisted, convoluted markup language like MediaWiki or its twisted preprocessor. The most logical assumption was that nowiki tags simply stopped the expansion of the text, not that they embedded special characters which are left inside to kill calculations or cause comparisons to mismatch and fail. Those results are just more bugs as to how it should have worked, logically. The nowiki tags should only be used in limited cases, not in calculations, at the present time. -
Wikid7714:34, 6 January 2011 (UTC)
More like "kids, don't play with things you don't understand". The way the preprocessor, and the parser, work is to chop out bits of wikitext that they've finished processing, and replace them with strip markers: {{padleft:|40|<nowiki/> <nowiki/>}} produces , which you can see is most of a complete strip marker. The <nowiki>...</nowiki> block is seen by the preprocessor, and snipped out and replaced by a strip marker; the control character you see at the start of the marker (there's one at the end as well) is there precisely to be impossible to generate in actual text; it is impossible to create a 'fake' strip marker in raw wikitext. The contents of the block are processed (escaped in this case to prevent any further parsing) and stored separately in the parser output. A <ref>...</ref> or <source>...</source> block would be treated in exactly the same way: {{padleft:|37|<ref>Foo</ref>}} produces . At the end of the preprocessor stage, all the strip markers are replaced with their contents, and all is well. Except here, where you have horribly abused a bug in padleft (a bug deliberately left unfixed, I might add, precisely in order to allow string manipulation) to cut off bits of the strip marker, meaning it is not recognised as complete and is not replaced with its processed contents. I can't show you what the end of the strip markers looks like because if you leave the whole marker, it is correctly replaced by its contents. {{padleft:|140|<ref>Foo</ref>}} produces : three complete copies of the strip marker are replaced by their processed contents, and one half marker is exposed in the output.
Happy‑
melon17:00, 6 January 2011 (UTC)
Thank you for describing the complex internal processing of the inserted markers. I will use that to explain to the "kids" so they can better understand the issues. -
Wikid7720:09, 7 January 2011 (UTC)
Other advice would be: "Don't expect a markup language to make a good programming language". Markup languages are designed for presentation; calculations and comparisons are secondary features. Mr.Z-man23:14, 6 January 2011 (UTC)
Well, conditional text formatting and typesetting is based on comparing text and making calculations, such as counting lines-per-page or incrementing header numbers in a Table of Contents. I don't know what else to say, except perhaps: "Computers: they're not just for reading email any more". The markup language should allow defining over 1,000
macros, nested hundreds deep, plus defining local variables (including line, word or page counters). Those features can be provided by a relatively simple one-pass preprocessor. The expansion limit of 40 nested templates is an unfortunate choice for the current complex template processing. I would raise to 60 (or perhaps 100). Regardless, it should be easy to allow a simple macro scripting language which defines local variables: we know templates consider "{{{x|25.0}}}" to be local when "x" defaults, but MediaWiki cannot retain that value of x as "25.0" for use later, only when passed into a template. I am not advocating we require macros to have
strong typing,
exception handlers, or
concurrent macro processing. We just need simple math, string operations, reliable if-logic comparisons, and local variables to hold the results to display on a page. -
Wikid7720:13, 7 January 2011 (UTC)
Need parser function to set a parameter: Since templates do not yet have "
local variables", but have parameters (set when a template is invoked), perhaps a new parser function {{#set: x|25.0}} could be created to simply insert "25.0" into parameter {{{x}}} as if it had been passed into the template. -
Wikid7714:31, 10 January 2011 (UTC)
Any plans to upgrade to simple macro scripting
There has been talk of changing to a more capable, simple, mainstream
macro scripting language to allow quick, easy typesetting macros which work reliably. The current markup language is so error-prone and limited that users have written many extremely complex templates to perform the simplest of tasks, in an effort to avoid MediaWiki or parser-function quirks. Some obvious questions:
Are there essays about upgrading to a quick scripting language?
Is
Javascript or some other browser language a better option?
Could the other language be triggered by a top-line switch?
Could conversion of parser-function templates be automated easily?
How many times faster could a macro scripting language format articles?
Currently, despite the ongoing problems, people have coded 4 million or more cases of using "home-made" string templates to support typical macro-style typesetting routines. There is obviously a major need to move to modern technology, so I am wondering what plans are being made to move forward, and which pages describe the alternatives. -
Wikid7719:59, 6 January 2011 (UTC)
There is no fixed plan atm. The holdup is the lack of an experienced developer that has enough skill, free time and a thick enough skin to actually take on such a huge project. Most volunteer devs are only casual contributors doing patch work and other small incremental work. Paid devs have plenty of other work. LUA is the most often suggested scripting option as far as I know. —
TheDJ (
talk •
contribs)
21:20, 6 January 2011 (UTC)
There was an objection to using things like Lua because they create a dependency on an executable or a PHP extension that would need to be compiled. This would make Wikipedia content unusable on some crappy shared hosting service where users don't have shell access or PHP is restricted from running other programs. So any scripting language would need to be implemented in pure PHP, even though we wouldn't use that implementation. Personally, I think that's stupid. But, its unclear whether that restriction is still in effect, or who would make the decision on it. Mr.Z-man23:21, 6 January 2011 (UTC)
Need parser function to set parameters: Meanwhile, since templates do not yet have "
local variables", but have parameters (set with values when a template is invoked), perhaps a new parser function {{#set: x|25.0}} could be created to simply insert "25.0" into parameter {{{x}}} as if it had been passed into the template. That would allow storing partial calculations, or setting
status variables, such as {{#set: gotmatch|true}} to avoid all the duplication in the current template markup coding. -
Wikid7714:37, 10 January 2011 (UTC)
Hi, normally I don't talk here much but today in fiwiki we encountered quite frustrating situation with
Google Sitelinks:
If Finnish person search Wikipedia from Google our sitelinks are "Pussy, Sex, Talking Machine, Adolf Hitler, Justin Bieber, Sodomia..." (you get the point, not very flattering topics). If person search Wikipedia using Deutch sitelinks are "Science, Technology, History, Arts and Culture..." (excellent result!). If user search Wikipedia using English instead of sitelinks Google shows Wikipedia search.
I tried to look
http://de.wikipedia.org source to locate possible place where these sitelinks could be manipulated, without luck. Because English Wikipedia has largest user base I hope that here are some SEO experts who could help fiwiki to get rid of inappropriate sitelinks. --
Agony (
talk)
20:55, 9 January 2011 (UTC)
No, I don't see any reason to believe this is a virus, just Google giving poor site links (as mentioned in their article it's an automated process, it's possible someone is abusing it somehow to purposely make it give those links) (
link to Google search). Seems to be Google's problem. -
Kingpin13 (
talk)
21:58, 9 January 2011 (UTC)
I did wonder about that as well =p. All the topics are fairly high interest, "Sodomia" and "Khagendra Thapa Magar" seem the oddest, ranking just 807 and 3884 in traffic (so not necessarily reflective of people's search terms, see
here too (which isn't necessarily reflective of what people are searching for on Wikipedia), with
this seeming to be the most interesting term, shot up in volume just recently, with news reference volume staying very similar. Of course that could be unrelated to Wikipedia. Also check out the regions, looks suspicious? (or not - language)), while the rest all rank higher than 100. But it seems that sitelinks are based more on the architecture of the site itself. Of course, Google are rather secretive about how they actually get them.. -
Kingpin13 (
talk)
22:24, 9 January 2011 (UTC)
For the past couple of months, Luxo's, Soxred's edit counters and Yet another edit counter, are not working for unicode user names. For example -
this user's soxred page comes up
like this. It used to work earlier (i think i last checked in november 2010). Is there a way to get around this?--
Sodabottle (
talk)
16:05, 10 January 2011 (UTC)
User:VolkovBot continues to add
it:Fistula acquaria to
Roman lead pipe inscription (and the German and French articles), even though the Italian article has nothing to do with epigraphy. I asked the operator of VolkovBot to stop his bot from adding the article - but his workaround did not work. Even removing simultaneously all language versions, as proposed, does not help, since some dude on the French WP
decided otherwise.
I am fed up by now combating a rampant bot and I don't want a single user in another language version to decide on the interwiki links for all language versions (which effectictively is the case, since a single addition, as that by the French user, triggers the bot to add the Italian link to all language versions). It cannot be the case that human users are thrown into the battle against a stupid and unrelentent bot. Either the bot gets it or it should get its plug pulled.
Gun Powder Ma (
talk)
20:24, 6 January 2011 (UTC)
Also worth having a go at just adding {{nobots}} or {{
bots|deny=VolkovBot}}, since it uses pywikipedia it may be complaint with this (depending on options) -
Kingpin13 (
talk)
22:58, 6 January 2011 (UTC)
Nobots should be a temporary fix only. There are a few cases (Monocieous/Monecieous springs to mind) where it has been hard work getting the interwikis sorted "permanently" but in my opinion it is worth it to avoid having a growing number of articles with crufty bot-directive templates on them. RichFarmbrough,
16:16, 7 January 2011 (UTC).
Could someone who understands safesubst: please rewrite that section of
Help:Substitution? It is utterly incomprehensible at the moment. It does not make it clear what safesubst: is even for, let alone how it works or how to use it. — This, that, and the other (talk)00:55, 30 December 2010 (UTC)
I understand it, but I'm not sure how to rewrite it to make it clear. I can try to explain it to you, maybe between us we can rewrite it.
The problem is in making a template that works correctly both when transcluded and when substituted. Consider a hypothetical template {{hello world}} that should display "Hello, world!" if no parameter is given, and "Hello, world+name!" if a parameter is given.
One way to write that is "Hello, world{{#if:{{{1|}}}|+{{{1}}}}}!". It will work correctly when transcluded, and when substed it will look correct but the page's wikitext will contain "Hello, world{{#if:|+{{{1}}}}}!". We don't want that parser function showing up in the wikitext!
We can't just throw "subst:" in the template code, becuase that would subst it as soon as we saved the template. We have to hide that subst until the template is actually used, e.g. with <includeonly>.
We can get the substed text to be correct by changing the template to "Hello, world{{<includeonly>subst:</includeonly>#if:{{{1|}}}|+{{{1}}}}}!", but then a transclusion displays "Hello, world{{subst:#if:|+{{{1}}}}}!" in the page!
The old solution was to make the template be "Hello, world{{{{{subst|}}}#if:{{{1|}}}|+{{{1}}}}}!". This works right on transclusion, and if you pass a parameter |subst=subst: when substuting it works right then too. But that's not really good, because then people have to remember to specify |subst=subst: or it gives the same ugly wikitext as in #1.
We can't change the behavior of "<includeonly>subst:</includeonly>" to work like we want it to, because that would break other things. So "safesubst:" was created to be like subst except that it doesn't screw things up when used like in #3. So now "Hello, world{{<includeonly>safesubst:</includeonly>#if:{{{1|}}}|+{{{1}}}}}!" works just like we want, correct on transclusion and leaving the page wikitext as "Hello, world!" when substed.
Another slightly shorter way to do basically the same thing as "<includeonly>safesubst:</includeonly>" is "{{{|safesubst:}}}", i.e. making "safesubst:" the default value of the parameter with an empty name. Of course, that does allow someone to do {{subst:hello world|=}} to give wikitext like in #1 or {{hello world|=garbage}} to really screw things up, but it's unlikely anyone would do that.
I'm afraid not... For me to understand it, I must first have a clear concept of what safesubst: exactly does in it's simplest inception, as that is the core piece of information that is missing form the documentation. For example, it is completely unclear what this code does: safesubst:{{{1}}} during both transclusion and substitution. Once that becomes clear, I would have no problem undersatanding its more complex incarnations. —
Edokter •
Talk — 17:51, 8 January 2011 (UTC)
As Kotniski already noted below, safesubst and subst do the same thing except that {{<includeonly>subst:</includeonly>foo}} when transcluded will display "{{subst:foo}}" while {{<includeonly>safesubst:</includeonly>foo}} will in turn transclude
Template:foo.
Anomie⚔19:24, 8 January 2011 (UTC)
I've just been trying to tidy up other parts of
Help:Substitution, and chucking out a lot of fairly meaningless or useless text, but there's still a long way to go (expert help would be appreciated). I think if the whole of the page were to be written to make sense, then the safesubst thing would fit into it quite logically. But first you have to understand the problem which it's trying to solve, which is kind of described elsewhere on the page, but not very clearly at present. (AIUI safesubst does exactly the same thing as subst, except in the kind of situation Anomie describes above - I'm guessing that safesubst causes substitutions to be performed after transclusion, while ordinary subst can only happen before transclusions.)--
Kotniski (
talk)
17:59, 8 January 2011 (UTC)
If I'm understanding
the code correctly, it seems that subst and safesubst both allow template replacement during the pre-save transform (PST) phase (i.e. substitution), but safesubst additionally allows template replacement during the parsing phase (i.e. transclusion) where subst doesn't.
Anomie⚔19:24, 8 January 2011 (UTC)
Here's a testcase that I knocked up based on the above description:
User:This, that and the other/subst and safesubst. Notice that only the final bullet point under each heading (safesubst:) produces the correct output every time (i.e. when transcluded AND when substituted). I think the main problem with safesubst: is its rather opaque name. It would have been better to call it substonsubst: or something. Still, we're stuck with it as is. — This, that, and the other (talk)09:36, 9 January 2011 (UTC)
Well you have explained the function of safesubst: to me well enough, I have looked at that documentation page a number of times, and improved what I could, but it never explained the salient point of safesubst: so I could never improve the wording, it's good to see it getting done properly.
I would like more detail (maybe a footnote) on "We can't change the behavior of "<includeonly>subst:</includeonly>" to work like we want it to, because that would break other things." I believe it would be worth a fair amount of breakage to do away with the subst:/safesubst: distinction. RichFarmbrough,
18:48, 9 January 2011 (UTC).
I've done a major revamp of the page
Help:Substitution, including (I hope) a better explanation of safesubst. Experts on the subject are invited to take a look, check I haven't made any major errors or discarded anything important (there's still a link to the Meta page which contains the more extensive text). Non-experts are invited to see if they now understand it any better (and what still needs to be made clearer).--
Kotniski (
talk)
12:46, 10 January 2011 (UTC)
I'd like some mention of the importance of placing the subst: immediately before the template name, without intervening spaces or newlines. I've carried out some tests, and found that there are some fairly simple rules, but these rules vary depending upon whether it's templates, or variables/parser functions. These rules are:
A space or a newline is permitted between the {{ and the subst:
For templates, a space is permitted after the subst:
For variables and parser functions, there cannot be a space after the subst:
Newlines are not permitted after the subst: in any situation.
You might also want to mention the {{{|safesubst:}}} idiom. I know {{{subst1|subst:}}} is mentioned, but {{{|safesubst:}}} isn't obviously the same deal.
Anomie⚔14:46, 10 January 2011 (UTC)
Thanks, I'll add those two things to the page. I'm also planning on doing some off-and-on tidying work on
Help:Template, which seems to be another hugely overgrown page (though at least it's a bit more understandable than
Help:Substitution was). --
Kotniski (
talk)
17:01, 10 January 2011 (UTC)
I can actually understand it now! (Although I'm a programming enthusiast, so I can't speak from the perspective of those not burdened with extra technical knowledge.) Thanks. — This, that, and the other (talk)06:24, 12 January 2011 (UTC)
Reverting
I don't know if it's Twinkle, Chrome or what, but if I go to history and click on the latest version and say 3 versions back to compare, I don't get the 'Restore this version' in Chrome but I do in Firefox. I've also lost Twinkle's rollback. This seems erratic as sometimes I don't have the problem.
Dougweller (
talk)
19:07, 9 January 2011 (UTC)
Is it feasible to introduce two new parameters into the Ogghandler? They would be to view/listen to a chosen portion of the original; something like:
|starttime = 0:00:00
|endtime = 0:00:00
I think this could reduce the amount of space used up by derivatives of larger files with points of interest at a particular portion (I think an example of what I mean is
here, and the movie files really take up a lot of space, although I cannot verify at the time) This could also have other applications for verifying particular parts of the media without getting lost in the rest. I imagine the specified parameters to either point to the positions (as in having the original file load) or truncate the selected part; either way I think this is a capability that would be beneficial to a lot of projects. -
Theornamentalist (
talk)
17:21, 10 January 2011 (UTC)
Thanks DJ, I did not know about that tool. I am curious though about building a parameter to point to a start and an end within a clip. That tool saves the file separately, which takes up a lot of space. Is messing around with the script something someone is willing to do? I hate coming in here and asking, yet never being able to return the favor directly. -
Theornamentalist (
talk)
18:58, 10 January 2011 (UTC)
I am attempting to create a book but when i download it as a PDF it says that the file is corrupt and cannot be retreived and when i download it as an ODF it downloads a blank document
Admiralalexmann (
talk)
19:55, 10 January 2011 (UTC)
Can someone take a look at
this? For some reason, there is a problem creeping up between endnotes #66 and #67. I cannot figure out what the problem is, much less how to solve it. All I know is, the numbers are being thrown off, leading to further problems in other parts of the page, as references that do not match examples have been removed as seemingly irrelevant. Does anyone have any idea what is going on here? Thanks for your time. ---RepublicanJacobiteThe'FortyFive'21:24, 11 January 2011 (UTC)
Er, what problem? Which browser and OS are you using? Works for me Windows XP with IE 7.0.5730.13, Firefox 3.6, Google Chrome 8.0.552.224, Opera 11.00 --
Redrose64 (
talk)
21:37, 11 January 2011 (UTC)
Does Wikipedia have an IP recording utility I can use?
Hello, I'm a journeyman level editor and I edit by fits and starts depending on when the spirit moves me so I haven't really advanced in skill level. I have been editing primarily from the secure site for a number of months. It appears that the login program used sometimes kicks me out so that I will do an edit thinking I'm still logged in but then find that the system has recorded it as an anonymous edit. When this happens I like to go to the Talk User:IP number talk page and leave contact info. But my ISP uses dynamic IPs so sometimes I lose track. The same happens when I'm at work and can't log in and do an edit (very rarely nowadays since my company changed servers).
So my hence my question: is there any easy to use utility program hidden in the bowels of Wikipedia which I can use to track my lost IP edits and go back and leave contact info on the talk pages?
Trilobitealive (
talk)
15:26, 8 January 2011 (UTC)
Thanks. That was what I was afraid of. I don't want to apply to become an admin or to get more user rights, (I still don't know what to do with Reviewer rights for instance) I just want to try to keep up with my own edits. Will look at the link you gave on the logged out issue.
Trilobitealive (
talk)
16:45, 8 January 2011 (UTC)
Let's face it, the poor man's checkuser isn't very helpful. It bases some (all?) of its data on edits like
these, so it's only useful for careless users like our friend Jimbo :) — This, that, and the other (talk)10:10, 9 January 2011 (UTC)
If you are looking for anonymous edits by you, you don't need checkuser to find these edits. You just need to figure out what IPs you'd have used during those eras (which Checkuser couldn't find anyway). Generally ISPs have a limited IP subset range (depending on ISP and market). It can range from (generally) a pool of about 256 IPs to around half a million (for something like Comcast in Los Angeles, or an Italian ISP). If you repeatedly hit
a place like this while rebooting your modem, or check a website that keeps historical data on you (gmail keeps a bit), you can discern the range to some degree, and then use a Wikipedia scanner (or the CIDR gadget in
Special:Preferences) to find them. --
Splarka (
rant)
22:34, 9 January 2011 (UTC)
Never mind the CIDR gadget. I can't even READ the
Classless Inter-Domain Routing page. The last hardware program I wrote was for a 300 baud modem so I'm a bit dated in my computer skills. I'll just have to hit up the whatismyip.com site, which is what I've already been doing when I have the time. Trouble is, they expect me to actually work when I'm at work so that cuts down the time I can edit Wikipedia. Maybe that is a good thing.
Trilobitealive (
talk)
01:10, 10 January 2011 (UTC)
Splarka's suggestion actually isn't too hard. In my (limited) knowledge, I usually just replace the last placeholder with an asterisk (on
Special:Contributions) and it finds all the contribs within that IP's range (at least, I think so).
Killiondude (
talk)
08:35, 10 January 2011 (UTC)
Next time, ensure login: Trilobitealive, before you login again, click on "login for 30 days" BEFORE entering your username, to remain all day or a month. If you cannot risk 30-day login, then logout after several hours, and re-login to avoid a login timeout which stores your IP address. Also, when you edit a page, put a 4-tilde (~) live username signature at the TOP OF THE PAGE you are editing: that way, during edit-preview, you can see if you get logged out after being away several hours (or days). If you forget to remove the 4-tilde (~ ~ ~ ~) signature in an article, don't worry: someone else will remove it for you, soon enough (tolerance for Wikipedia's 2-month and 6-month vandalism hacks has made people tolerant of even simple mistakes). It is more important for you to keep edits attributed to your username than leaving a few signatures in articles, where most people really just skip over that when they are reading. It is not unusual for an article to be viewed "27,000" times before people fix spelling or grammar. Most people come to read topic details, regardless of typos. Anyway, please try to use your login username when you want to keep track. -
Wikid7722:15, 11 January 2011 (UTC)
I installed
Gadget850's green-save-button thing on 4 March 2010. I no longer notice that it's green; but I do notice when it's grey. This means I've become logged out; and when this happens, I start Wikipedia on another browser tab, log in there, return to my first tab, click "Show preview" and the save button turns green again, so I can safely save with the edit attributed to my login. --
Redrose64 (
talk)
22:26, 11 January 2011 (UTC)
For the past week or two I haven't been getting a "You have new messages" banner when my talkpage is edited. I don't recall editing my preferences or anything, but who knows. Is there any setting in preferences that controls the talkpage banner? (I'm sure there is stuff like that in personal CSS as well, but I definitely have not edited my monobook.css or anything like that.) I am currently in China, where Wikipedia (and the internet in general) gets wacky sometimes, so I don't know if that might be the problem. rʨanaɢ (
talk)
23:02, 11 January 2011 (UTC)
Solved, I simply removed the {| and |} from Template:Items then made the header template open and had : |} at bottom of all templates.
Adamtheclown (
talk)
18:29, 12 January 2011 (UTC)
How can we implement pending changes on talk pages? Mine is continually vandalized by a sock IP, so I don't want them removing stuff, but I do want legit IPs to be able to leave queries. CTJF83chat20:23, 12 January 2011 (UTC)
My user page is protected all the time. My talk page, however, where the harassment occurs is not always protected (I prefer it to remain unprotected) so the harasser can freely vandalize my talk page. That's why I want pending changes on talk pages, so he (
User:Brucejenner) can't remove stuff from my talk page. Oh, and as for blocking, he has well over
100 blocked sockpuppets, and apparently won't stop creating them to vandalize/harass LGBT pages/users. CTJF83chat21:50, 12 January 2011 (UTC)
User talk pages are not protected unless they are very heavily vandalized (i.e. > once per 5 minutes) and even then that is only temporary.
Prodegotalk00:03, 13 January 2011 (UTC)
Somebody kindly remind me how I use a URL like
http://www.google.co.uk/search?q="David+Allen+Green"&ie=utf-8&oe=utf-8&aq=t&rls=org.mozilla:en-GB:official&client=firefox-a#q="David+Allen+Green"&oe=utf-8&rls=org.mozilla:en-GB:official&client=firefox-a&um=1&ie=UTF-8&tbo=u&tbs=nws:1&source=og&sa=N&hl=en&tab=wn&fp=acdf426d122b1dae without it breaking? (It's for a talk page) Andy Mabbett (User:Pigsonthewing);
Andy's talk;
Andy's edits12:51, 13 January 2011 (UTC)
Perhaps, but that's not what's needed: http%3A%2F%2Fwww.google.co.uk%2Fsearch%3Fq%3D%22David%2BAllen%2BGreen%22%26ie%3Dutf-8%26oe%3Dutf-8%26aq%3Dt%26rls%3Dorg.mozilla%3Aen-GB%3Aofficial%26client%3Dfirefox-a%23q%3D%22David%2BAllen%2BGreen%22%26oe%3Dutf-8%26rls%3Dorg.mozilla%3Aen-GB%3Aofficial%26client%3Dfirefox-a%26um%3D1%26ie%3DUTF-8%26tbo%3Du%26tbs%3Dnws%3A1%26source%3Dog%26sa%3DN%26hl%3Den%26tab%3Dwn%26fp%3Dacdf426d122b1dae Andy Mabbett (User:Pigsonthewing);
Andy's talk;
Andy's edits15:41, 13 January 2011 (UTC)
I use
this page which also explains it a bit. If you put a URL in the box there, check 'URL-Safe Encoding' and then the 'covert' button that does similar. Chzz ► 01:28, 14 January 2011 (UTC)
Reduce the size of non-edited portions when showing changes
Forgive me if this has been asked before, I couldn't find any record of it. Can we reduce the size of non-edited portions when showing changes?
If you select 'Show changes', the edited text is shown. Either side of the edited text is non-edited text. For an example of what I mean see
this edit.
Sometimes the non-edited text is so large that it makes it difficult to see edited text above and below. This is particularly a problem if you do multiple small edits throughout an article. You then have to scroll up and down to see the changes. I was told that this was an API issue but I don't really know who has control of that. How do we reduce the number of visible non-edited words/characters?
Lightmouse (
talk)
19:59, 1 January 2011 (UTC)
In general, articles should have smaller paragraphs, so that less text is show in the diff before-versus-after blocks. A trick I have used is to save as 2 separate edits: the first edit adds line breaks or adjusts the blank lines (most people don't care about extra line breaks); then the 2nd edit has the real changes to see during a diff. Decades ago, people realized how the size of the multi-line diff-bracket affects the comparison to re-sync lines, so that diff-bracket needs to be a user-specified option, especially in lists where the data is similar, and the diff-bracket is the wrong size so that old lines, which should be considered unchanged, are instead shown as part of a larger difference. There are so many simple changes which could be made to simplify double-checking of edits, but perhaps copy 2 revisions of a page to your PC and use diff-tools to compare versions in your disk files (and count the number of changes, etc.). -
Wikid7720:02, 2 January 2011 (UTC)
Perhaps some <tr> class to identify rows which contain gray cells (td.diff-context) and facilitate hiding them in one’s monobook.css. That would make the line-number headings unreliable however, so you probably would want to suppress those as well. ―
cobaltcigs16:48, 4 January 2011 (UTC)
The default diff-bracket seems larger than I remembered: it is displaying TWO unchanged lines above and below the changed line, where a blank line can count as one of the unchanged lines. I made 2 edits to "
Boulder, Colorado" while aligning the infobox keywords, and the 2nd edit forced a line break to trigger a diff; note how the differing text is surrounded by 2 unchanged lines (paragraphs) above and 2 lines below:
Boulder-diff-link]. An option for "set unchanged-text size" would be a good option to have under
Special:Preferences (as opposed to some of those other rather useless options). A long unchanged line could be shown truncated as "<...>" beyond the unchanged-text size. I don't know who to contact to help sort out their priorities as to what features should be preferences. I'm still in shock about the 40-depth expansion limit in the 21st century! -
Wikid7720:29, 11 January 2011 (UTC)
Is there a good way to find pairs of articles "X people" and "X language", to verify that "X" is a dab page rather than a rd to just one? It's likely that many won't be categorized. Thanks —
kwami (
talk)
00:31, 12 January 2011 (UTC)
The following is a list of all “X” where articles “X people” and “X language” exist, neither is a redirect and “X” is not a disambiguation page, generated using a query on the toolserver.
Svick (
talk)
I don't believe I made any comment on what you requested; I merely clarified that deleted articles are indeed treated no differently to extant ones; all edit screens are noindexed, whether or not they are deleted, and no view pages are, whether or not they are deleted. I generally agree with you that they should be.
Happy‑
melon10:33, 13 January 2011 (UTC)
Thank you for bringing this up. Basically, the issue is "Wikipedia sockpuppets of x" categories, where a user had their real name as their Wikipedia name and now when someone (presumably employers) do a google search of them, the Wikipedia page comes up. The linked discussion above is the second such request that has been made, the first of which I noticed when going through empty categories at one point, and the second having been this one when someone e-mailed me about it. Even though it's a deleted page, it would be nice to have something there preventing indexing.
VegaDark (
talk)
01:46, 13 January 2011 (UTC)
The issue is far wider than that. It is a general one of trying to get Google to forget about deleted articles. Never mind one person fussing over having been accused of sock puppetry. I am thinking about the hundreds of spam articles we delete every day. —
RHaworth(
talk·contribs)01:56, 13 January 2011 (UTC)
Filing this in bugzilla, if there isn't already a bug for it, would be a good first step towards getting this noticed and fixed. I would expect it to be a very simple fix.
Happy‑
melon23:10, 13 January 2011 (UTC)
HTTP 404
Cracked it - I hope! In the (article) namespace, deleted or never-existed titles return an
HTTP 404 Not Found code. This a stronger signal to a search engine than noindex,nofollow. But in the Category: namespace all pages return 200 OK. (I have not checked any other namespaces.) I assume the idea (misguided in my view) is to support cases where articles have been put in a certain category but no category page has been created. I have raised
this ticket at Bugzilla proposing that all deleted or never-existed categories should return a 404 code. —
RHaworth(
talk·contribs)18:32, 14 January 2011 (UTC)
Not quite correct. Any category that has ever had any pages in it will return 200, even if it is now empty; a category page for a category that has never had any pages in it will return 404 under the same conditions as pages in "normal" namespaces.
Anomie⚔23:45, 14 January 2011 (UTC)
can't wikilink within <ref>?
After revising the text of the first <ref> that appears on
Cult of the Holy Spirit, it seems there's some snafu with the wikilink I inserted. Content within [[square brackets]] appears as plain text instead of a link, but for the life of me I can't figure out why... Any thoughts? —Preceding
unsigned comment added by
67.246.119.98 (
talk)
11:25, 14 January 2011 (UTC)
Linking works, but you're not allowed to use the
WP:pipe trick within refs, for some reason (this is no doubt one of those bugs that we've been waiting for years to get fixed). --
Kotniski (
talk)
11:34, 14 January 2011 (UTC)
The pre-save-transform parser is not applied to the contents of tags; subst and signatures don't work either for the same reason. It is indeed the longstanding
T4700, which has just seen in its sixth New Year.
Happy‑
melon12:06, 14 January 2011 (UTC)
I have a question about the page limit for deletions. I've several times noticed that user talk pages with under 5,000 edits—often considerably under—display the "can't be deleted" message, supposedly because they have over 5,000 edits. I just noticed this with
User talk:Mbz1, which has suddenly started displaying the message,
[3] though it has 2,700 edits.
[4] Does anyone know how these messages are generated, and why pages with just a couple of thousand edits are displaying it?
SlimVirgintalk|contribs22:15, 12 January 2011 (UTC)
I believe it's an estimate of db row count, rather than an exact value (which could be an expensive query in its own right for enormous pages), so it's more order-of-magnitude than a precision measurement.
Happy‑
melon22:31, 12 January 2011 (UTC)
Thanks. I'm wondering what would cause that message to appear and disappear. That same talk page, five minutes beforehand, did not display the over-5000-message. Then suddenly it did. And then later it didn't again (or, rather, the archive that page's edits were moved to didn't). That's why I'm wondering whether these messages are automatically generated, and what causes the change. I've noticed the same issue with other user talk pages in the past.
SlimVirgintalk|contribs22:55, 13 January 2011 (UTC)
Page moves cause issues. I believe you can delete any page by moving it before deleting it. Or perhaps you just have to move it enough.
Prodegotalk01:30, 14 January 2011 (UTC)
You can't delete it either; not while it's the main page, anyway... ;D And don't you remember that Drama
™ when someone history-merged 4,000 null revisisions into the main page in order to bring it over the threshold? Tim manually purged them from the database, IIRC...
Happy‑
melon11:45, 15 January 2011 (UTC)
So does anyone know how these messages are generated, i.e. are they always automatically generated, or can they be added manually? The reason I'm wondering is because they appear then disappear again.
SlimVirgintalk|contribs11:27, 15 January 2011 (UTC)
They're automatic, it's tested when you try to load the delete form. The method used to estimate the revision count is somewhat inaccurate, and in a database which is changing as rapidly as wikipedia's, it can fluctuate rapidly. The data will also be being selected from a randomly-chosen one of enwiki's five slave databases rather than the master, each of which will be lagging a slightly different length of time behind the master database, which can explain instantaneous variations.
Happy‑
melon11:45, 15 January 2011 (UTC)
banner to editor that there's a message via article or edit page
A registered editor recently began contributions and the list is about 14 minutes long. Some, at least, were inappropriate and a caution appeared on the user's talk page, followed by a more specific caution, followed by a warning, followed by a block, which was indefinite. From the first message to the block took 7 minutes. In my experience, users get to see banners about messages for them on the login acknowledgment page (I think that's where I've seen mine). So I wonder if this user would have known about the cautions, warning, and block until they tried to edit again and found they were blocked without an apparent prior problem. Does anyone know? For example, would the you-have-a-message banner have shown up on an article page or an edit page during the user's session?
Nick Levinson (
talk)
05:12, 15 January 2011 (UTC)
When I load the edit page, the font looks smaller for a fraction of a second, and the left menu bar looks different. I don't know if this is just an issue with the Vector skin. Here's a crop of a screen cap of what I see for a fraction of a second on the Black Swan page:
link. Compare to the edit screen that you see there when you click edit.
Doc StrangeMailboxLogbook04:46, 16 January 2011 (UTC)
It "flashes" because that's when the browser loads the JavaScript, which applies formatting changes to the sidebar and edit box. It is especially noticeable when the page is slow to load. Gary King(
talk ·
scripts)05:01, 16 January 2011 (UTC)
If you have multiple tabs open and change the "zoom" on one of them, the JS takes a bit to catch up with the resolution. That's when I really notice this issue.
Skier Dude (
talk)
05:05, 16 January 2011 (UTC)
I never zoom on anything, but I did see this on another computer which I had never logged in on before (and indeed hadn't, I clicked edit as an anon). So its Javascript and others notice it? That's all I needed to know. The edit screen thing isn't happening right now after I restarted Firefox, but I might see it again. Good to know its nothing to worry about if I do.
Doc StrangeMailboxLogbook05:15, 16 January 2011 (UTC)
Mobile Wikipedia error
Internal Server Error
The server encountered an internal error or misconfiguration and was unable to complete your request.
Please contact the server administrator, [no address given] and inform them of the time the error occurred, and anything you might have done that may have caused the error.
More information about this error may be available in the server error log.
Additionally, a 500 Internal Server Error error was encountered while trying to use an ErrorDocument to handle the request.
Apache/2.2.14 (Ubuntu) Server at en.m.wikipedia.org Port 80
Comment
I get the above error on mobile version of site. Error seems permanent. Regards,
SunCreator(
talk) 02:17, 14 January 2011 (UTC)
I am able to access the site through the secure.wikimedia.org (that's how I can post here) but no access apparently to any page through the en.wikipedia.org because of above error. In the pass I had chosen the option to never use the mobile version of the site. Still set with cookie perhaps. Using an iPhone with OS Version 4. Regards,
SunCreator(
talk)02:32, 14 January 2011 (UTC)
Is there any way to perform a search and replace on one of the parameters supplied to a template? (Why I need this is a long-winded, convoluted story; fact is, I do need it, there's no easy way around it.) --
GutzaTT+22:08, 15 January 2011 (UTC)
(
edit conflict) The
AutoWikiBrowser has a feature where you can replace a deprecated parameter name with another. But before you fire it up and do such a run, consider whether it is really worth it. One reason could be that the parameter name was causing too much confusion that resulted in editors adding the wrong kind of data. For example, {{Japanese episode list}} had a parameter named |JapaneseTitle= which was for the Japanese
transliterated (romaji) episode title. However, editors were confusing it with the translated English title, which was suppose to go into |EnglishTitle=. So to help clarify what belonged in that field, the parameter was renamed to |RomajiTitle=. However, if it is simply a matter of preferring one parameter name over another, then it's best to leave the transclusions alone. —Farix (
t |
c)
22:28, 15 January 2011 (UTC)
I knew I had to include more info... :-) No, that wouldn't work -- I need to present one set of data to the reader, while using another set of data in the background. The two are interchangeable with a simple search and replace, but the magic needs to happen inside the format. --
GutzaTT+22:29, 15 January 2011 (UTC)
No, that would be fixed with a switch or what's-it-called. I need full search/replace, as in "100 foo" needs to be used once as "100 foo" and once as "100 bar" (obviously, this needs to work with any number). --
GutzaTT+22:58, 15 January 2011 (UTC)
For the record, the summary for my previous edit should've read "moar" instead of "moah". I wasn't being annoyed, I was being casual. --
GutzaTT+23:02, 15 January 2011 (UTC)
Need more detail. I'm guessing that what you should do is have two parameters, as template string functions are generally horrible.
OrangeDog (
τ •
ε)
00:16, 16 January 2011 (UTC)
(
edit conflict) No good way to do that, although it's possible you could fake it something like {{italic title}} does with disambiguators. If it's always "foo" and "bar", just have the template take the number and append the "foo" or "bar" in the template. Otherwise you could take the number and the unit as separate parameters.
Anomie⚔00:17, 16 January 2011 (UTC)
I just realized I have the perfect example: solve the {{Birth date}}microformat for
Cicero. You would need to change "106 BC" to "-106", and "43 BC" to "-43", while still using "106 BC" and "43 BC" in order to point to the relevant articles about the years. --
GutzaTT+00:34, 16 January 2011 (UTC)
Can a subtemplate solve it? The users template "T:Infobox X" has one parameter: "Foo = 123" for user (editor) input, the template calls a subtemplate with two parameters. So T:infobox X calls "{{Infobox X/core|Show = {{Foo}} year|OtherUse = {{Foo}}/100 century}}". Now the core has two separate input value for you to use. I used it in {{Infobox IPA}} (the subtemplate being {{Infobox IPA base}} -
DePiep (
talk)
16:32, 16 January 2011 (UTC)
Thank you all for the responses; (un)fortunately, OrangeDog's note makes all the difference, because this was exactly meant to treat ad/bc years, and the fact that the hCard format doesn't support non-Gregorian dates makes the problem go away (simply because it doesn't apply any more). --
GutzaTT+19:30, 16 January 2011 (UTC)
Removal of citation when editing page section not flagged as error until page published
When I removed a faulty citation while editing a page's individual section, the fact that another later section also referred to it was not flagged up in Show preview, only after Save page, when it appeared in large red letters (and presumably publicly) in the Citations section. As an occasional newcomer, I found this disconcerting. Have I missed something here? --
Peteradamson (
talk)
08:52, 16 January 2011 (UTC)
That is a correct analysis. The section-edit-preview looks only at one section, so it has no idea of dependent refs in other sections. Obviously the whole-page-version that is the article itself takes into account all the sections, and therefore finds the error, giving you the big red letters. It is unlikely to be a major problem for long though: a human volunteer or a bot will probably come and fix it in a timely fashion, though fixing the error yourself saves time. -
Jarry1250Who?Discuss.10:36, 16 January 2011 (UTC)
Although be aware that then you have the opposite problem: if you remove the last in-page reference without also removing the LDR definition, you'll get a different big red error.
Anomie⚔14:06, 16 January 2011 (UTC)
The system is described at
WP:REFNAME. As you say, preview doesn't reveal the problem. The only hint of the problem when you edited the section would have been that the reference in the wiki source started <ref name=...> instead of <ref>. In that case an editor should look at the whole article to see if the ref is used elsewhere and either remove other uses or copy the body of the reference to an other use (assuming is isn't already copied but that's rare). And yes, after clicking Save you see the same public version as everybody else.
PrimeHunter (
talk)
14:37, 16 January 2011 (UTC)
Question at Help desk re disable mobile site function
Hi. i`m from ko.wikipedia. I recently created
Template:Dated maintenance category on ko.wiki. but, i can`t categorize former articles that did not appointed date. so, i want to appoint date with my bot
ko:wiki:user:Altobot. How to command to my bot to appoint date on maintenance template? Of course, i don`t know when the templates attached.--
Altostratus (
talk)
10:21, 17 January 2011 (UTC)
Converting 4 km² into square miles, I get two different results: 4 km2 (2 sq mi) or 4 km2 (1.5 sq mi). Being a metric man, which one should I take?
Gun Powder Ma (
talk)
12:39, 18 January 2011 (UTC)
Depends on how much precision you need, which itself depends on context and on precision of the initial data.—
EmilJ.13:21, 18 January 2011 (UTC)
To expand on Emil's answer, the answer lies in the 0 and 1 in each template you used. This is the degree of rounding you want to use. So, when you want to round to zero significant digits, you are saying you want to round "1.5444086342 sq mi" to "2 sq mi". If you choose one significant digit, you are rounding "1.5444086342 sq mi" to "1.5 sq mi", two rounds to "1.54", three to "1.544", and so on. Everything outputted should be correct, it just depends on the amount of precision you want. If you choose to not include that parameter, then the template makes an assumption that you want the same number of significant digits output as the figure you input. So, if you input "4.5328 km2", it will output "1.7501 sq mi". Hope this answers your question. —
Huntster (
t@c)13:30, 18 January 2011 (UTC)
Thanks, you've answered my question indeed. 13:39, 18 January 2011 (UTC)
Gun Powder Ma (
talk)
Login problems
Hello everyone manning the pumps. I already posted this at the help desk but then thought it sounded more like something requiring a good seeing to rather than like a superficial problem.
I'm experiencing a very annoying sort of login problem: Somehow, Firefox has started to forget that I had already logged in for 30 days and has asked me to re-login every time I opened the browser (without me signing out or clearing my Firefox history or anything). I changed the password to rule out someone having a laugh with my account. Now I'm trying to login from a different computer and can only access the page immediately after the login (with my WP layout, links to my preferences and watchlist etc.). However, every time I try to access any other page, they tell me I'm not logged in anymore. Any help would be appreciated. Best,
212.117.123.14 (
talk)
13:13, 18 January 2011 (UTC)
Sorry - hadn't tried that before out of daftness: login via the secure server was no problem, it worked fine. What could be the reason why it didn't work via the normal route?
212.117.123.14 (
talk)
13:23, 18 January 2011 (UTC)
Do you have over-zealous security software configured to zap all cookies on sight, and which is running constantly? The loss of a cookie can cause logout. --
Redrose64 (
talk)
14:58, 18 January 2011 (UTC)
Done Please check it once and make sure everything's all right. I'm not sure if the q star should be a or , so if you want me to change it, I shall. Actually you don't need to be a MATH guru to do this. I have hardly any experience at all using LaTeX (none on Wikipedia)... You just need to know the basic markup. Try it!
ManishEarthTalk •
Stalk17:42, 18 January 2011 (UTC)
Collapsible list in reference
I am trying to write a template for the Portuguese wikipedia, that references to the Histories of Herodotus, and then prompts with many on-line editions of the text. However, there's a minor annoyance in the template, and I have no idea how to solve it - but it's probably something with the template Collapsible list and not the template I wrote.
Let me explain by giving an example. If I write (in the Portuguese wikipedia):
Leonidas was the king of Sparta<ref>{{citar heródoto|1|100}}</ref>
Heródoto, Histórias, Livro I, Clio, 100 [Expandir]
Which means: Herodotus, Histories, Book I, Clio, 100 and a small box with "expand!". The problem here is that this formatted text is preceded by a CR-LF, making it ugly to read.
Look like this is being used to list different versions of a reference. Only the actual version being used as a source should be directly cited. The other versions would be better in a further reading section. ---—
Gadget850 (Ed)talk18:43, 18 January 2011 (UTC)
Mysterious 'Grey' Bar at top of pages
Has anyone else experienced this? I've been getting this slim 'grey bar' at the top of the pages, on & off. It appears right under the project page, discussion, edit this page, history, watch directives.
GoodDay (
talk)
03:46, 14 January 2011 (UTC)
Maybe a screenshot & paste of the source code of the page with the grey bar (into a sandbox somewhere) would help. –
xenotalk16:52, 19 January 2011 (UTC)
Could everybody experiencing this please state OS, browser and skin - I have seen nothing amiss in Windows XP/Firefox 3.6/Monobook. --
Redrose64 (
talk)
17:10, 19 January 2011 (UTC)
Windows 7, IE, Monobook. As I said over at the help desk, the gray bar starts appearing for me whenever something triggers IE to switch to "compatibility view".
Deor (
talk)
14:28, 25 January 2011 (UTC)
I am totally at loss but maybe someone here has an idea what's going on. For a few days two of the user scripts I'm using, rollbackSum.js and revisionjumper.js from the Preferences, suddenly do not work anymore. rollbackSum.js does not prompt me for a message on rollback, and revisionjumper just gives me an empty dropdown box next to the "← Previous edit" link. There has been no update on rollbackSum.js for over a year. I have completely purged the browser cache, disabled addons in Firefox, but nothing helped. Any ideas?
Nageh (
talk)
21:15, 17 January 2011 (UTC)
No errors or warnings pertaining to the user scripts. I can even verify that the script's addOnloadHook() function is in the onloadFuncts[] array, and I can call it from there but nothing happens. I tried to have a look at
User:Porchcrop/rollbackSum.js but unfortunately I'm not too familiar with regular expressions. The script should add an onclick handler to a link like <a href="https://en.wikipedia.org/wikipedia/en/?title=User:Nageh&action=rollback&from=Nageh&token=f5dd1ccb2e4d62fe2b0760da5d52f9e4%2B%5C" title=""Rollback" reverts edit(s) to this page of the last contributor in one click">rollback</a> but it doesn't happen.
Nothing seems to be wrong with the RegXs. The script works when I put it into my console... I also imported it... Works fine.... (I'm using chrome). Normally these problems are caused by a syntax error elsewhere (If there's a syntax error in your monobook/vector js, it won't even be imported.), but your vector.js looks fine (and you haven't touched it for months, which rules out the possibility completely). Maybe it's a firefox error?
ManishEarthTalk •
Stalk07:36, 18 January 2011 (UTC)
When a new article is created, it is listed at
Special:Newpages for up to 30 days so that it can be patrolled by a
new page patroller. If the article hasn't been marked as patrolled after 30 days, it falls of the Newpages list forever. I've recently been working on a bot which keeps a list of the articles which fall off the Newpages list. The original intention was that this additional backlog of articles would eventually get patrolled and we wouldn't miss anything. However, in the course of testing things out, it appears that articles cannot be patrolled after 30 days. I believe this is because the rcid of their creation edit expires after 30 days, and a valid rcid is required to patrol the article. I'm starting this discussion to find out the answer to the following questions:
Is it technically possible to extend the expiration time of the rcid past 30 days (perhaps by changing
mw:Manual:$wgRCMaxAge)?
If it is technically possible, is there a consensus to make such a change?
If the answer to #1 or #2 is "No", is there any other workaround that is possible? For instance, is it possible to modify the patrolling mechanism such that it doesn't require a valid rcid? Or, at the very least, it doesn't require a valid rcid if one doesn't exist?
There is no reason that I can think of why new articles shouldn't be able to be marked as patrolled after 30 days (even if they are no longer on
Special:Newpages). I'm also not clear on exactly why an rcid is required to mark a page as patrolled. What is the purpose of the rcid with respect to marking a page as patrolled? Anyway, in closing, marking pages as patrolled is an excellent way of tracking new articles, and making this change would allow us to track new pages even after they fall off the Newpages list. The bot has only been running for a few days, but in the last 3 days alone there have been 175 articles that fell off the Newpages list without being patrolled. Thanks.
SnottyWongsqueal00:33, 19 January 2011 (UTC)
1. Yes and no. There is no technical limitation that prevents it from being set higher than 30 days, but this is unlikely to happen. This was discussed in October, the reasons are given in detail
there. There isn't any point to marking a page not on Special:Newpages as patrolled because Special:Newpages is the only place where "patrolled" status is actually displayed. There's the patrol log, but since it doesn't show what isn't patrolled, its not particularly useful for newpage patrol. An rcid is required to match the page to the corresponding entry in the recentchanges table for the edit that created the page. Mr.Z-man07:05, 19 January 2011 (UTC)
The October discussion was begun by a user based on my initiative. The suggestions for an extension beyond 30 days was turned down by consensus, but not on technical grounds. --
Kudpung (
talk)
15:09, 19 January 2011 (UTC)
See the comments by MZMcBride. One of the Wikimedia sysadmins was consulted and said that it was unlikely to be increased. Mr.Z-man15:30, 19 January 2011 (UTC)
Is it technically feasible to remove the requirement that a valid rcid is required for marking a page as patrolled? I understand that an rcid is required to correlate the page to the recentchanges table, but I still don't understand what purpose that serves. That information doesn't seem to be preserved. You can't go to the patrol log and find the rcid that was patrolled, or even the revision that was patrolled (without manually noting the patrol timestamp and checking it against the article's edit history). The rcid eventually dies anyway and (presumably) is reused for another edit some time down the road. So, why can't we just mark a page as patrolled without providing an rcid? That would solve the problem of articles being unpatrollable after 30 days without the resource problems posed by extending the rcid expiration time.
SnottyWongprattle15:37, 19 January 2011 (UTC)
A page can't be marked as patrolled after it's removed from the recentchanges table because there's nothing left to patrol after that. You're not actually patrolling the page or revision, you're patrolling the recentchanges entry for the revision. That's why you need rcid: it corresponds to rc_id in the recentchanges table. See the
recentchanges table documentation on MediaWiki.org. Once it's gone from the recentchanges table, there is nothing that can be patrolled. I have confirmed that it works this way after examining the code.
Reach Out to the Truth19:00, 19 January 2011 (UTC)
Thanks. So it appears that the only way to make this happen is to extend the rcid expiration time, and it sounds like the side effects from doing that would be too troublesome to justify making the change for this purpose. I just wanted to make sure there wasn't a practical way to make this happen, and it doesn't appear like there is. Thanks all for the answers.
SnottyWongspout19:37, 19 January 2011 (UTC)
Search suggest order
The order of search suggest items used to be alphabetical, now it seems based on number of hits. Is that correct? How can I change my own MediaWiki wiki to do this as well?
Is there a way\an option to get Rollback to add pages to you watchlist that have been rv by the editor in the same way that you can with normal editing?
Simply south...
..17:35, 19 January 2011 (UTC)
I keep getting prompted to log in
Every time I start a new WP session, I start on the Main Page, and I'm prompted to log in. But when I go to the login page, I am already logged in. What's going on, and how do I stop this from happening? Corvus cornixtalk18:42, 19 January 2011 (UTC)
Foe some reason I can't stay logged in while using Google Chrome but can in IE. This problem started <20 minutes ago and I have used Google Chrome while editting Wikipedia for >1 year. I click log in, put in my password, get 'Login successful' page, then I go to another page and I'm no longer logged in. The same thing happens if I log in on the secure and standard server. J04n(
talk page)01:45, 13 January 2011 (UTC)
I can't duplicate this - I'm using chrome 8.0.552.224 on Windows Vista - tried both the "regular" and "incognito" modes with both my regular account and this one. Perhaps it's a different version of Chrome - can you give your version & the OS that you're working with and perhaps someone with a similar 'flavour' can duplicate this.
Skier Dude2 (
talk)
02:32, 13 January 2011 (UTC)
Hmm, well I'm also using chrome (10.0.634.0), I have no problems, and I don't recall having any in the past. If it is a bug with chrome it's a bug with chrome, nothing we can do. It isn't in my version.
Prodegotalk06:13, 13 January 2011 (UTC)
Also fine on 9.0 Have either of you been fiddling around with your cookie settings recently? Try taking a look at Tools>Options>Under the bonnet>Content settings>Cookies and make sure they are allowed and not blocked. -
Kingpin13 (
talk)
06:16, 13 January 2011 (UTC)
If I can throw my hat in the ring, I'm not getting the issue on Chrome 9 for Ubuntu. Are you sure the solution isn't to actually file a bug with Chrome? Also, if I can patronize everybody and remind you to turn on cookies, and also clear your cache and cookies.
Magog the Ogre (
talk)
01:53, 15 January 2011 (UTC)
Appears to be resolved at bugzilla - "commented out the $wgCentralAuthCookieDomain = in CentralNotice.php" was the issue.
Skier Dude (
talk)
06:24, 15 January 2011 (UTC)
I experienced the same problem since 14 January and now solved. Try to merge your accounts by clicking
Special:MergeAccount. If you have a message "Login unification Not complete!", then probably the cause of your problem is
m:Help:Unified login. Enter the password for the merge completed. ――
Phoenix7777 (
talk)
00:36, 21 January 2011 (UTC)
"Invitation to edit" template and m.wikipedia.org
Is there anything that could be changed in
Template:Invitation to edit that would make it appear collapsed by default, or even better, not appear at all on
articles in m.wikipedia.org? This is a fairly urgent problem.
Cheers,
Looks like it is already collapsed by default. As for hiding it on the mobile site, you could use the {{SERVERNAME}} magic word to check. In other words, wrap the whole template in {{#ifeq:|{{SERVERNAME}}|en.m.wikipedia.org|all your template stuff in here|}}.
SnottyWongcomment00:57, 19 January 2011 (UTC)
A nice idea thank you. But unfortunately it doesn't seem to work. See
[6]. It is reporting the server name as en.wikipedia.org. — Martin (
MSGJ ·
talk)
11:25, 19 January 2011 (UTC)
Hrm, you're right. That seems like a bug to me. Does that seem like a bug to anyone else? If you're browsing wikipedia on the en.m.wikipedia.org server, then {{SERVERNAME}} should resolve to en.m.wikipedia.org, not en.wikipedia.org, right?
SnottyWongspill the beans17:35, 19 January 2011 (UTC)
Meanwhile TheDJ seems to have fixed the immediate problem with the template by using class="metadata". (Thanks for that.) — Martin (
MSGJ ·
talk)
20:52, 19 January 2011 (UTC)
SERVERNAME returns the name of the server the page is stored on. After all, magic words are resolved into HTML on every purge. en.m.wiki is basically a (legitimate) mirror site of wikipedia with a different stylesheet. So it'll return en.wiki no matter what.
ManishEarthTalk •
Stalk04:24, 21 January 2011 (UTC)
Emil is right, but to answer your question, editcounts and all change every second. That means, pages using these templates will have to purge constantly. Not good... The only way to program new ones would be if you had commit access to mediawiki, or if you developed your own extension. But you could only use it if you included it in your own wiki.
ManishEarthTalk •
Stalk04:18, 21 January 2011 (UTC)
Book Error
I was merrily downloading a PDF version of
a book I created through Wikipedia, only to find after about 80% of the way through the following error ocurred:
An error occured on the render server: RuntimeError: command failed with returncode 9: ['mw-render', '-w', 'rl', '-c', 'cache/07/0710d0f851b106e5/collection.zip', '-o', 'cache/07/0710d0f851b106e5/output.rl', '--status', 'qserve://localhost:14311/0710d0f851b106e5:render-rl', '--template-blacklist', 'MediaWiki:PDF Template Blacklist', '--template-exclusion-category', 'Exclude in print', '--print-template-prefix', 'Print', '--print-template-pattern', '$1/Print', '--language', 'en']
I checked the collection and it renders fine (manually, not through the "webinterface"). I believe the problem is the sheer size of the collection: fetching the resources and rendering took ~50 minutes on an otherwise idle machine. Currently there is a time limit of 60 minutes after which render jobs are killed - and that's what probably happened. Could you please check if splitting the collection in two parts works? I will check if I can increase the time limit.--
Volker.haas (
talk)
15:36, 20 January 2011 (UTC)
answered (by xeno; even double solutions!) -
DePiep (
talk)
18:24, 20 January 2011 (UTC)
I have checked
the SP tools without result. My question is: I want to create a link to a predefined Special Pages list, preferably wiki-internal not URL. How to?
Example & exactly: on my User:-homepage I want a wikilink to all my subpages. At the moment I do: 1. via WP toolbox menu, click page "Special pages", I enter the requested prefix "User:DePiep", and -- yes, the list OK, not by a click. Can I single-click this as a link? -
DePiep (
talk)
18:06, 20 January 2011 (UTC)
Yes it is there, but how could I get to the helpful page... . Still, as the Mexican says when asked for a "
whoomahh" and than by plane touched down in Rome: gracias'. -
DePiep (
talk) 19:49, 20 January 2011 (UTC) oops, this could be read as offencive. Which is not what I intended. I tried to say: "thank you" in a different language, and with a twist. -
DePiep (
talk)
20:00, 20 January 2011 (UTC)
Request action for bots to take care of
Help changing more then --> more than
A brief description for those who aren't familiar with the difference between "then" and "than"
Then is used to describe time:
Example: All if-then statements. If I buy a car then I will be in debt. I arrived before then. Then I will always be ready.
Than is used to compare things:
Example: I have more than you. The poor man is less sad than those taken over by greed. We are closer to the heavens than the animals.
I came across two users in the past few hours that made this grammatical error, and decided to look into it. A quick search of "more then" yielded enormous amounts of results using it incorrectly, and that does not cover all the other incorrect uses being made like "less then" and "more advanced then", etc.
Unfortunately there are situations where "more then" can be correct i.e. "If I buy more then I will have more", and despite being far less common than the incorrectly used "more then" a blanket sweep would remove some legitimate uses.
Is there any way to program a bot or device to help with this, or should we perhaps make an official guideline to describe common grammatical errors(comma usage) and how to fix them? --
AerobicFox (
talk)
07:06, 19 January 2011 (UTC)
How many are we talking about?
AWB with a competent editor can handle 5 - 10 a minute, and I'm not sure I'd trust a bot with the complexities of English grammar.
Someguy1221 (
talk)
07:25, 19 January 2011 (UTC)
Perhaps hundreds of thousands of misuses of then and than. The main problem though is people are often quoted using them incorrectly, and we shouldn't misquote them. I suppose something like AWB will need to be used though. Can we just put people on alert to this? It is probably the most serious, heinous, and despicable epidemic of grammar within the English language, and it has invisibly rooted itself deep into Wikipedia.--
AerobicFox (
talk)
07:31, 19 January 2011 (UTC)
Perhaps we could have an essay, "
WP:Editing for grammar" to warn of the most common issues, and link that essay in various places, as part of larger edits. We want to avoid clogging Wikipedia with 1-word revisions, because all are kept (so far) to show copyright authorship, despite my plans for "
WP:Pruning article revisions". The good news about grammar errors: "
Where there's smoke, there's fire" - when you spot use of "more then", it is likely just the smoke indicating more problems in an article; hence, searching for grammar problems is a great way to find articles needing major rework (assuming the subject is worth your time). Even when quoted, perhaps consider adding "[<b/>[[sic]]<b/>]" with a spelling problem. Also, remember the bots which were fixing the spelling errors in vandalism jokes? Bots were just flooding WP with more useless revisions, of properly worded vandalism, as the Most Smartest Bots. Meanwhile, we DO NOT want bots polishing bad articles to "unsmoke" them; instead, "bad grammar predicts bad data" and we want to find poorly written articles to improve all of their content, and references, not just the grammar. From that viewpoint, bad grammar is a
good thing, as an indicator of an article's overall poor quality. Correcting only small mistakes is a big mistake. -
Wikid7711:27, 19 January 2011 (UTC)
I disagree with you on two things:
First, I can see only one problem with lots of small edits – it's flooding watchlists. And even that is just a minor problem. There really aren't any technical problems with that. Also,
WP:DWAP.
Second, bad article with good grammar is (slightly) better than bad article with bad grammar. If you see an article, you can identify whether it's good or bad regardless of the quality of its grammar. And I don't think any significant amount of searching for grammar errors is going on, apart from trying to fix just them plus some mostly automated fixes (most likely using AWB). So, no, grammar errors shouldn't be kept in articles, they should be fixed. Correcting only small mistakes is bad, not correcting anything is even worse.
My computer used to open PDF documents in the same window as the webpage. Now, it has started downloading them and then opening in a separate window. I cannot work with it like this, especially when it is necessary to go from one page of a scanned journal to another and another and back - end up crashing the bottom taskbar like that.
It was affecting my research and editing of Wikipedia. Anyway, I've managed to reset the blessed thing so that it behaves itself. Thanks anyways, will remember that page next time I need it.
Mjroots (
talk)
16:21, 22 January 2011 (UTC)
I am attempting to create a book but when i download it as a PDF it says that the file is corrupt and cannot be retreived and when i download it as an ODF it downloads a blank document
Admiralalexmann (
talk) 19:55, 10 January 2011 (UTC)
The list of pages in my book are:
Series's [Rename]
Remove Show Star Trek: The Original Series
Remove Show List of Star Trek: The Original Series episodes
Remove Show Star Trek: The Next Generation
Remove Show List of Star Trek: The Next Generation episodes
Remove Show Star Trek: Deep Space Nine
Remove Show List of Star Trek: Deep Space Nine episodes
Remove Show Star Trek: Voyager
Remove Show List of Star Trek: Voyager episodes
Remove Show Star Trek: Enterprise
Remove Show List of Star Trek: Enterprise episodes
Remove Movies [Rename]
Remove Show Star Trek: The Motion Picture
Remove Show Star Trek II: The Wrath of Khan
Remove Show Star Trek III: The Search for Spock
Remove Show Star Trek IV: The Voyage Home
Remove Show Star Trek V: The Final Frontier
Remove Show Star Trek VI: The Undiscovered Country
Remove Show Star Trek Generations
Remove Show Star Trek: First Contact
Remove Show Star Trek: Insurrection
Remove Show Star Trek Nemesis
Remove Show Star Trek (film)
Remove Ships [Rename]
Remove Show USS Enterprise (XCV 330)
Remove Show Enterprise (NX-01)
Remove Show USS Enterprise (NCC-1701)
Remove Show USS Enterprise (NCC-1701-A)
Remove Show USS Enterprise (NCC-1701-B)
Remove Show USS Enterprise (NCC-1701-C)
Remove Show USS Enterprise (NCC-1701-D)
Remove Show USS Enterprise (NCC-1701-E)
Remove Show USS Voyager (Star Trek)
Remove Show Delta Flyer
Remove Show USS Equinox
Remove Show Deep Space Nine (space station)
Remove Show USS Defiant
Remove Show USS Excelsior (Star Trek)
Remove Show USS Reliant
Remove Ship Classes [Rename]
Remove Show NX class starship
Remove Show Constitution class starship (Star Trek)
Thank you so much! I don't know why it would not work for me. I am currently doing a whole load of Star Trek related books, that general one and one specifically for each series. Ifyou are interested let me know on my talk page.
Admiralalexmann (
talk)
19:21, 23 January 2011 (UTC)
Latest computer-math precision problems
When using the math function floor(x) in
parser function #expr, I have been adding a correction factor to avoid tiny differences in the precision of the numbers.
Particularly, rounding by -5 is leaving small differences (a known bug), so I am adding 1E-9 (0.000000001) to compensate, rather than rounding twice (as "round -5 round 0"). Here are the latest tiny precision differences for values rounded to 100,000, 200,000,... 900,000 as rounded by -5:
{#expr: 1 - ( 80333 round -5)/1E5} = 0
{#expr: 2 - (180333 round -5)/1E5} = 0
{#expr: 3 - (280333 round -5)/1E5} = 0
{#expr: 4 - (380333 round -5)/1E5} = 0
{#expr: 5 - (480333 round -5)/1E5} = 0
{#expr: 6 - (580333 round -5)/1E5} = 0
{#expr: 7 - (680333 round -5)/1E5} = 0
{#expr: 8 - (780333 round -5)/1E5} = 0
{#expr: 9 - (880333 round -5)/1E5} = 0
These are current precisions for 100,000 (etc.) rounded by -4:
{#expr: 1 - ( 96333 round -4)/1E5} = 0
{#expr: 2 - (196333 round -4)/1E5} = 0
{#expr: 3 - (296333 round -4)/1E5} = 0
{#expr: 4 - (396333 round -4)/1E5} = 0
{#expr: 5 - (496333 round -4)/1E5} = 0
{#expr: 6 - (596333 round -4)/1E5} = 0
{#expr: 7 - (696333 round -4)/1E5} = 0
{#expr: 8 - (796333 round -4)/1E5} = 0
{#expr: 9 - (896333 round -4)/1E5} = 0
The use of "round -4" did not show any precision problems, as of this writing.
I had thought the precision with "round -5" had been off by 2.9E-13 a few weeks ago, so I wonder why the precision is now 15 digits. How long will the precision remain at 15 digits? -
Wikid7711:35, 19 January 2011 (UTC)
The expressions above bare little resemblance to what you are describing (where are the 1E-9, 100000, etc.). Why are you attempting calculations to this precision using parser functions? Why do you expect them to perform better than standard 64-bit IEEE floating-point arithmatic?
OrangeDog (
τ •
ε)
23:09, 19 January 2011 (UTC)
Sorry, I had skipped several steps in the explanation, above, to avoid a
shaggy dog story. I am rewriting parts of
Template:Convert to handle ranges of billions without exceeding the tiny expansion limit of 40 nested if-else or nested templates. The rewrite uses the math function floor(x), which implicitly uses extreme precision, because "floor(2.0)" gives "1" when the internal value of 2.0 is just a tiny amount below 2. The examples, above, can be expanded as follows:
So, the specific problem is when floor(2.0) yields "1" due to the internal value for 2.0 really being 1.999999999999997 (or such). The fix adds 1E-9 (0.000000001) as being floor(2.0+1E-9) yielding "2" as intended. Internally, adding 1.999999999999997+1E-9 generates x=2.000000001 with floor(x)=2. I hope that clarifies all the missing steps, above. -
Wikid7719:35, 23 January 2011 (UTC)
The formulation is somewhat confusing: the internal value of 2.0 is exactly 2, the problem is that the round function can give an internal result deviating from the actual round number, which does not affect the displayed result, but can affect subsequent computations.--
Patrick (
talk)
00:54, 24 January 2011 (UTC)
Last time I looked into it (two years ago maybe?), the precision of parser function arithmetic actually depended on which server did the processing. Some servers were giving 15/16 decimal digits while others gave 12/13 (I think?). I understood at the time that 15/16 would become the standard as an incidental consequence of server upgrades planned at that time, but maybe that hasn't been the case. In general though, parser functions are not designed for accurate arithmetic, and I would suggest that it is probably best to avoid using them in ways that would require very precise arithmetic. (They actually could be tweaked to make roundoff errors quite rare at the expense of some precision and more expensive operations, but developers where against those efforts. Actually, some developers were generally opposed to anything that would make parser math more useful, on the grounds that they didn't want to encourage its use.)
Dragons flight (
talk)
23:45, 19 January 2011 (UTC)
Thanks for the background details. I had suspected some of the servers were running precision as 12/13 digits, with other servers running 15/16 digits. -
Wikid7719:35, 23 January 2011 (UTC)
Deleting text by using <references>?
I understand that using the <references> tag can delete text. This is what happened. I made
this edit, opening just a section. Before saving, I checked the references by adding <references> to the bottom of the section, and preview. Without removing it (as was the intention, clearly) I saved it. Then the article missed all text below the inserted tag. Deleted (see section "Works" in the link).
Is this a know bug? -
DePiep (
talk)
02:17, 23 January 2011 (UTC)
The problem appears to be that the inclusion of <references> on line 92 (before "Works " Section) should have been <references/> or {{Reflist}}. The inclusion of a beginning reference tag has the software looking for a closing reference tag, and when it's not found will produce this type of issue. In addition, the tag {{Reflist}} was already on the "Notes" section, which will reflect all of the correctly formatted reference tags above it, so the inclusion of the <reference> tag wasn't needed.
Skier Dude (
talk)
03:51, 23 January 2011 (UTC)
Sure it wasn't needed, except temporarily to check refs (in preview) when editing a single section (I should have deleted it afterwards). And indeed, I saw the closing <.../> slash is prescribed, with sense as you describe. But, this joke throws away all following text? Wow. That's the worst typo I can think of. Not even an error "no closing tag" or so, or nullifying the effect. Anyway, thanx for the answer. -
DePiep (
talk)
09:27, 23 January 2011 (UTC)
It doesn't "throw it away", per se. The parser sees an opening tag, and starts bundling up all the containing text until it finds a closing tag, then it hands the text off to the processing function for references to do with it as it wishes... which in this case is nothing because the <ref /> doesn't use the contents of the tag. In the same way that if you leave an unclosed <source> tag, it syntax-highlights the rest of the page, or an unclosed table puts the rest of the text into a table cell, this error turns the rest of the page into contents of a tag. It's only for <references> specifically that this results in the content not being included.
Happy‑
melon11:55, 23 January 2011 (UTC)
I get it, if after saving next I edit out (delete) the opening tag, the missing (i.e. unshown) content is visible again. Still quite distressing for a semi-pro like me. At least, with a spoiled table I see some effects. Now to learn something extra, if I propose "this tag requires closing tag or it does error/noeffect", at which (wikimedia) level is there a No? -
DePiep (
talk)
13:36, 23 January 2011 (UTC)
One way to work around the issue is to not add <references /> by yourself, but add the code importScript('User:Js/ajaxPreview.js'); to your
Special:Mypage/skin.js. That adds a Preview button near the right top corner of the edit box, which shows references automatically.
Svick (
talk)
13:50, 23 January 2011 (UTC)
I was pointed here from the Helpdesk for help with this problem. I have simply quoted what i posted there.
I have two accounts.
User:thisisftg and this account
User:darkskynet. The "Thisisftg" account was made while i was in high school. (Thisisftg was my online identity at the time). The old account has a creation date of January, 2007 and the new account that i currently i am using has a creation date of 2009. Is there anyways a Admin could make the creation date on
User:Darkskynet to be the date for the old account and simply delete the old account. The old account has no edits so it would not mess with the database. I just would really like the old date. This would make my century. Cheers, --
Darkskynet (
talk)
12:30, 23 January 2011 (UTC)
Is it possible, easily, to exempt pages in User space from the blacklist restrictions that mean that titles with certain forms can't be created? (
This is the sort of situation where it causes unnecessary problems.)--
Kotniski (
talk)
08:39, 24 January 2011 (UTC)
It can be done on an individual basis as needed, but just exempting userspace in general would mostly defeat the purpose of the blacklists. Mr.Z-man20:29, 24 January 2011 (UTC)
font-related class="IPA" disrupts a #switch:
The template IPAsym uses a #switch: to output an article name from an input (IPA) symbol, e.g. the symbol "p" (=regular letter p):
Now when the editor uses the {{
IPA|p}} template for input (not unexpected, since this is within IPA-world), the template does not find it (returning the #default=red error):
{{
IPAsym|{{IPA|p}}}} → Lua error in Module:IPA_symbol at line 51: bad argument #3 to 'format' (string expected, got nil).
My question is: should/can I determine or "undo" that class="IPA" from incoming input? Any other way of thinking? (Of course, warning & documenting left alone, and within a template I am in control).
Details: 1) This problem pops up only now as I am working on nested templates & infobox, I know no situation when an editor did so in a straight edit. {{IPAsym}} is existing, it is well used/tested). 2) Same effect with {{unicode}}. -
DePiep (
talk)
14:21, 24 January 2011 (UTC)
No. The output from {{
IPA|p}} is <span title="Representation in the International Phonetic Alphabet (IPA)" class="IPA">p</span>. Literally all that template does is wrap its input in that titled span. If you don't want the content to be in the span, don't call the template. Calling the template to wrap the span, then looking for another mechanism to undo that effect, is totally circular processing.
Happy‑
melon15:10, 24 January 2011 (UTC)
I can live with the "no". I think your word 'circular' is not correct ("circular" as in "tail-baiting" calling? Or even "recursive"?), I understand it to mean "once enveloped, you cannot un-envelop; even 2nd envelop cannot undo the 1st envelop". Thanx for answering. -
DePiep (
talk)
16:59, 24 January 2011 (UTC)
Looking for a MediaWiki page
Hello! Which MediaWiki page displays the box saying "Wikipedia does not have a user page with this exact name."? Also, is there a page showing all MediaWiki pages the English Wikipedia uses?
HeyMid (
contribs)
17:53, 24 January 2011 (UTC)
I display my rollback privilege on my user and talk page and was recently granted autopatrol status -- but when I try to put the icon on my user and talk page, it displaces my rollback icon rather than joining it. How do I avoid this problem? DRosenbach(
Talk |
Contribs)02:50, 25 January 2011 (UTC)
Well, I don't understand what to do. The instructions are not explicit enough and they don't directly address my problem. I've tried putting that text into the icon display HTML code in various ways and none of them suffice. DRosenbach(
Talk |
Contribs)20:35, 25 January 2011 (UTC)
The automatic date formatting provided by the {{#formatdate}} magic word doesn't seem to be working. I have my date format set to the UK form like 24 January 2011, but the
current issue of the Signpost is giving the date as "2011-01-24", in ISO 8601 date format. Has something recently changed in MediaWiki that might have caused this problem? Graham8703:49, 25 January 2011 (UTC)
see
bugzilla:18479 where it was disabled and the server admin log: January 2: 11:48 logmsgbot: hashar synchronized php-1.5/wmf-config/InitialiseSettings.php '18479 - Set $wgUseDynamicDates = false for the English Wikipedia'
Can we file another bug then? I assumed we wanted formatting for links turned off, but still allow {{#formatdate}} when we wanted it (e.g. interface messages, newsletters, etc.).
OrangeDog (
τ •
ε)
12:39, 25 January 2011 (UTC)
The consensus was that we should not use date autoformatting in articles. Among other reasons, in general, it does not work. For example, the commas can't be made to work right if the American style is Obama will give the state of the union address January 25, 2011, in the evening. If date autoformatting were still in use, the international style would be rendered Obama will give the state of the union address 25 January 2011, in the evening. Newsletters were not discussed, but the same problem applies to them.
Jc3s5h (
talk)
14:03, 25 January 2011 (UTC)
As OP demonstrates, the correct use is {{#formatdate:2011-01-24}} (international standard), not to use a different regional format. We have many pages other than articles, including interface pages and pages explaining how {{#formatdate}} works. Presumably there are all affected too.
OrangeDog (
τ •
ε)
14:19, 25 January 2011 (UTC)
Cross-compiling C++ programs for Mac OS X on Linux host
Thanks for your confidence in Wikipedia however these questions are out of scope and should be targetted to a Linux forum. Of course it is possible but I haven't a clue which libraries you would need, or if any one has done it. --
ClemRutter (
talk)
12:12, 25 January 2011 (UTC)
I am trying to write some code to parse out occurrences of "plain text" external link additions. For example,
http://www.example.com would be one (using no brackets). Per
Help:URL there is a certain set of permitted characters for URLs. The comma is not among them, but it seems to be parsed as part of the URL. For example,
http://www.example.com,bla includes the comma in the URL. Other characters break the link (including a space). For example, see the code behind for
http://www.example.combold. Was the comma omitted from this list? are there others? or do I have some misunderstanding?
Similarly, ahttp://www.example.com does not turn to hypertext, but boldhttp://www.example.com does. It seems like certain mark-up characters are "space equivalent". Is there a list of these anywhere? All I want to do is take some wiki-markup and write the regular expressions to pull out these links. Obviously, when people use the typical bracketed syntax -- life is much easier. Thanks,
West.andrew.g (
talk)
15:36, 25 January 2011 (UTC)
You might be able to simplify this somewhat without too many problems, but AFAICT this is the procedure MediaWiki follows to find bare links: First you have to expand templates and avoid considering text involved in quite a bit of other wiki markup (e.g. internal links, bracketed external links, and text in comments, <nowiki> tags, and the like) and text between the < and > of recognized HTML-like tags. Then you have to find out the value of
$wgUrlProtocols, which if different from the default may or may not be in the configuration files publicly available at
http://noc.wikimedia.org/. $wgUrlProtocols is most likely an array of strings; escape any regexp metacharacters and join them all together separated by "|" (call this "$protocols"). Then you pull out everything matching the PCRE regex \b(?:$protocols)[^][<>"\x00-\x20\x7F]+, i.e. any substring that doesn't follow a word character (a–z, A–Z, 0–9, or "_", IIRC), and consists of any of the protocols in $wgUrlProtocols followed by one or more of any character other than controls (U+0000–U+001F or U+007F), space, square brackets, less-than, greater-than, double-quote. Then post-process each found string as follows:
Truncate at the first instance of "<" or ">".
Consider the characters comma, semicolon, backslash, period, colon, exclamation point, and question mark. If the string doesn't contain a left parenthesis, also consider right-parenthesis. Remove any of these characters from the end of the string, e.g. change "http://www.example.com/?foo=1!?!?!," to "http://www.example.com/?foo=1".
Remove any whitespace characters and U+00AD, U+1806, U+200B, U+2060, U+FEFF, U+034F, U+180B, U+180C, U+180D, U+200C, U+200D, or U+FE00–U+FE0F from the hostname portion of the URL.
Right, here we go, xeno. Charlie Chan in the Secret Service will do to demonstrate the glitch. That's what it looks like. I use Windows Vista/IE. It disappears for me, too, when I refresh, only to pop up again on another page. I'm still seeing it; so I know nobody's come up with an answer yet.
Kelisi (
talk)
17:02, 25 January 2011 (UTC)
All right, it's Snippet #82300 on dpaste. Can you see it? The URL says
http://dpaste.org/NrAs/ but I didn't think that was very useful. Can you tell from my choice of article that I'd never even heard of pastebins?
Kelisi (
talk)
17:32, 25 January 2011 (UTC)
I seems to me that it probably has something to do with the random banners. You could try adding the code
here to your
skin.js and
clear your cache; that should display the name of the currently-selected banner in the upper-left corner of the page, and you can see if the grey bar shows up in combination with a particular banner.
Anomie⚔18:20, 25 January 2011 (UTC)
Can't log in to the secure server
Hi, I'm having a problem where I try to go to my watchlist, log in successfully (hello monobook!) on the secure server, then the next page I load via secure server I am no longer logged in. Anyone else experiencing similar problem? Browser=Firefox 3.6.13.
192.131.232.56 (
talk)
19:57, 25 January 2011 (UTC)
Not sure where my question goes, or perhaps it's already been answered here and I don't know enough to understand -- but a new user edited
Susan G. Komen for the Cure with a small bit of actual information, and the edit also included a bunch of weirdly placed <noinclude> and <onlyinclude> markers. I suppose the editor could have added them himself, but the edits in question were his first edits ever and that seems unlikely. I edited the article to remove the bits of code, but I didn't understand it and thought I'd mention it.
Sweet kate (
talk)
21:07, 25 January 2011 (UTC)
Diff in question
[8]. The code would've rendered a short blurb about how much they've rasised since inception (had you transcluded the article page). Ask them maybe =] –
xenotalk21:14, 25 January 2011 (UTC)
I would say that this is pure testing by a newbie. They've spotted this menu thing below the "Save page" button, changed it from "Insert" to "Wiki markup", and then clicked a few things, not really realising what they were doing. Give them a {{
subst:welcome}} and direct them to the sandbox. --
Redrose64 (
talk)
21:30, 25 January 2011 (UTC)
Those were not "random" noincludes; they very carefully isolated this section of the article when included into another page:
"Since its inception in 1982, Komen has invested over $1.5 billion for
breast cancer awareness, research, and health services, making it the largest breast cancer
charity in the world. Today, Komen has more than 100,000 volunteers working in a network of 125 affiliates worldwide."
Many editors, seemingly new with a new account name, are actually experienced editors who might be editing a controversial area, afraid of repercussions with their other username(s). Some admins have blocked a person's username for 1 week or 1 month when disagreeing with other editors. Technically, a blocked person should not edit with any username while blocked, but their other username(s) do not record the stigma of the blocked username, and so appear to be quiet users who rarely enter quarrels. Also, the use of "noinclude" tags allows an article to be included within another article where the no-included portions will be omitted there. Editing an article that way should be addressed at the Village Pump Policy discussion (
WP:VPP), to see if there are restrictions to using "noinclude" tags in an article. Some people think cancer (or earthquakes) are God's punishment, so as usual, be careful when contacting a short-term user, where there is little edit history to check for mental stability. As you might know, the U.S. has a severe mental health crisis, with handguns everywhere, and Wikipedia is a target website, aka a
psycho magnet. So, consider using a "2nd anonymous username" (see restrictions under
WP:SOCK) when contacting unknown people. -
Wikid77 (
talk)
21:57, 25 January 2011 (UTC)
I think mobile wikipedia articles need links to other languages.
In smartphone, I use the mobile wikipedia but articles don't have links to other languages. For example, if I find "동물"(Korean word, meaning is animal) in Korean mobile wiki, and I needs more information in English, I can't go to English mobile wikipedia article "animal" easily. So I think mobile wikipedia articles need links to other languages.
When I put in a certain username in the title searching through
Special:Log/renameuser, it usually means that the old name of the user has to come first. Is there a version of MediaWiki available that can include both the target username and the old username when searching through the log using a single title (e.g. a system with "User:Crat" renamed "User:X" to "User:Y" that would be included along with "User:Crat" renamed "User:Y" to "User:Z" were one simply to search for "User:Y")?
:| TelCoNaSpVe :|05:31, 26 January 2011 (UTC)
Image names appearing as errors because of commas
I've recently been uploading to Commons lots of images of buildings in historic districts, and to distinguish them, I typically give each file a name of "Street Address", "Building Name", "Historic District Name".jpg — notice the presence of the commas. Tonight, when I go to one of these images and click the little "en" tab at the top (I'm using Monobook; I don't know what the tab is called in Vector) to come here, I consistently get the "Bad title" message, and the URL displays oddly. For example,
File:First Street 1123, Bruner House, Vinegar Hill HD.jpg has a URL of
http://commons.wikimedia.org/wiki/File:First_Street_1123,_Bruner_House,_Vinegar_Hill_HD.jpg, but when I click the "en" tab, it goes to
http://en.wikipedia.org/wiki/File:First_Street_1123%252C_Bruner_House%252C_Vinegar_Hill_HD.jpg: for some reason, the commas have been converted into their
character entity references, and the system doesn't like having them in the URLs. Any idea what's happened, either here or at Commons? Just one day ago I didn't have this problem: the URL would come here and still have the commas instead of the %252C bits. The images display properly here if I replace each %252C with a comma. I'm running IE9, but I doubt that has any effect here.
Nyttend (
talk)
05:52, 26 January 2011 (UTC)
Installed MediaWiki version (live): 1.43.0-wmf.13 (8eaf4e5)
The developers are preparing to release dot-17 (
MediaWiki 1.17), but not yet according to the WP release id. That type of title issue is one of their efforts: to handle text after quotemarks in a URL, or reject other bad data, but not specifically for images, so perhaps someone else knows more. Watch for the MediaWiki version to switch to 1.17xxx (this month?), and perhaps the URL changes will allow the encoded commas then. -
Wikid7707:52, 26 January 2011 (UTC)
I think you may find that it is a issue with the global js/gadget which is creating the "en" tab is what is causing the issue to exist.
Peachey88(
T·C)10:49, 26 January 2011 (UTC)
Can I prevent someone creating subpages under my user page?
My guess is that anyone can simply create a subpage under my user page. And assuming that is correct then is there something I can do to stop the
toolserver from divulging any extra information when/if someone creates the subpages EditCounterOptIn.js and EditCounterGlobalOptIn.js?
How can I monitor for creation of those two subpages? Can I just put them into my watchlist even though they currently don't exist?
How can I monitor the meta: namespace to see if someone creates the pages there? AFAIK I've only ever used the en wikipedia, but somehow I have logons for some others also. I've no idea how they came about but it bothers me that these things happen without me realising what happened. Can I remove my logons from those other sites?
How can I monitor for creation of any new subpages with any other name? It seems the watchlist can't catch all possible future subpages, and having to check Special:PrefixIndex periodically is not really a nice solution.
*.cs and *.js pages in userspace are automatically protected by Mediawiki. They can only be edited / created by the corresponding user and/or administrators.
Dragons flight (
talk)
03:54, 27 January 2011 (UTC)
Okay, good, that basically answers all of 1 and 2 and the first part of 3. Just leaving the question of those other rogue logons and monitoring any other new subpages someone might decide to create.
HumphreyW (
talk)
04:07, 27 January 2011 (UTC)
To add to the answer to 2, you can watch non-existent pages. As for 3, when you log in to this Wiki and then visit another Wikimedia Wiki (such as meta), an account will automatically be created for you. See also
Special:Mergeaccount. You cannot remove those accounts (you can perhaps have them renamed and then avoid visiting the site again, but I honestly can't see much of a point in that). I don't think it's possible (question 4) to watch for that; you'll have to watch Prefixindex.
Ucucha04:18, 27 January 2011 (UTC)
With regard to subpages, take a look at my main userpage (it's in my sig) to see what I do; the short, surprising answer is that you can transclude
Special:Prefixindex just as if it were a template. That means you can list every page in your userspace if you want to, and look for random page creations you don't recognize. —
Gavia immer (
talk)04:19, 27 January 2011 (UTC)
Yeah, you're right, it doesn't seem to work for that. It will show new page creations, which I took to be what HumphreyW was concerned about. Obviously, pages that one knows about can simply be watchlisted. —
Gavia immer (
talk)05:07, 27 January 2011 (UTC)
I guess it's too late: I see you didn't actually talk about Recentchangeslinked. Of course, transcluding the prefixindex and then checking it is not much more efficient than just checking it.
Ucucha05:15, 27 January 2011 (UTC)
{| class="wikitable"
! This is a table header.
|-
|This text is "in" the table.
|}
This text is below the table.
See the example to the right. I know the table syntax used is incorrect, but wouldn't it make more sense for the parser to display this as if it were correct (i.e., with a pipe before the "This text is "in" the table." line)?
Ucucha01:22, 27 January 2011 (UTC)
re
Why discuss "incorrect syntax"? But alas, with the pipe (marked "x") it would be & look like:
This is a table header.
xThis text is "in" the table.
This text is below the table.
Code used:
{| class="wikitable"
! This is a table header.
|-
| xThis text is "in" the table.
|}
This text is below the table.
Yes, I just said that. The question is in what I wrote before; it's the sentence that ends with a question mark.
Ucucha01:41, 27 January 2011 (UTC)
re2
Ah, the question mark! Then the answer is: no. My expansion, if you allow me, is: no, it would not make more sense, because it is breaking the logic. So it would take sense away. The wikitable doc is quite clear about what happens with information, pipes and newlines in their positions. The parser is not weird, it's just doing that job of interpreting. -
DePiep (
talk)
01:58, 27 January 2011 (UTC)
How exactly does it make sense to put content that is inside a table above it? Why is the behavior here different from what a brower does when it sees <table><tr><th>Header</th></tr><tr>Text</tr></table> (the equivalent HTML syntax)?
Ucucha02:20, 27 January 2011 (UTC)
That is, in fact, what
HTML Tidy sees. And it tidies it by moving the misplaced text above the table. Which happens to match what Firefox does with it, BTW; other browsers may interpret the
tag soup differently.
Anomie⚔02:25, 27 January 2011 (UTC)
Yes, you're right; not sure what I did wrong when I tested that syntax in Firefox before. In that case, the behavior makes a little more sense to me.
Ucucha02:33, 27 January 2011 (UTC)
re3
You got a point. This is example code from
documantation:
{|
|+ The table's caption
|-
| row code goes here
|-
| row code goes here
|}
Which produces:
The table's caption
row code goes here
row code goes here
Only thing: the row code ... is pseudo code, i.e. not to be encoded literally. That leaves us with the "undefined" escape: the situation is undefined, so the outcome is undefined. The parser could state that text as 'unused' -- which is a feature. -
DePiep (
talk)
02:37, 27 January 2011 (UTC)
Not weird parser but weird neat feature: Try to imagine neat uses for the wiki-weird features, such as omitting the bar "|" on a column allows shoving text above the table, as the feature pre-table text. As another example, when a person cannot read "{{template|{{{{x}}}{{{y}}{{{z}}}{{{q|{{{qq}}}}}}}}}}}" then you know they are normal; when a person cannot read "{{{{{1}}}}}" then they have normal
dislexia. Think of Wikipedia as an exercize in spaznology: just how spastic can things be and yet the human mind can still get results with it. Of course, some day, parameters could be changed to allow outer square-brackets, "[{1}]" as easier to read when nested: {{mytemplate|[{1}]|[{size|[{siz|4}]}]}}. However, meanwhile, all the excessive nested "{{{{{}}}{{}}}}}" are proof of the human mind's ability to overcome adversity and make progress. Even then, we would remember the days when people seeing unreadable "{{{{}}}{{{}{}{{}}}}" was a test for normalcy. -
Wikid7714:26, 27 January 2011 (UTC)
Using a Logo from German Wikipedia (de.wikipedia.org) in English Wikipedia
Yes, you would need to upload a local copy. Download the image to your computer, then reupload it as a logo to en.wikipedia, and make sure to use the appropriate liscensing and fair use tags. --
Jayron3218:26, 27 January 2011 (UTC)
Out of interest, is the image archive for international Wikipedias hosted on the same servers as the English? I ask as it might be cost-saving to combine the image archives to one giant one for all languages. This sounds large but would save duplicate uploads across versions.
QuickHare (
talk)
01:10, 2 February 2011 (UTC)
Odd behavior from Reflist
At
Malcolm X#Footnotes, Reflist has been set as {{Reflist|colwidth=30em}} for a long time. On my wide-screen monitor, the article has always shown three columns of footnotes—until today, when it shows only two columns. The following section, "Works cited", uses {{Refbegin|colwidth=30em}} and it shows three columns. Any thoughts? —
Malik ShabazzTalk/Stalk00:54, 28 January 2011 (UTC)
Reflist was overhauled some two months ago in order to make plain references look the same as reflist. This means some recoding in the CSS for reflist as well. The problem seems to be that reflist is now wrapped in two nested divs, with the smaller font size set in the inner div, but the columns set to the outer div. Since the outer div has the original non-reduced font size, it causes the the calculated column width to be slightly wider then those using the refbegin templates, which delays triggering the 'next' column. Basically, to get three columns back, you need to decrease the column width (on my 1280px wide screen, it jumps to three columns at 27em). I did anticipate this, but didn't think the difference would be this big. Perhaps the inconsistency can be removed, but I need my thinking hat for that. —
Edokter •
Talk — 01:18, 28 January 2011 (UTC)
I use Firefox, and I have no similar issues with any web site but Wikipedia. This has been coming on for months, but used to be only once in a while.
Over the Jan 15-17 weekend, I noticed Wikipedia was really slow, no matter the page or action. You could take little mini-naps waiting. And the Timed Out message with the "try again" option kept coming up. Wikipedia's speed recovered after the weekend, but the timed out messages have gotten worse. Sometimes it's every other click of the mouse, or every third or fourth. Way too much. It's only Wikipedia. My cache was cleared.
Maile66 (
talk)
14:23, 21 January 2011 (UTC)
Wikipedia is often slow, even down sometimes. It happens to all of us. If you're only having the problem with Wikipedia and have cleared your cache etc., then that just simply makes it more likely that the website and not your browser, Internet connection, etc. is the problem. Gary King(
talk ·
scripts)05:49, 22 January 2011 (UTC)
One thing that can sometimes make a difference (here in the UK anyway) is the server that you log in through. In the normal course of things the secure server is very slightly slower than the normal server, but if the normal server goes dead slow you may find that the secure server is running at its usual speed.
When you go for the "
Log in / create account" link at upper right, you get to the login screen for the normal server. If you don't actually log in there but read down the page, you'll see "Consider logging in on the
secure server. That takes you to another login screen, where you can log in using your normal ID and password. --
Redrose64 (
talk)
11:51, 22 January 2011 (UTC)
This problem is probably being caused by difficulties connecting to bits.wikimedia.org, responsible for serving various site scripts, due to routing issues – are you in the UK, by any chance? (
[9]) This problem is alleviated by using the secure server since those scripts etc. that are usually served by bits.wikimedia.org in the non-secure environment are served directly from the secure server.
217.137.122.112 (
talk)
13:09, 22 January 2011 (UTC)
As a UK user, I've experienced the same problem, and tried the secure server. However, on every WP page I go to, I have to dismiss the MS "do you only want to view the secure content" pop-up. Can this be overcome just for WP? (I'd like the pop-up to remain for other sites). Arjayay (
talk)
16:36, 22 January 2011 (UTC)
I'm using secure server and have the same issues as
Maile66. I'm in the UK and the problem does get worse as the day goes on(likely as more US traffic). At times I get a 50% fail rate to posting. I rarely do a 'show preview' anymore because the fail rate is high. I just submit and if it fails.....after about 15 seconds, hit back and submit again. I've only double posted once. Still it's not a fun way to edit. Regards,
SunCreator(
talk)16:54, 22 January 2011 (UTC)
I am experiencing the exact same issue; always thought it was something to do with my network... I am logging in from Sri Lanka.
Rehman02:59, 23 January 2011 (UTC)
Wikipedia was just offline to me for a period of about 7 minutes. I guess the time of day means lots of traffic in America now? Anyway, thought it worth noting the downtime. Regards,
SunCreator(
talk)03:32, 23 January 2011 (UTC)
"The server at en.wikipedia.org is taking too long to respond." is part of the error message. What someone mentioned as "50% fail rate" is sometimes the norm, regardless of weekend/weekday or time of day or night. Frustrating.
Maile66 (
talk)
03:38, 23 January 2011 (UTC)
More bugaboos
Reference my earlier post about slowness and timing out. I use Firefox. It's still doing that, even worse. But now it has an added kick. It's logging me out. That nonsense just started about half an hour ago, but thought it was worth mentioning, in case anyone else is experiencing Wikipedia's mood swings.
Maile66 (
talk)
01:23, 26 January 2011 (UTC)
I also noticed terribly slow response time, the past few days, where perhaps 1 in 10 edit-previews would time-out (So, I made mirror edits in a text-editor window to avoid losing all edits due to time-out). The developers might have restricted user resources, for a few days, in preparation for the dot-17 release (
MediaWiki 1.17) which has numerous tiny changes to slightly adjust internal data structures, and totally remove some sections of deprecated code (rather than just leave it, who cares). Some group has created a children-will-like-it upload screen for Wikimedia Commons, with a cartoon explaining the process (I'll bet the upload-cartoon doesn't cover analyzing a fair-use rationale to comply with changes to international
copyright law!), just the basics for users new to Commons upload of "free" images. They are improving secure login to allow returning to non-secure browser windows (but any celebrities had best continue editing with secure-post mode). A lot of clerical changes: it's mostly like re-arranging deck chairs (or coal shovels) on the Titanic, rather than building new ships to solve major concerns. Bigger changes usually happen to templates or policies than to the MediaWiki features. Remember, last year we removed the
WP:CANVAS censorship which warned people not to tell others about new articles: now if you write a new article and tell 9 people, you can't be accused of conspiracy to provide too much valuable information. I think we will need to have a 2nd Wikipedia, which can risk major new features without the fear that the ONE Wikipedia will die if a new feature has a severe bug. -
Wikid7706:49, 26 January 2011 (UTC)
Aha. Yes, I experienced being logged out too. It seemed to me that it happens after leaving Safari (I use FF and Saf), with (or without?) FF open & logged in.
Re my report below about being logged out, I use Firefox, but do have IE open and minimised in order to run AWB, which I do from my NSH002 account. --
NSH001 (
talk)
11:58, 26 January 2011 (UTC)
I had similar issues with firefox, only it was 100% logging me out. Switching to rekonq solved the problem. Not sure what was going on.
Sailsbystars (
talk)
14:56, 26 January 2011 (UTC)
Connection timed out
I've been getting an increasing number of "connection timed out" messages in the last few days, and just wondered whether anyone else had noticed the same thing.
SlimVirginTALK|CONTRIBS05:26, 26 January 2011 (UTC)
Yes, in various locations, for over 5 days, while we ran other websites quickly. Sometimes, 1-in-10 edit previews would time-out (leading to, "I had better copy this edit-buffer to an external text editor or risk losing all"). Currently, the
Arabic Wikipedia is as slow as a tourist walking through the deep sand in the
Sahara. See concerns above:
I suspected the developers are testing new MediaWiki 1.17 on some servers, but maybe not. I didn't see any great, giant new features in 1.17, but all computer-math is showing 15/16 digit precision, rather than some having 12/13 digits, a few weeks ago (so that indicates system upgrades). I am working on essays to improve performance, such as making {Cite web} & {Cite book} run 4x times faster & smaller, which are 95% the wikitext of scholar's articles. Also, we need to avoid those large bottom navboxes which are causing "50,000" articles to reformat all the time. Over 1.1 million navboxes are typically doubling the size of "every" article on Wikpedia (it should be running twice as slow for everyone). --
Wikid7707:31, 26 January 2011 (UTC)
I don't believe this is the case: although some servers may have been taken out of rotation for other reasons, testing 1.17 is unlikely to be one of them. 1.17 is still in code review.
Happy‑
melon15:22, 26 January 2011 (UTC)
Thanks, Wikid, I was about to contact my ISP. I know that citation templates slow things down (I wish we could find a way to stop encouraging people to use them), but which are the navboxes that are causing the problem?
SlimVirginTALK|CONTRIBS23:56, 26 January 2011 (UTC)
I have also had an unusual number of connection time outs, 'server not available' messages, and general slowness of response, over the same recent periods as mentioned by the posters above. These problems are only related to WMF sites. I connect from Thailand using MacOS with FFox and Safari through good DSL (at home and in the office) and
EGPRS (EDGE) when on the move (and sometimes a smart phone).
Kudpung (
talk)
01:48, 27 January 2011 (UTC)
being logged out
This has happened to me 2 or 3 times this morning. Very annoying. Unlike others above, Wikipedia doesn't seem any slower than usual. Just thought I'd mention it here, in case others are having the same problem. --
NSH001 (
talk)
11:30, 26 January 2011 (UTC)
Saying "It doesn't work" is pretty useless in trying to get this fixed. At a minimum, your location, your ISP, connection details, etc would be helpful to the people who can even begin to look at this.
QTC01:20, 27 January 2011 (UTC)
Q, if the problem were my ISP/connection/location, or similar, this would not be the place to discuss it. I posted the message to discover if the problem might be with the servers or their software. Fortunately the problem has not recurred since I reported it above. --
NSH001 (
talk)
08:35, 27 January 2011 (UTC)
It happens to me too with increasing frequency. Fortunately I notice before pressing the 'save page' button, because the skin goes back to the vector default, whereas I, wisely, use monobook which is far more stable and has a much more practical interface.
Kudpung (
talk)
01:48, 27 January 2011 (UTC)
It looks like a rogue server. Can you identify the server which showed you as logged out? (View source, and look at the "Served by" at the very bottom)
Platonides (
talk)
23:37, 28 January 2011 (UTC)
Formatting numeric templates on Arabic WP
I do not speak Arabic, so I am wondering if anyone here knows of a CSS style that forces left-to-right text on the Arabic Wikipedia (or other right-to-left languages). I have ported the
Template:Convert subtemplate for temperature ranges, {{Convert/Dual/LoffAoffDxSoffT}}, but the output, usually of the form "10–20 °C (50–68 °F)" is being scrambled in Arabic due to Latin letter "F" within parentheses "( )". Currently, I am hiding the letter "t" (as font color=white) at the start/end to force Latin-text alignment, as left-to-right:
Hiding "t" at start/end: t10–20 °C (50–68 °F)t.
However, is there a browser style directive (or some template, because {nowrap} still scrambles parentheses), which does not introduce hidden text, but can directly format text as left-to-right in the Arabic Wikipedia? There's no hurry on this, because the 2 hidden t's work okay for now. -
Wikid7721:10, 25 January 2011 (UTC)
In depth, Unicode Bidi Algorithm covers this for most browsers. Hardcoding, you could use uniciode bidi-formatting 'control' characters
Unicode_character_property#Bidirectional_writing. The special characters LRE, RLE, LRO, RLO, PDF ('pop') can control/overrule the F-property 'latin'. They are invisible, but just take a look how they work. Amazing, innit? 10–20 °C (50–68 °F). Need more?-
DePiep (
talk)
22:29, 25 January 2011 (UTC)
Adding: the algorithm UAX#9 is
here, but is not easy reading. Some 28 rules must be applied to a character string (in memory), using the 19 bidi-types in the above linked table, while interpreting paragraph breaks, linebreaks and looking at neighboring character properties. A group of rules is specific for LtoR numbers in RtoL scripts, but as you write an "F" or so might break the logic. (detail: in LRO and RLE &tc, O=override, E=embed, and 'Pop Directional Format' (pop) is the general closing bidi-bracket for the latest (closest, deepest nested) LRx/RLx bidi-bracket). -
DePiep (
talk)
22:47, 25 January 2011 (UTC)
So would this solve it: the number & dimension sequence must be LtoR in Arab. Simple would be: the template outputs: LRO{{#ifeq:|on|<span style="display:none">{{padleft:{{FORMATNUM:10|R}}|16|0}}{{#ifeq:{{#expr:(20)*0}}|0|{{padleft:{{FORMATNUM:20|R}}|16|0}}}}</span>}}{{convert/{{#if:1|-}}|{{FORMATNUM:10|R}}|{{#ifeq:{{#expr:20*0}}|0|0}}|20|C|F||r={{#ifeq:{{{sp}}}|us|er|re}}|d=LoffAoffDbSoff|s=}}PDF. The ()-bracket are kept too. Will be more difficult when there are Arabic words etc in: say "From 10 to 20." would have to translate into: "RLOFrom LRE10PDF to LRE20PDF.PDF", so: "From 10 to 20.". (using Embedding within an Override. Do NOT change the sequence of the characters: in memory they will stay LtoR always. The browser will reorder them as defined when sending them to the display) -
DePiep (
talk)
23:00, 25 January 2011 (UTC)
There was also this question: is there a browser style directive (or some template [...]), which does not introduce hidden text, but can directly format text as left-to-right in the Arabic Wikipedia. -
DePiep (
talk)
01:58, 26 January 2011 (UTC)
Thank you both. In this case, the unicode characters worked to override each temperature word and retain the flow of the surrounding Arabic text: using "‮F°‬" reversed the text as "°F" without disturbing the nearby Arabic text. They have a special symbol for "F" Fahrenheit, but at least, now,
Arabic Wikipedia has the temperature-range conversions which they have been wanting for nearly a year now. Considering the mind-boggling differences between the Arabic Wikipedia and enwiki, this 1-day one-year solution was as you said, "Amazing" ("Amazing"). Thanks again. --
Wikid7705:25, 26 January 2011 (UTC)
You might want to consider using templates to insert those characters. Allows easy changes, and {{LRO}} is probably easier to read than ‮ and friends. (perhaps we should add a magicword/tag or something to MediaWiki itself ? —
TheDJ (
talk •
contribs)
11:41, 28 January 2011 (UTC)
centralNotice inconsistency
… .notice-collapsed-wrapper-stew2011-noms {
margin: 2px auto 0; …
Because of this margin, even with a cookie saved to hide the notices at the top, pages "randomly" jump around 2px every now and then (when the "Candidates for the 2011 steward elections…" notice is chosen). This is not a big problem, in fact I couldn't even reproduce it on Windows, but it is annoying, and I do find it unnecessary. That particular notice has a significant amount of custom CSS inserted and I don't really understand why. Shouldn't these notices be more or less uniform? At least the ones that are just text anyways? ¦
Reisio (
talk)
07:53, 26 January 2011 (UTC)
This should be fixed once the cache clears (5 minutes). Had to remove a superfluous div that was not included in the collapse definition.
Kaldari (
talk)
22:39, 28 January 2011 (UTC)
Keymanweb alternate language input facility safe?
Hi, Can someone help me with any feedback regarding the
User:Keymanweb/Keymanweb virtual keyboard? The monobook.js page suggests that I check out the Village Pump if in doubt so here I am. By the way, I did a search for keymanweb but did not see many results. My main wory is that the user page and user talk page for keymanweb did not help at all in veryfing the safety/robustness of this tool. Any help appreciated.-
Deepraj |
Talk20:20, 27 January 2011 (UTC)
Its from the commercial website www.keymanweb.com by Tavultesoft. It loads external JS and has the potential for keylogging, but seems wider used than just here. If you want to be sure, you need to do a full review of the code (i don't have the time). How much do you trust the company ? :D —
TheDJ (
talk •
contribs)
12:24, 28 January 2011 (UTC)
You need to add the parameters group1=whatever, group2=whatever, to have each sub navbox show up. I don't really see the point of that #if parser, so you might as well remove the {{#if:{{{group1<includeonly>|</includeonly>}}}{{{sect1|}}}{{{section1|}}} and related for group2. /
ƒETCHCOMMS/04:49, 28 January 2011 (UTC)
Perfect! I'll have to take a look at what you did to fix. I really appreciate it, the double collapsible w/columns was a bit ambitious for me. Looks great, thanks again to both for taking a look.
The Interior(Talk)17:33, 28 January 2011 (UTC)
The vandal duplicated most of the page content inside a comment in
this edit. Whoever tried to remove it manually didn't notice that, and edited the part inside the comment instead of the text that was actually being rendered. I think I reverted it to the last good revision for you, but someone should probably read it over to make sure more crap didn't sneak in earlier.
Anomie⚔12:14, 28 January 2011 (UTC)
singing with tildes doesn't work
I have recently changed my username, and as you will probably notice as you read on, signing with four tildes doesn't seem to work. I still get a notification on my talk page that i have to sign my posts. any ideas? Rafytalk— Preceding
unsigned comment added by
Rafy (
talk •
contribs)
16:17, 28 January 2011 (UTC)
In order for SineBot to detect your signature as a proper signature, it has to contain a link to your user or talk page. The bot has no way of knowing that you have changed your username, you should fix the links to point to your current user page.—
EmilJ.16:37, 28 January 2011 (UTC)
Are you signing with four tildes? Three tildes only produces your signature without a timestamp. Your signature does not contain a timestamp, which is causing signbot to think you're not signing your posts. —
Edokter •
Talk — 16:36, 28 January 2011 (UTC)
When I'm editing a page, I'll often type in my edit summary and hit the return key to submit. However, every once in a while I'll do that and it will trigger the 'show preview' button rather than the 'submit' button. I can't figure out what causes that difference. does keymapping to buttons change via some undocumented feature, or am I miskeying, or is this just some reasonably frequent but random fluke? Not a big issue (it's just confusing), but if there's an intentional way to trigger a show preview rather than a submit that would be useful information. --
Ludwigs216:18, 28 January 2011 (UTC)
The Save Page button should always be activated when hitting Enter on the Edit Summary because it appears first in the HTML, even though there are three submit buttons on the edit page (Save Page, Show Preview, Show Changes). I've never had the problem that you are describing. Has this been happening for a long time? Gary King(
talk ·
scripts)21:47, 28 January 2011 (UTC)
The Tab key can move between buttons but in my browser it takes several Tabs before Show preview or Show changes becomes active.
PrimeHunter (
talk)
00:56, 29 January 2011 (UTC)
@ Gary: no, it happens occasionally (I'd say once or thrice a month). It could be a Safari-dependent thing - a semi-random rendering mistake, a weird undocumented 'feature' that I'm somehow triggering accidentally (and yeah, it's not a tab key thing - that would take it to the 'minor edit' checkbox first). Unfortunately I can't replicate it on demand; I was just hoping someone else had a clue about it. --
Ludwigs203:03, 29 January 2011 (UTC)
All I can say is it's probably something on your end, as something has got to seriously be out of whack for the page to be served that way to your browser. Meaning, even if your browser rendered the page strangely, the order of the submit buttons shouldn't change. For instance, maybe it's a Firefox extension if you're using that browser, etc. Gary King(
talk ·
scripts)04:40, 29 January 2011 (UTC)
Talk: Cherokee is a lengthy talk page with many sections, but there is no autogenerated table of contents at the top. The archive pages for this talk page do have autogenerated tables of contents (and I do see these tables on other long talk pages, so I assume it's not something in my preference settings). I've looked at the code and I don't see anything obvious to account for this. I suppose a solution would be simply to add manually the code for a table of contents, but I would like to understand this function better, so I would appreciate it if someone could explain why the autogenerating function isn't working there. Thanks,--
Arxiloxos (
talk)
17:00, 28 January 2011 (UTC)
It has to do with the deprecated assessment subpage, which I've just deleted after pasting it onto the talk page. –
xenotalk17:34, 28 January 2011 (UTC)
Spaces in external links
How can I encode spaces in external links? To explain: I would like to let users write {{
gbmappingsmall|NH 714 684}} so that it displays prettily: as NH 714 684. But unfortunately it breaks down totally: NH 714 684. I need a function that will remove each space or convert each space to %20 or _ when forming the external link. Does such a function exist? —
RHaworth(
talk·contribs)22:37, 28 January 2011 (UTC)
Urlencode, I guess: NH+714+684. You might want NH%20714%20684 instead, which replaces spaces with %20 instead of +.
Ucucha23:13, 28 January 2011 (UTC)
Many thanks - just what I wanted. Yes, the WIKI and PATH options seem to have been dropped. As Our Ford would have said "you can have any separator you like as long as it is + But {{anchorencode}} is available to give NH_714_684 providing you don't need its other translations. —
RHaworth(
talk·contribs)05:36, 29 January 2011 (UTC)
Numbers often 3-digit but what length are strings
What is probably the most-common string length in English Wikipedia usage, such as with counting (by {{
Template:Str_len|abcdef}}=6)? In designing shorter ways to handle character strings (and numeric strings), I have found that, by far, the most common numbers in conversions are 3-digit integers, such as 115 in {{convert|115|km|mi|0}} → 115 kilometres (71 mi). Perhaps 70% of conversions are 3-digit integers, then 3-digit decimals, with 2-digit integers less common (scanned by wiki-searching for Convert option "abbr=on"). However, I can't yet think of an easy way to scan through uses of {str_len} to find the typical size of character strings counted on English Wikipedia. The
WP:Database_reports list {Str_len} as used 480,510 times. However, many uses are buried in the inner reaches of complex templates, so I dread sampling article edits from the nebulous
Special:WhatLinksHere/Template:Str_len, where an article's text has nothing, but uses an infobox layer invoking {str_len} deeply hidden within. Anyway, I guess I could edit through several many pages in that list. Does anyone know of any count data, already gathered about this issue, or typical infobox string codes? There's no hurry on this. Just curious if the typical length is 3 or 4 or 7 (etc.). -
Wikid7708:20, 28 January 2011 (UTC)
Average string length is italicized article-title length: 19? Okay, it appears the average string length for {str_len} is the average length of italicized article titles: 19 characters (based on a sample of 10,000 titles). The titles are those "353,026" pages using
Template:Italic_title to italicize films, books, albums (etc.), up to the left-paren. "(" suffix. So, I think I found Template:Italic_title is the "elephant in the room" of string usage, using {{
Str_find}} 4 times for every title with "(" and hence, using {str_len} 5 times for every such title. Because {Str_find} uses 99 {str_left} (with {padleft}) for every string searched, then {Italic_title} is using 450 calls to {padleft} for every title containing left paren "(" in the title. {Italic_title} always invokes {str_find} 1 or more times, so that's at least 35,303,000 calls to {padleft}, plus 350 more uses of {padleft} for each italiziced name with "(" in the title, probably nearly 30 million. So, {Italic title} is using more than 64 million calls to padleft. Based on a sample of 10,000 italicized titles, where 2336 titles had parentheses "()", the indication is that 82,500 articles using {Italic title} have "()". Hence, the processing for "()" adds nearly 82,500 x 350 = 28,875,000 more calls to {padleft}. I have mentioned this under
Template_talk:Italic title, for possibly improving operation, long term. More later. -
Wikid77 20:18, 28 January 2011, revised
07:48, 29 January 2011 (UTC)
Signature Issues
Hi, i'm having trouble with my signature as it's too big to fit into the box and i haven't even finished.
What i have so far is "
Jenova" but it should be in bold and have a black background around just the "Jenova" part with 20 just in black at the end.
Can anyone explain how to do this?
Thanks
09:24, 28 January 2011 (UTC)— Preceding
unsigned comment added by
Jenova20 (
talk •
contribs)
If you use <u> tags you can save perhaps (4 − 1) × 2 × 6 = 36 bytes and also make the underlining color match the text color instead of being a continuous red or blue bar. ―
cobaltcigs16:12, 28 January 2011 (UTC)
In particular, the yellow "n" is very difficult to see against the background colour of this page (the orange "e" isn't much better). The background is a pale greenish-blue here, but other talk pages differ: you shouldn't make assumptions about the background colour of any page. You'll notice that in my own signature, where a colour is set the background is also set (i.e. <span style="color:#d30000; background:#ffeeee">Red</span>), and there is a high degree of dark/light contrast between the two forced colours. --
Redrose64 (
talk)
17:50, 28 January 2011 (UTC)
Yellow could be "color=gold" then use "lime" with color=black on "20", set background:#999 for gray, plus "<b>" for bold, as:
Jenova20, with quotemarks removed in the wikitext: [[User:Jenova20|<b><span style="background:#999"><span style="color:red">J</span><span style="color:orange">e</span><span style="color:gold">n</span><span style="color:lime">o</span><span style="color:blue">v</span><span style="color:purple">a</span><span style="color:black">20</span></span></b>]]. I don't think quotation marks are needed with the one-word colors. The Firefox browser doesn't use color=0, so put color=black for "20", but Firefox also allows omitting each end-tag </span> to shorten the signature. -
Wikid77 08:27, 29 January 2011, revised
11:10, 29 January 2011 (UTC)
Maybe the n is supposed to be so faint that other editors have to guess which letter it is? I know what I would guess, and I believe it's not in line with the username policy.
HansAdler10:43, 29 January 2011 (UTC)
Quotation marks are indeed not required with one-word colours when used in the <font>...</font> element. They are only required in HTML element attributes when the attribute data contains any character other than letters, digits, period or minus. So, <font color=blue> does not require quotes, but <font color="alice blue"> and <span style="color:blue"> both do (because of the space in the former, and the colon in the latter).
Firefox may permit the omission of the closing </font> in certain circumstances (see
below) but that doesn't mean that other browsers do too. You cannot assume that all target users have the same browser, nor even that some browser quirks are universal. It's not good HTML to leave any tag unclosed, with very few exceptions: within the <body>...</body> section, the only tags which ordinarily permit non-closure that I know of are the inherently self-closing tags (<br>, <hr> and <img>); certain block elements (such as <p>) and certain elements which are intended only for use within blocks (such as <dd>, <dt>, <li>, <td>, <th> and <tr>). I know of no inline elements (and both <font>...</font> and <span>...</span> are inline elements) for which non-closure is valid HTML. --
Redrose64 (
talk)
16:42, 29 January 2011 (UTC)
Gadget: Black Background
I was curious as to whether a gadget similar to the "Use a black background with green text on the Monobook skin" could be created for the vector skin. With the monobook gadget enabled and using the vector skin, it seems mostly alright other than the main header, which if I recall can be changed through CSS settings. Sorry if this was posted in the wrong section, and thanks for your time -
NeilHynes -
TalkEdits06:34, 29 January 2011 (UTC)
Oh yes, forgot to add this as well: The boxes which constitute the front page seem to be hard coded as being white rather than transparent, making it pretty hard on the eyes while this gadget is enabled. Is there any way to fix this as well?
NeilHynes -
TalkEdits06:51, 29 January 2011 (UTC)
It might be a good idea to adjust the brightness of whichever screen you are using, on a day/night basis, to reduce eyestrain. Perhaps lower the brightness during dark hours. Also, avoid facing bright windows behind a computer, as focusing on computer screens might cause more intense viewing than normal reading of a printed book. There is a lot of medical research about eyestrain with computer screens. -
Wikid7710:15, 29 January 2011 (UTC)
Page protected from moving personal stub page to
I have recently created a personal stub page, currently located at
User:Trident13/Boticca which I wish to move to the page titled
Boticca, but am told by the system that I can not move it to that mainpage as the page
Boticca is presently protected. How do I find out the reasons for the protection, and would be grateful for advice on next steps. Rgds, -
Trident13 (
talk)
22:45, 29 January 2011 (UTC)
I know {{Deletable image-caption}} has already been converted to use dated category system to enable tracking and cleaning up of stale tags. I took a whack at trying to use the same dated system with this template, however the way the parser function #time operates it does not accept the current format that is passed by the log parameter. I would appreciate a way to resolve this issue, as I often find really stale tags that need removed, and these dated categories make it soo much easier to sort though the mess.
ΔT The only constant03:32, 30 January 2011 (UTC)
Using #switch: 'p' does not match 'p'
(Few days ago I did a neighbouring(?) report on this
font-related class="IPA" disrupts a #switch:). {{
IPAsym}} uses #switch:. Input is a IPA-character, output is the corresponding IPA article. But when entering the character using numeric reference like &#x0070;, it fails to find it. Example with the regular 'p':
According to the template's documentation, you can only enter IPA characters or IPA numbers; HTML entities do not work. —
Edokter •
Talk — 15:03, 26 January 2011 (UTC)
"p" is an "IPA character". But Unicode does not know or handle an "IPA character" differently: to you, me & Unicode it is just a regular "p". The point is: when entered by typing (or copypasting), it is recognized by "#switch:", when entered by p, it is not recognised. "p" is an easy example, non-US-keyboard chars (like ð) have the same behaviour/problem. -
DePiep (
talk)
00:26, 27 January 2011 (UTC)
This is expected behaviour, as p and p are not the same. Consider instead |, the encoded form of the pipe character... you wouldn't want that to be treated in the same way as its raw counterpart, would you? :DHappy‑
melon15:13, 26 January 2011 (UTC)
No. I expect p (rendered) and p to be the same. It is the same character. Another expected the same is p. Maybe you mean to say explainable different? -
DePiep (
talk)
00:26, 27 January 2011 (UTC)
re the pipe symbol & encoding: what you describe is the my Q other way around. There is nothing I may "want" in behaviour. We "expect" in behaviour. When I encounter pipe-problems, sometimes I can handle them (e.g. by using the template {{!}}). But that is understandable logic to me by now (I use that in #if: table-building). Saying "(A) =/= A is expected", as you do here, is not logic to me. -
DePiep (
talk)
00:42, 27 January 2011 (UTC)
HTML entities are decoded by the browser. Mediawiki never knows what the encoding means. This behavior is desirable in some cases, and is unlikely to change.
Dragons flight (
talk)
00:46, 27 January 2011 (UTC)
Nothing wrong with that. The option stays. In the current ~150 calls of {{
infobox IPA base}}, all input does so (either by character, or by "IPA number" -- none by hex). The question rose, when I tried to use the hex or decimal editors input (in a regular article page) to a valid IPAsym output article name. This would reduce the essential main input to two out of three: IPA-number -- symbol -- Unicode-decimal-value. E.g. the font does display well through Unicode-HTML-reference (but that same code does not return the IPAsym article name, as is the issue here). After this thread, the solution must be: editor expected to enter all three out of three. That will do, we'll just have to instruct (document for ) the editor. -
DePiep (
talk)
02:03, 28 January 2011 (UTC)
Using a #switch, to check separate characters, would need both values in each bar-branch: such as "p | p" to match either the literal "p" or its hex-code. A #switch can have over 1,400 values in the branches. Long term, I think we need to change the NewPP preprocessor to allow metacharacters, such as the UNIX "\b" or "\t" for a space or a tab, at least to treat some entities such as " " as the 1-character value   with string length of 1. Currently, using " " is a 5-character string (&-#-3-2-;), and " " is a 6-character string until displayed. UNIX has used the backslash "\" as the
escape character to denote text metacharacters because file path names used the forward slash ("/"). Using that method, prepending a backslash could denote "\p" as the 1 character "p" rather than the current literal 9 characters. Worrying about performance, I would expect the preprocessor to just place 1 character in the text, rather than 88-character marker text which "<nowiki>xx</nowiki>" currently generates. -
Wikid7713:36, 27 January 2011 (UTC)
Wikid: way beyond my scope, but I get the idea. For now, I am OK with the explanation by User:Dragons flight: "rendered in the browser", i.e. after the template is solved. That's it. -
DePiep (
talk)
02:03, 28 January 2011 (UTC)
It would be more controllable to introduce a parser function {{#decodeentities:...}} which fully-unescaped references. " " is a five-character string, in the HTML source as well as in wikitext. There are some situations where you do want to be able to evaluate it as one character, and some situations where you wouldn't. Introducing new behaviour for previously-unmagic notation is dangerous, and would require a huge amount of testing on old and current revisions. Implementing it through a parser function or similar provides the same functionality without impacting on existing wikitext.
Happy‑
melon13:28, 30 January 2011 (UTC)
Question about a possible change to Twinkle
I delete a fair number articles proposed for speedy deletion. I almost always check to see that the original editor of the article is notified (I don't always check attacks, vandalism or redirects.) In most cases the editor has been notified, probably because many taggers use
Twinkle and that generally does a notification automatically. However, some new editors getting into CSD work don't yet know about Twinkle, and occasionally Twinkle is claimed to miss one.
How hard would it be to make a minor modification to Twinkle, so that when it adds the CSD template to the article, then adds a notification to the original editor, it could also add a note to the CSD template indicating that "
user:foo received a notification". Then I could skip checking the 90% or so that do notify, concentrate on determining whether the CSD is warranted, and only track down notification issues in those cases where it hasn't happened?
(I know this ultimately belongs at Proposals, but I wanted to see if there were technical hurdles I've missed, or if I should be talking to
AzaToth or someone else.)--SPhilbrickT02:31, 30 January 2011 (UTC)
I can open any in-text link in a new browser tab (IE8), but when right clicking on an article tab such as "Discussion" or "Edit" I do not get this option, but rather the option set associated with right clicking on background or regular text. I can not be sure, as I use other Wikis, and may be confused, but I think this is new, and that I used to be able to open these in a new tab.
Peter (Southwood)(talk):
08:16, 30 January 2011 (UTC)
It's a known IE8 bug. Right-click right next to the text to get the proper menu, or "middle"-click (press the wheelbutton) the link to open it in a new tab immediately. —
Edokter (
talk) — 10:16, 30 January 2011 (UTC)
Oops, you're right, it's already on Commons. I've seen this error previously and as far as I recall you could make it go away by changing the file name (removing spaces or something). Somehow that did not seem to do the trick here, though. --
Morn (
talk)
12:41, 30 January 2011 (UTC)
Signature Length
Why can't the signature length box in "My preferences" be longer, so users can do more markups and such. CTJF8302:45, 29 January 2011 (UTC)
How can I propose to make the available space sightly longer? I just wanna add and link "chat" on the end of my current one. CTJF8303:14, 29 January 2011 (UTC)
I think you can omit quotemarks on word colors and every closing "</font>" when inside a span-tag (<span>), plus use color=gold (shorter than "yellow"), as:
CTJF83, now with the shorter wikitext: [[User:Ctjf83|<span style="color:red>C<span style="color:#ff6600">T<span style="color:gold>J</span><span style="color:green>F<span style="color:#0000ff">8<span style="color:#6600cc">3]]. You might need one closing "/font" but not all of them. Each font color changes the prior color, "<font color=blue>" as: "This is blue text".
HTML was NOT originally designed to demand a matching end-tag "</font>" for every font tag, but I don't know what the HTML fascists are planning. The
World Wide Web was designed by physicists, not by computer scientists, so it was peculiar that way: it allows "<center>" but not margin "<left>" or "<right>" which the center is between (so clueless). Now that computer experts are involved, it is still backward, and they should allow word "hue" (to avoid spelling bias against "colour"), plus denote non-nested tags perhaps as "<#" to use "<#font hue=red>" where the pound-sign "#" would indicate to omit the closing "/font" tag. Anyway, it seems like we need a WP essay about shortening signature text. -
Wikid7709:41, 29 January 2011 (UTC)
All of Wikipedia seems to fail the
Web Content Accessibility Guidelines, based on the brightness formula: ((Red value * 299) + (Green value * 587) + (Blue value * 114)) / 1000. The brightness of tan is: 191.352 compared to the brightness of red: 76, a difference of 115, not the required 125, so redlinks would likely be considered invalid brightness in note boxes. The focus needs to be on "almost passing" the guideline, as an acceptable approach. -
Wikid7712:34, 29 January 2011 (UTC)
We aren't discussing "all of Wikipedia", just this signature. If there are problems in other areas, then those need to be discussed and fixed as needed. ---—
Gadget850 (Ed)talk17:20, 29 January 2011 (UTC)
<font>...</font> is a deprecated element, use <span>...</span> instead, or <b>...</b> if you want to save some characters. You do need the tags to be properly nested, as Graham says, you're causing no end of problems for users of screendreaders etc otherwise. Wikid, the
HTML 3.2 specification, which introduced the <font> tag, clearly indicates that it "Requires start and end tags"; I don't know where you got your HTML-history from, but it's not accurate. The most recent version of HTML, HTML5, has moved to reduce the number of tags which need to be matched, loosening the restriction for various block elements like <p>, but that logic cannot be applied to inline elements.
The signature string is stored in a 255-byte database field, and increasing the size would require a database schema change, which I very much doubt will happen even if you ask for it.
Happy‑
melon11:11, 29 January 2011 (UTC)
I agree that HTML5 still has major design problems, but most browsers seem to keep working anyway. Hopefully, those screenreaders will be fixed to match the real world, and accept changing a font without using the end-tag "</font>" as has been done for years. If the font is blue, and black is needed, just put "<font color=black>" and keep going. I realize the omission of end-tags thwarts the use of nested font styles, but it needs to be expected, and so span tags limit the font scope. The word "
e-mail" has been misspelled as "email" in over 90% of webpages, so years ago, Google started matching either spelling as being the same word. It would make sense for all dictionaries to formally define "
email" as a valid alternate spelling. If most people set their screenreaders to announce font colors, then perhaps let users know that multi-color signatures will be cumbersome to hear. When a committee designs an unworkable language feature, it typically has a real-world patch to become usable, so expect font settings to continue in that manner. -
Wikid7712:34, 29 January 2011 (UTC)
Can I ask whether you've tried reading this page with the greenscreen gadget enabled? If not, please do so: go to your preferences and check the box labeled ""Use a black background with green text on the Monobook skin"". Then come back and read your comment above, and perhaps you'll understand what we mean by lack of accessibility. Black text is not part of the MediaWiki style, it is a browser default. Adding <font> is not resetting to the browser default, it is imposing black text, whether or not that's what users want, and regardless of whether that's even accessible to users.
Happy‑
melon13:14, 29 January 2011 (UTC)
This is an encyclopedia. I'm sorry, but if you can't get a short enough signature yourself, you're probably obsessing over it too much. Let's keep building an encyclopedia! (says the user who spent a good hour trying to design his own sig. :P) /
ƒETCHCOMMS/05:37, 1 February 2011 (UTC)
Having trouble editing/unintended vandalism
OK, so I was working on the
2010-11 Australian region cyclone season, and just adding the dissipation date for 06U (revision on 00:22 4 January), and although I had gotten the changes I wanted, I somehow exposed some formatting in the line of the table for TC Tasha. This was due to an infinite loop. What I did (if I remember correctly) was that I was submitting it, and it got stuck in the infinite loop. So, I either closed the computer (it was a laptop) or hit the Back button on the browser (I don't remember which), and then the formatting came out all screwy. I am using Safari version 5.0.3 on a Mac OSX Version 10.6.6 with a 2.4 GHz Intel Core 2 Duo processor. (The loop occurs even when I am editing small sections of a larger page, though the loop does not always happen, even when I am editing larger pages.)
No, there are no error messages, the screen is just frozen, but otherwise works normally. Interestingly, this does not happen when I am simply browsing around without editing. — Preceding
unsigned comment added by
VeryPunny (
talk •
contribs)
17:12, 31 January 2011 (UTC)
William H. Ryan, Jr.
The
William H. Ryan, Jr. Table of Contents has a line " * 4.1 Non-state Territories of the United States" appearing under References. The Commonwealth of Pennsylvania is one of the original thirteen states. Some technical person should investigate. Please tell me if I should be reporting this situation elsewhere, rather than here. --
DThomsen8 (
talk)
15:49, 31 January 2011 (UTC)
Which thumbnail software does Wikipedia use?
According to Mediawiki, it should be either ImageMagick or GD. But I'm not sure which, and how it's implemented (on commons, for an image, there's no thumb.php?size=xyz etc.). I'm thinking of fixing the bugs with Animated GIF thumbnailing. Thanks,
ManishEarthTalk •
Stalk16:19, 31 January 2011 (UTC)
Yes, it's ImageMagick. The parameters for calling it will be in svn somewhere - you could check related bugzilla issues which may lead you there.
OrangeDog (
τ •
ε)
18:55, 31 January 2011 (UTC)
Hi, I'm using Internet Explorer 8; I always check the stay logged in checkbox, but in the last few weeks Wikipedia won't remember my login for more than 20 minutes. I've tried clearing my cookies as suggested but that didn't fix the problem. Any ideas? Thanks.
Some guy (
talk)
20:28, 31 January 2011 (UTC)
Is there a bot that can add WikiProject templates based on categories or stubs?
I wonder if there is a bot that could check which articles in a given category (ex. Polish singers) or using given templates (ex. Poland-bio-stub) are not categorized with a corresponding WikiProject assessment template (ex. WikiProject Poland template), and then add the wikiprojects template to those talk pages (preferably, assessing it as stub if the articles have a stub template)? That seems like something a bot should be able to do with a good success rate. --Piotr Konieczny aka Prokonsul Piotrus|
talk 20:52, 31 January 2011 (UTC)
But the message following wikipedia flag/box has not gone away: "Constructs such as ibid. and loc. cit. are discouraged by Wikipedia's style guide for footnotes, as they are easily broken. Please improve this article by replacing them with named references (quick guide), or an abbreviated title."
I was wondering what I am missing RE formatting, etc...
The message is a note added by a user, not an automatically generated tag - it won't go away until explicitly removed. I've done so, now. In future, please do feel free to remove the notices when the issue's been resolved!
Shimgray |
talk |
21:24, 31 January 2011 (UTC)
This is bizarre
In
Double (basketball), Quadruple-double section, NBA subsection, in the notes, it reads:
"Olajuwon was originally credited a quadruple-double as shown by the box score; however, the NBA stripped Olajuwon of one assist assist after reviewing the game tape.[52]"
I see only one "assist", too. I can't think of any reason why that word would appear twice for you; it shouldn't be a cache problem, either, since "assist assist" hasn't appeared in the article before, at least not in the past few dozen edits. Gary King(
talk ·
scripts)03:03, 1 February 2011 (UTC)
en dashes and searching text
The MOS (in
MOS:ENDASH) suggests en dashes rather than hyphens in certain constructions, including in article titles. The only real problem with that is that it makes it very difficult to search for the term within the article text. Is there any way for en dashes to be counted as hyphens when searching, just as case distinctions are ignored, or is that entirely browser dependent? —
kwami (
talk)
01:48, 30 January 2011 (UTC)
It's browser dependent. You'll probably need a browser that supports regular expression searches in order for it to do what you want, or you could just do each search manually. Gary King(
talk ·
scripts)02:15, 30 January 2011 (UTC)
I was thinking of general accessibility to our readers. This would be an argument against implementing the MOS. —
kwami (
talk)
12:18, 1 February 2011 (UTC)
Left floated image overlaps table of contents
TOC and image overlaps
While reading the
Jingshan Park article I noticed that the TOC and image overlapped. I'm using Safari on Mac OS X and the bug only appears if the browser has a particular width, although at other withs the image border can overlap the text.--
Salix (
talk):
12:21, 1 February 2011 (UTC)
Problem with contributions not moving after username change
Last year, Feb 22 2010, I RTV'd per CHU
[10]. However, not all of my contributions moved over
[11], even though my old userpage claims the username is not registered
[12] and there is no "user contributions" link in the toolbox section. Can the remaining 3 or 400 contributions be moved over to the other username (see
[13] for new user name on 2/22/2010). If not, can they be moved to a similar named username. Rgrds. --
64.85.215.37 (
talk)
19:29, 1 February 2011 (UTC)
I recently put {{Expand}} (Hereinafter "Expand Tag). On the Reasonable doubt article. When I went to the article after saving it it had
Template:Expand where the Expand Tag box should be. Was the Expand Tag discontinued? Also if possible could someone put an alert on my talk page that my question was answered?
Thank You,
Etineskid (
talk)
02:26, 2 February 2011 (UTC)
I'm headed for watchlist bankruptcy again. Rather than dumping everything, I'd really like to have some sort of filter, e.g., that dumps heavily-watched pages, but keeps things that are on few watchlists. Alternatively, I'd be happy dumping anything that I haven't edited either the article or the talk page for, say, 90 days (especially user pages). Is there anything out there along these lines?
WhatamIdoing (
talk)
05:07, 29 January 2011 (UTC)
There also is a way to easily check in your watchlist if articles are redirects. That always gets rid of a couple of hundred articles for me.
Garion96(talk)17:33, 29 January 2011 (UTC)
Watchlist bankruptcy isn't usually invoked because of a technical restriction, but when someone finds they are having trouble managing their watchlist (see
Wikipedia:Don't overload your watchlist!). I believe there are some technical issues when you exceed a certain number - "9800" is mentioned on the previous link and
Wikipedia:WATCHLIST#Size limitation - but I was well beyond that number (I think close to 17000) when I declared watchlist bankruptcy last October, so that figure is probably a bit dated. –
xenotalk16:38, 31 January 2011 (UTC)
FWIW, the record was set by the now-departed
User:Wik, who I believe had over 50,000 articles on his watchlist. (I may be wrong.
[14]) He did so because he believed it was important to say, for example, Mongolia was in central Asia,
[15] & would revert people endlessly to keep it so. (This is why he is a "now-departed user".) In any case, a short watchlist is a happy watchlist. --
llywrch (
talk)
23:44, 2 February 2011 (UTC)
I gave up trying to manage my watchlist when it got past 2,000. It's one of the few major MediaWiki features I never use. The issue is that there are always some gadget(s) or program(s) that I don't realize has an auto-add-to-watchlist setting and I run a couple thousand pages in it. /
ƒETCHCOMMS/05:39, 1 February 2011 (UTC)
It's less than 2000 pages at the moment, but I'm not keeping up. I've done some manual trimming, but it's time-consuming. Garion, the redirects on my list are commonly things that I'm watching for a reason, e.g., to make sure that they stay redirects. Δ, thanks for the offer. I'll probably email the list to you later this month.
WhatamIdoing (
talk)
20:21, 2 February 2011 (UTC)
About redirects, the ones I still have I have for a reason. But old page moves redirects or vandalism redirects I can get rid off. See
User:Garion96/monobook.css for the script. It just strikes through the redirects but doesn't remove them.
Garion96(talk)22:38, 2 February 2011 (UTC)
This category just had a number of pages added, mostly Boston Red Sox players such as
Tim Wakefield. None of which belong. Anyone know why? (Where should such anomalies be reported - I saw some other mistakes earlier today)--SPhilbrickT01:01, 2 February 2011 (UTC)
Oddly, I was just looking at that diff, but nothing sunk in. Adding a hangon to an article not speedied, adds it to the list? Odd.--SPhilbrickT01:08, 2 February 2011 (UTC)
That's quite purposeful. An extremely high percentage of users accidentally replace the CSD nomination with the hangon template, and keeping the category makes sure that admins notice.—
Kww(
talk)
01:11, 2 February 2011 (UTC)
But that's just a subcategory--listing it there automatically lists int in the main category also. And so it should. Some of us who patrol speedy check the critical subcategories first, some go by the time the speedy was placed, some--like myself--go through it alphabetically. Just as Kww said, many new users who don't carefully read the instructions think that by placing the hangon tag they can remove the original speedy tag. They can't , but for the hangon tag to continue to list in the category makes sure we deal with the articles just the same. (The only problem is when someone adds a hangon tag to contest a prod or an AfD nomination, & thus gets the article listed in speedy also, which was certainly not their intent. When we admins see that, we just remove the tag & if necessary, explain what to do.) DGG (
talk )
05:16, 2 February 2011 (UTC)
From
WT:Invitation to edit: Does anyone know how to get the number of page views (per month is good enough) specifically served to readers who are not logged into an account? (We're talking about ~20 pages, no more than six [specific] months each.)
WhatamIdoing (
talk)
20:22, 2 February 2011 (UTC)
EasyTimeline not working correctly?
In the
distributed.net article, there is a timeline that shows the project's many milestones. However, the sections for OGR and DES had to be split into separate bars due to overlapping. The bars were previously named OGRa, OGRb, OGRc, DESa, DESb and DESc.
This seemed a little awkward, so I decided to use ditto marks instead:
However, this caused all of the data to be merged into one bar. It did not work like described in the help files. Am I doing something wrong, or is this a bug? --
Ixfd64 (
talk)
01:29, 3 February 2011 (UTC)
Possible bug in moving files and listing in categories
Looks like any subsequent edit to the categorized page (including null edits) will update a sort-key which defaults to {{PAGENAME}}. ―
cobaltcigs05:02, 3 February 2011 (UTC)
Currently,
Special:NewPages lists new pages in chronological order from newest to oldest. Users have to make an effort to patrol at the back of the backlog. I have two questions.
Is it possible to make
Special:NewPages list pages in reverse chronological order from oldest to newest?
Have we ever considered going in reverse order before?
Oops. I should have clarified that. There are over 1,000 users patrolling the front of the list and only a handful patrolling the back of the list. So several hundred articles fall off the end and aren't reviewed each month. If the list is reversed, then maybe more people would patrol at the end instead of the front. - Hydroxonium (
H3O+)
17:36, 1 February 2011 (UTC)
I doubt they'd be willing to change it. It's newest-to-oldest probably because every single page listing edits on Wikipedia are newest-to-oldest (Recent Changes, history pages, even nomination pages like
WP:FAC,
WP:RFA, etc.). And people are used to newest-to-oldest already. Gary King(
talk ·
scripts)17:38, 1 February 2011 (UTC)
and there's an advantage in trying to check the new ones , for this gets the worst problematic material deleted fastest. (of course, it also gets some things deleted too fast when they're still under construction.) Perhaps we should make the option to go from the end a little more prominent , to get a better balance. DGG (
talk )
05:20, 2 February 2011 (UTC)
Most of the Newpage backlog pages arent patrolled as they are sticky subjects (notability, etc). We should tag and patrol with a special tag, so BLP/Notability experts can check it out.
ManishEarthTalk •
Stalk11:50, 2 February 2011 (UTC)
or more simply, those who consider themselves such experts should be encouraged to go from the back. When I patrol, if I want to do something easy, I look at the front, if I'm up to actual thinking, I look at the back. DGG (
talk )
20:22, 2 February 2011 (UTC)
This might be a good chance to advertise
this new bot. Pretty good if you're having trouble staying on top of the new page list cut off. -
Kingpin13 (
talk)
15:53, 3 February 2011 (UTC)
Article page is a dabpage, talk page is a redirect
I really am stuck! Please see
this sequence of HTTP request and response headers made from the wolfsbane toolserver. The first pair are from the use of login() in
Wikibot.php5. The next is a GET which sends the expected six cookies apparently correctly. But the response to it has completely ignored the cookies. Can anybody spot what I am doing wrong? That report was generated with
this call and the source of the script can be
seen here. —
RHaworth(
talk·contribs)01:54, 3 February 2011 (UTC)
enwiki_session (and possibly centralauth_session) should only be sent when making edits and doing other stuff that requires tokens - you get a new one every time you fetch an edit (or whatever) token.
MER-C03:05, 3 February 2011 (UTC)
No, they can be sent whenever you make a request, as long as you update them whenever the server happens to send back a new one. Most HTTP toolkits will actually handle that automatically for you, anyway, and every one of them will send those cookie with every request.
Anomie⚔03:19, 3 February 2011 (UTC)
(
edit conflict) I see no indication in that transcript that the server ignored the cookies, and only the first query (for Special:WhatLinksHere/Template:Oscoor) would show any difference anyway. Although I do note that when I query that page logged out the Content-Length is 19924, while logged in it is 21329 due to extra Javascript variables and such; in your transcript the Content-Length is 22171, which might indicate that your login was successful. Redo the transcript with a query for
http://en.wikipedia.org/wiki/Special:MyPage, that will tell us for sure whether you're logged in or not because it will serve an HTTP 302 redirect to either your userpage or the userpage of the IP address.
I recommend you check whether you are still logged in in this fashion. It is pointless for read requests to attach cookies unless you want a token.
MER-C08:50, 3 February 2011 (UTC)
If you don't pass the cookies for a read request, then MediaWiki won't see you as being logged in and you won't get the advantages of apihighlimits.
Anomie⚔01:48, 4 February 2011 (UTC)
Automate stock information through RSS feeds
I am not an expert on matters of stock markets, but wouldn't it be possible to use a reliable RSS feed to guide a bot to update
Template:Infobox company? We could place a flag like NASDAQ = ????, and we could start on only a few companies to begin with. The bot could update with most recent closing price and most recent stock volume, and the template could automatically generate a company value (assuming that those numbers would equal that, I'm not certain myself). Does this sound like something that would be feasible to do? ▫ JohnnyMrNinja05:57, 3 February 2011 (UTC)
Sounds like the bot would have to make thousands of edits a day. It makes more sense to just put market capitalization values "as of January 1" or the last quarter, instead of transforming Wikipedia into an up-to-date stock ticker. Gary King(
talk ·
scripts)06:16, 3 February 2011 (UTC)
There are many bots that make thousands of edits a day. But sure, it could be made to only work once a month or once a week (weekly would probably be best at any rate). Many articles on publicly-traded companies contain well-out-of-date financial info. I'm more interested to know if it is technically possible, for a bot to respond to an RSS feed. ▫ JohnnyMrNinja07:02, 3 February 2011 (UTC)
An RSS feed is just
XML. Most languages with bot software should have an RSS parsing library somewhere out there (e.g.
[17]), and if not then it is relatively easy to roll your own.
MER-C07:18, 3 February 2011 (UTC)
True that some bots can make a huge amount of edits, but I think that few of them constantly update the same page over and over again. I was just thinking how useless it might be to have a history page for a rarely edited article filled to the brim with updates by a single bot. (I do realize that some bots do constantly update pages over and over again, such as WP 1.0 statistics pages, but those exist explicitly for the bot to edit.) Gary King(
talk ·
scripts)16:09, 3 February 2011 (UTC)
Yes but why? Constantly updating information like this is not what an encyclopaedia is for. If people want stock info they should go to a stock market website. If they want financial news they should consult a financial news agency.
OrangeDog (
τ •
ε)
11:23, 3 February 2011 (UTC)
I’m not going to argue about that. A specialized (non-encyclopedia) wiki may find this technique useful however. ―
cobaltcigs23:24, 3 February 2011 (UTC)
If done correctly, this can be achieved with one bot update every week/month. The bot will collect all the information into one file with a huge {{#switch:}} as cobaltcigs suggested above. I still would not support it, from reasons similar to what OrangeDog mentioned, but this belongs on the proposal, not on the technical discussion, so I'll voice it there. --
Muhandes (
talk)
16:30, 3 February 2011 (UTC)