Jump to content

Wikipedia:Village pump (technical)

From Wikipedia, the free encyclopedia

This is an old revision of this page, as edited by Apoc2400 (talk | contribs) at 20:27, 20 June 2009 (→‎Preferences changed by them selves: new section). The present address (URL) is a permanent link to this revision, which may differ significantly from the current revision.

 Policy Technical Proposals Idea lab WMF Miscellaneous 
The technical section of the village pump is used to discuss technical issues about Wikipedia. This page is not for new feature requests. Bugs and feature requests should be made at the BugZilla or the Village pump proposals page because there is no guarantee developers will read this page. Problems with user scripts should not be reported here, but rather to their developers (unless the bug needs immediate attention).

Newcomers to the technical village pump are encouraged to read these guidelines prior to posting here. Questions about MediaWiki in general should be posted at the MediaWiki support desk.

Someone tried to active / take control of Google Analytics for WP

I marked this for CSD and it was quickly deleted, but I am concerned that this was truly a hacking attempt (albeit not terribly effective) and wanted to raise it to the the attention of security experts.    7   talk Δ |   00:28, 1 June 2009 (UTC)[reply]

Original discussion from Gimmetrow's talk page:

== Google Analytics ==

Hi - saw you declined AIV action on this. I just wanted to make sure you understood this was a hacking attempt to gain control of using google analytics over all of WP. Seems pretty severe to me. Thanks    7   talk Δ |   23:15, 31 May 2009 (UTC)[reply]

The file is deleted. What exactly would this have allowed? Gimmetrow 23:33, 31 May 2009 (UTC)[reply]
Google Analytics will give people stats on links in/out, hit's per day, referrer, IP, etc... Well, it might not have been a very effective hacking attempt, because this "verification" step is only the first part of using http://www.google.com/analytics/. The second part would be adding a script to pages (ideally I suspect the poster would have tried to do it on all pages, but at the very least they could do it on some pages they created or on their user page. I guess my point is that even if it was harmless and the user couldn't have done much with it, it was still a hacking attempt.    7   talk Δ |   23:39, 31 May 2009 (UTC)[reply]
My other concern is that this verification step is to ensure that the user has control over the site (which they prove by posting a file), and as long as the site doesn't return error404 then google assumes the file is there - so in theory even though the file is deleted it is probably not returning a 404.    7   talk Δ |   23:41, 31 May 2009 (UTC)[reply]
If there is a potential privacy hole here, shouldn't someone try it out and report the results? Gimmetrow 23:44, 31 May 2009 (UTC)[reply]
Yes, a white-hat should. I have no idea if this new user is acting with the consent of WMF and in the project's best interest.    7   talk Δ |   23:51, 31 May 2009 (UTC)[reply]
Well, one thing I notice is that some stats are limited to 5M pageviews per months for non-adwords users. That wouldn't get too far here ;) Gimmetrow 00:07, 1 June 2009 (UTC)[reply]
I completely agree that this wouldn't be the most effective hack even if it worked, but again my point is it seems like something should be done about it regardless of how effective it is. Is there perhaps another place I should mention this?    7   talk Δ |   00:10, 1 June 2009 (UTC)[reply]
I think WP:VPT would be the best place. Gimmetrow 00:13, 1 June 2009 (UTC)[reply]

The verification does work to some extent. If there is any concern here, we could add /.*google.*html/ to the spam/abuse filters. Gimmetrow 00:36, 1 June 2009 (UTC)[reply]

May also want to add google-analytics.com/urchin.js and google-analytics.com/ga.js filters, as those are the scripts that users would try to add (actually, I suspect the entry parser may already block them?).    7   talk Δ |   01:00, 1 June 2009 (UTC)[reply]
I don't think we need to worry about any of this. The google authentification file must, I believe, be placed in the root directory, which can't be achieved through the interface, and it's not possible for non-admins to add any javascript that would automatically be run by users to collect the data. Yes, people may attempt to create those articles, but they do no harm, and can simply be deleted as they come in. Amalthea 11:15, 1 June 2009 (UTC)[reply]

Don't be ridiculous. If the user could have added a script to all pages, then we would have a very serious XSS security vulnerability, and it would be fixed within an hour or two. There is no reason to be alarmed here. — Werdna • talk 13:10, 1 June 2009 (UTC)[reply]

Sorry you felt this post was ridiculous.    7   talk Δ |   15:02, 1 June 2009 (UTC)[reply]
I've posted a note at User talk:Michkol1, inviting the editor to explain, here, what he/she was trying to do. If there is no explanation provided, I suggest that it may be appropriate to block the account. -- John Broughton (♫♫) 19:09, 2 June 2009 (UTC)[reply]
I am using a content management system which uses /word1.word2.word3 style URLs. One of my users created a page called "Google*** html" and added the website to his Google Webmaster Tools. I was checking if it works on Wikipedia. I was able to add http://en.wikipedia.org/wiki/ to my Google account. I didn't change any settings in Webmaster Tools. Google's verification system is insecure because it doesn't check file contents - verifying a site requires only creating a file/page. Michkol1 (talk) 20:13, 2 June 2009 (UTC)[reply]
This regular expression matches all Google verification files: Google[0-9a-f]{16}\.html Michkol1 (talk) 20:46, 2 June 2009 (UTC)[reply]
It's nasty that anyone could get even this far, and it should be stopped, as part of a general strategy of defense in depth against blended threats. If these pages are a way to convince Google that someone "owns" Wikipedia, could their generation be blocked using the abuse filter or page name blacklist? Has anyone contacted one of the developers about this, to see if this can be blocked globally in MediaWiki itself? -- The Anome (talk) 23:22, 2 June 2009 (UTC)[reply]
m:Title blacklist works for all Wikimedia wikis. Mr.Z-man 00:45, 3 June 2009 (UTC)[reply]
I'm not a meta admin: could someone make a suitable blacklist entry there, please? I believe "google.*\.html" would probably do the job in a reasonably future-proof way, assuming that the blacklist documentation is right about matching being case-insensitive. -- The Anome (talk) 13:21, 3 June 2009 (UTC)[reply]
It is case-insensitive. However, I'm not clear why we need to block such titles. Could someone give a short & clear explanation?  — Mike.lifeguard | @en.wb 22:52, 3 June 2009 (UTC)[reply]
It is the first step (verification process) of proving that the person who posts it is the site owner for use in google webmaster tools - please see this link.    7   talk Δ |   01:47, 5 June 2009 (UTC)[reply]

The fact that the offending page must be put in the root directory is quite deliberate on the part of Google and others, because they know that subdirectories frequently include unprivileged or semi-privileged input. There is no risk, please relax. — Werdna • talk 00:07, 18 June 2009 (UTC)[reply]

is there a way to search several sections with one search?

Is there a way to input several prefix parameters to search? For example, is there a way to search the "Wikipedia:Votes for deletion" section and the "Wikipedia:Articles for deletion" for "minor leagues baseball"? Is there any way to do this with one search? --stmrlbs|talk 02:09, 10 June 2009 (UTC)[reply]

not at the moment, but I might hack it up if I get some time over the weekend. --rainman (talk) 09:02, 10 June 2009 (UTC)[reply]
boy, that would be nice. It would be really nice if a person could pass a list of "prefixes" that indicated which sections to search. But, hey, I'll take any way you can set up to do it.  :) --stmrlbs|talk 18:50, 10 June 2009 (UTC)[reply]
this works now, thanks to rainman, and I have updated the search documentation for anyone wishing to use prefix for searching more than one namespace/section.--stmrlbs|talk 07:15, 17 June 2009 (UTC)[reply]

Dumb Template Trick of the Week™

Well, I've got the basics down, but could somebody help make this fully functional… support for octal and binary numbers would be good too. You might even help find a use for this. — CharlotteWebb 07:00, 14 June 2009 (UTC)[reply]

o_O Do you have to? — Werdna • talk 21:27, 17 June 2009 (UTC)[reply]

Install StringFunctions and we'll stop using the hacks :P Happymelon 21:37, 17 June 2009 (UTC)[reply]
Don't spend too long writing templates that use those padleft hacks, some day they might stop working. -- Tim Starling (talk) 13:54, 18 June 2009 (UTC)[reply]
And then there will be howls of outrage, like there was the last time the hack was blocked. I'm with you all the way that the hack is just that, a horrible hack, but the fact that it's seeing widespread use is indicative of the need for this functionality to improve editing and viewing experience. If you enable a proper, parser-friendly, non-hackish implementation, I will personally hunt down and replace every instance of the padleft-based hacks, and then break out the champagne when the hack is blocked. But arbitrarily removing used and needed functionality without providing a replacement is not a constructive way to improve MediaWiki. Happymelon 11:48, 19 June 2009 (UTC)[reply]
Actually, the correct phrase is enable the string functions. The string functionality made it into the now live version of ParserFunctions, but is disabled by default with a global configuration setting. Dragons flight (talk) 21:53, 17 June 2009 (UTC)[reply]
It made it through code review? I'm very surprised (per our discussion at mw.org). But as you say, it's still a config line away. Happymelon 22:23, 17 June 2009 (UTC)[reply]
Tim made several modifications (e.g. fixing a parser state issue I had ignored, changed several other minor things, and added the global disable flag), but yes, the functionality is still there. Dragons flight (talk) 22:35, 17 June 2009 (UTC)[reply]
Tim did comment that he would prefer mw:Extension:Lua instead. But no word on whether we'll get that any time soon either :( Anomie 23:59, 17 June 2009 (UTC)[reply]
I've heard other devs be quite skeptical of lua, which would give us a roughly Turing complete function system and requires compiled binaries to operate (limiting its availability for other sites). But who really knows. Dragons flight (talk) 00:10, 18 June 2009 (UTC)[reply]
I'm reasonably convinced that with the padleft hack and an unlimited preprocessor node count, the current parser with ParserFunctions is Turing-complete. It would be an interesting challenge... Happymelon 11:56, 19 June 2009 (UTC)[reply]
With neither looping nor recursion, I don't believe you can do it. Lua would allow both of those things. Dragons flight (talk) 12:07, 19 June 2009 (UTC)[reply]
With an infinite page base, you can simply create as many copies of the function you want to recurse as you want (since there's no way to return values from 'functions', our 'program' flow is almost entirely by going deeper into the transclusion stack anyway, potentially without limit). I agree it's a rather pedantic distinction when you need to give yourself infinite resources to get around the restriction, but true Turing-completeness depends on that anyway. Looping is actually harder than recursion; if you allow database actions to be performed as part of the machine's cycle (that is, can you say that you need to edit a page a few times and replace whatever's there with subst:something to clock the machine over) then it's definitely possible. It's certainly not something I'd want to try and build an OS from :D Happymelon 12:52, 19 June 2009 (UTC)[reply]
To be honest though, I think I'm with Werdna here, in wondering Why? I fail to see where those conversions would actually get used. Dragons flight (talk) 21:59, 17 June 2009 (UTC)[reply]

Something based on this could be used to choose complementary colors for foregrounds/backgrounds of templates, etc. by subtracting it from 0xFFFFFF. Picture something like:

<td style="color:#{{{1}}}; background:#{{invertcolor|{{{1}}}}};"</td>

Maybe I'm getting ahead of myself, shrug… — CharlotteWebb 17:34, 20 June 2009 (UTC)[reply]

Magic word for Special:WhatLinksHere

Is there a magic word that returns the number of links in a "what links here" page? I'm thinking something along the lines of {{PAGESINCATEGORY}} but for Special:WhatLinksHere. rʨanaɢ talk/contribs 13:30, 14 June 2009 (UTC)[reply]

Presumably you have looked on the pages that would document such a feature if we had one (mw:Help:Magic words, mw:Help talk:Magic words, Help:What links here). You could search bugzilla for: Special:WhatLinksHere to see if anyone has requested this. There is a Special:MostLinkedTemplates page which isn't what you want. --Teratornis (talk) 19:23, 14 June 2009 (UTC)[reply]
In short, no, there isn't. What were you thinking of using it for? haz (talk) 07:12, 15 June 2009 (UTC)[reply]

I imagine the purpose would be to determine when an article is no longer {{orphan}}ed, so that the template in question can invoke a different category (or even make itself disappear) as appropriate. — CharlotteWebb 16:53, 15 June 2009 (UTC)[reply]

Nah, nothing that useful...I was mainly just thinking of using it as a way to do this template without having to use a category. Right now it uses a category that is populated by a userbox, but the same thing could be accomplished with WhatLinksHere...the only difference is that using a category allows me to have that template (using PAGESINCATEGORY), but WhatLinksHere doesn't. Some people objected before to having this category (it was CfD'd), but whatever, it's not a huge deal, just thought I'd ask. rʨanaɢ talk/contribs 04:31, 19 June 2009 (UTC)[reply]

Written Language Auto-Traslation Systems : Language barrier in Wikipedia

Language barrier in Wikipedia or anywhere in the world of communication can be overcome atleast for the written media through autotranslation; link for this is given given here to convert any of languages listed in the program to another; conversion is automatic and on-the-fly. The translation may not be linguistically accurate in usage or contextual meaning.

  • Translates from original with original not shown [1]
  • Translates from original with original shown side by side [2]

With program systems such as above, it seems a Wikipedia page or any written text can be traslated to another language on the list in the programs or similar program as may come to light. A high level discussion is needed that the Wikipedia main page should include availabilty of Auto-Translation, though says distinctly that that page is only for the content of that page. Above matter, actually pertains to Wikipedia opening page with the globe and multiple languages of the world shown, but has no discussion facility on that page. This is the page where some form of this matter can be incorporated and will receive wider exposure of availabilty of Auto-Translation Systems. I will attempt to insert the same on the Main page also; this one is from a beginner with above link deleted twice from the External Links on Air France 447 page where I had established a translation using the the first Translation system link #1 above to translate the Brazilian Air Force Portal with lot of information on the progress of accident investigation. Feel free to help these re-establish link on Air France 447 page if appropriate. Patelurology2 (talk) 01:08, 15 June 2009 (UTC)[reply]

We have other language Wikipedias for people who don't speak English... Stifle (talk) 08:52, 15 June 2009 (UTC)[reply]
Search through the archives— this has been discussed several times. Machine translation is not reliable enough for our purposes. ---— Gadget850 (Ed) talk 10:07, 15 June 2009 (UTC)[reply]
See Wikipedia:Wikipedia Signpost/2009-06-15/News and notes for the announcement of the Google translator toolkit. It is not automated, but it should be more reliable. ---— Gadget850 (Ed) talk 12:52, 16 June 2009 (UTC)[reply]

Javascript not working?

My use of User:Ais523/hidetopcontrib.js has suddenly, this morning, stopped working. The "hide top contrib" tab still shows up, but clicking it makes no difference. Also, my Twinkle rollback links have disppeared from my contribs-page. Can anyone help? ╟─TreasuryTagsecretariat─╢ 10:35, 15 June 2009 (UTC)[reply]

Nothing has changed, BTW. I have tried on two different computers. ╟─TreasuryTagquaestor─╢ 11:10, 15 June 2009 (UTC)[reply]
I had a look, but I can't quite fix it. I think the problem must lie in a change to the html of Special:Contributions caused by the recent update of the MW version that en's running. Someone more familiar with the changes might be able to advise. - Jarry1250 (t, c) 11:39, 15 June 2009 (UTC)[reply]
The script checks whether you are the top contributor by looking for <strong>(top)<strong> in the HTML for each entry. However, the (top) text is no longer bolded using the strong tag - it's set to <span class="mw-uctop">. Thus, the script thinks you are not the top contributor for anything, and does nothing. You'd be best off asking the script developer to fix it. Ale_Jrbtalk 12:04, 15 June 2009 (UTC)[reply]

<< And presumably TWinkle works by putting "rollback" links next to top contribs using the same method? ╟─TreasuryTagsundries─╢ 12:12, 15 June 2009 (UTC)[reply]

It probably does something similar, although I can't actually be bothered to look through the twinkle code and find where it broke. MediaWiki was updated around 4 hours ago (still ongoing, possibly) which will be what broke it. Developers will need a while to fix things. Ale_Jrbtalk 12:14, 15 June 2009 (UTC)[reply]
Hmm, I just looked, it does – can you think of what specific coding should be used for the correction, that'd probably make it easier, if I can tell the scripter exactly what could be done as a fix? Thanks!! ╟─TreasuryTagconstabulary─╢ 12:21, 15 June 2009 (UTC)[reply]
I'm looking at possible solutions for my version of User:Markhurd/hidetopcontrib.js now. Mark Hurd (talk) 14:03, 15 June 2009 (UTC)[reply]
Fixed for Twinkle, thanks. Amalthea 15:06, 15 June 2009 (UTC)[reply]
Fixed my User:Markhurd/hidetopcontrib.js. Also added userHideAllSubsequent, which when =true will also hide all but the most recent edit, so it is more like watchlists. Mark Hurd (talk) 15:35, 15 June 2009 (UTC)[reply]
Where is the code for the hide subsequent tab? You must have read my mind, because I dreamt of having this feature earlier today.Ost (talk) 20:51, 15 June 2009 (UTC)[reply]
userHideAllSubsequent is not another tab, just a user setting you can specify in your monobook.js. E.g.
userHideAllSubsequent=true;
importScript('User:Markhurd/hidetopcontrib. js');
Mark Hurd (talk) 16:20, 16 June 2009 (UTC)[reply]

Sorry, that all started out making sense to me then quickly surpassed my understanding. I used those flags all the time to see if someone had replied to a message, added information to an article or whatnot. Knowing whether I was the last contributor or not made my use of WP much more efficient. Will there be some kind of fix in place one way or the other? Matt Deres (talk) 23:37, 15 June 2009 (UTC) (who came here to ask that very question)[reply]

Ah, all better now; never mind! Matt Deres (talk) 10:30, 16 June 2009 (UTC)[reply]

Autosig not working ~~~~

Alright, who is playing with the autosig?[3] That extra "Contributions/" is totally unneeded. Contributions/199.125.109.102 (talk) 20:37, 15 June 2009 (UTC) (supposed to be 199.125.109.102 (talk) 20:37, 15 June 2009 (UTC))[reply]

Hmmm, I don't see anything here: MediaWiki:Signature-anon. --MZMcBride (talk) 20:47, 15 June 2009 (UTC)[reply]
"$2" in there is malfunctioning as "Contributions/xx.xx.xx.xx" instead of just the IP address. A change has been proposed to remedy this until it is fixed. I'm not sure what $2, the "nickname" for the IP is however I have never seen an anon have another signature besides its own IP address. Triplestop (talk) 21:20, 15 June 2009 (UTC)[reply]
Why do we use the variable $2? Looking at the source, it appears as if $1 is actually the IP, and $2 is... something else? Maybe this is causing the error. Of course, I'm probably wrong about this, but my proposed remedy is to change the $2 in the source to $1. The Earwig (Talk | Editor review) 21:37, 15 June 2009 (UTC)[reply]
I'm looking through the code revisions however I am unable to find the culprit. See [4] Triplestop (talk) 21:39, 15 June 2009 (UTC)[reply]
Go to [5] and find where it says "function getUserSig( &$user ) {" Triplestop (talk) 21:55, 15 June 2009 (UTC)[reply]
(ec) I applied that workaround. I don't think anons can ever have a nickname option set here on enwiki, so it can probably stay in permanently. Amalthea 21:57, 15 June 2009 (UTC)[reply]

Fixed. Thanks. 199.125.109.102 (talk) 23:52, 15 June 2009 (UTC)[reply]

Well, the symptoms are fixed. The bug is filed as bugzilla:19232. Amalthea 10:45, 16 June 2009 (UTC)[reply]

Surprise pipe-trick. $2 is the "nickname" of an IP address which doesn't have one, so it was blank, giving us something like:

--[[Special:Contributions/127.0.0.1|]]

which expands itself by copying everything between the colon and the vertical bar. — CharlotteWebb 19:17, 16 June 2009 (UTC)[reply]

Aaah ... stupid pipe trick. Thanks, Amalthea 19:26, 16 June 2009 (UTC)[reply]

{{coord}} / CSS malfunction?

Is anyone else seeing a display problem with coordinates at the page title level? Suddenly tonight I am seeing the coordinates baseline running slightly below the horizontal rule underneath the title. See Lawrence, Massachusetts, Lodi, California, London which I have checked. I can't remember whether it was above or below the rule before - I think it was below. Happens in monobook on FF 3.0.11/IE7 on WinXP and FF 3.0.11/Safari 4/Opera 9.64 on Mac OS X 10.5, whether I am logged in or out. I removed custom javascript and purged the page, cache etc. and still always displays crashed into the rule. Sswonk (talk) 01:07, 16 June 2009 (UTC)[reply]

This is a perennial, or recurring issue. 199.125.109.102 (talk) 01:14, 16 June 2009 (UTC)[reply]
Yes, for example in Archive 38 from a year ago. I asked whether anyone else is seeing it, although based on what I tested I'm pretty sure everyone is. Looks messy, and the template page was last edited in February. I understand it is the template that needs to follow changes to the default CSS/js, any ideas who to notify? Sswonk (talk) 01:46, 16 June 2009 (UTC)[reply]
FWIW, importScript('User:TheDJ/movecoord.js'); mentioned on that archive page fixes it for me. Sswonk (talk) 01:53, 16 June 2009 (UTC)[reply]
And Archive 11 before that. Looking at some old screenshots the coordinates were normally below the line unless you were logged in, in which case they were well above the line if there was a central notice. There used to be a little globe too, and the top of the globe was just at the bottom of the line, in between the word Coordinates and the coordinates. 199.125.109.126 (talk) 03:30, 16 June 2009 (UTC)[reply]
I'm pretty sure this is a side-effect of bugzilla:19219, which has been causing a lot of problems with those "topicon" things. The Earwig (Talk | Editor review) 10:30, 16 June 2009 (UTC)[reply]

License update, mediawiki messages

The license update for en.wikipedia has been implemented by Eloquence [6]. The new copyright warning is now four-lines long and contains several external links that may need to be made in plain text, and the format of edittools rethought as well. As of now, it's not very user-friendly. We must follow those instructions. Cenarium (talk) 01:14, 16 June 2009 (UTC)[reply]

Could the line "Re-users will be required to credit you in any medium, at minimum, through a hyperlink or URL to the article you are contributing to." be removed? That's all covered in the terms of use. (I'm not going to do it myself because I don't want to really mess with the legal stuff). And maybe it could be enclosed in <small> tags? –Drilnoth (T • C • L) 01:21, 16 June 2009 (UTC)[reply]
No, Erik has been fairly insistent on that bit. Dragons flight (talk) 01:34, 16 June 2009 (UTC)[reply]
Which part? Or is it both? –Drilnoth (T • C • L) 01:40, 16 June 2009 (UTC)[reply]
Pretty much everything beginning "By saving, ..." is considered required information. In particular, Erik has been insistent about the "Re-users..." sentence. Dragons flight (talk) 01:53, 16 June 2009 (UTC)[reply]
Also, maybe <span class="plainlinks"> could be put on each of the external links? –Drilnoth (T • C • L) 01:25, 16 June 2009 (UTC)[reply]
Plainlinks should be fine. Dragons flight (talk) 01:34, 16 June 2009 (UTC)[reply]
Thought not at all simpler, it is interesting to see this version at Commons which has the virtue of being pretty (and also pretty huge). Dragons flight (talk) 01:34, 16 June 2009 (UTC)[reply]
Please no template! Really, on commons, users almost never actually edit, but here, we don't want to scare away new users with such notices. Let's be concise and to the point. Cenarium (talk) 01:39, 16 June 2009 (UTC)[reply]
No. It looks nice, but as Cenarium said that might scare people away and it is way to big. –Drilnoth (T • C • L) 01:42, 16 June 2009 (UTC)[reply]
Just want to point out that something in between, like:
is probably possible, which is two lines on my screen. In other words, make it smaller, but also highlight it in some way that encourages people not to overlook it entirely. Dragons flight (talk) 01:49, 16 June 2009 (UTC)[reply]
I'd say let's see how simply changing in plain links renders first for now (I'm no expert in this stuff, so I prefer not to do it.), then seek more input. But yes, something in those lines may be possible. Cenarium (talk) 02:03, 16 June 2009 (UTC)[reply]
I've put in the plainlinks because the arrows would drive me nuts. Dragons flight (talk) 02:06, 16 June 2009 (UTC)[reply]
Sounds good. –Drilnoth (T • C • L) 02:09, 16 June 2009 (UTC)[reply]
Did the same for edittools. Now we can write local copies, so we can link internally. Cenarium (talk) 02:23, 16 June 2009 (UTC)[reply]
Local copies, done. Dragons flight (talk) 04:53, 16 June 2009 (UTC)[reply]

The attribution permission is important because some users have found our terms confusing in this regard in the past, insisting that in spite of fairly clear usage guidelines, attribution must happen by crediting every single contributor by name. We want to make sure that there's absolutely no confusion about this point for anyone who contributes. But text refactoring within reason, abbreviating the licenses, changing the font size etc. may all be ways to make it less bulky.--Eloquence* 01:55, 16 June 2009 (UTC)[reply]

Would it be possible to change "GNU Free Documentation License" to "GFDL" and the CC license to "CC-by-SA-3.0"? Would that be legally possible, and if so do you think that it would get to confusing doing that? The latter more confusing than the first. –Drilnoth (T • C • L) 02:09, 16 June 2009 (UTC)[reply]
Is "some users" a significant number, or just a couple of trolls on a mailing list? --Apoc2400 (talk) 17:58, 16 June 2009 (UTC)[reply]

Maybe move "Encyclopedic content must be verifiable" down to the "Please note" section below the character insertion box? –Drilnoth (T • C • L) 02:11, 16 June 2009 (UTC)[reply]

My thanks

My thanks to whoever decided that the XML produced by the Wikipedia API could use an unannounced change -- one of my bots spent half the day falsely informing people that they'd uploaded unsourced images. --Carnildo (talk) 01:37, 16 June 2009 (UTC)[reply]

Which part of the API? Dragons flight (talk) 02:01, 16 June 2009 (UTC)[reply]
"prop=revisions": the <rev> element now has a 'xml:space="preserve"' attribute, which causes naive XML parsers such as XML::Simple to convert it into a data structure rather than a string. I haven't had time to check for other changes, so there may be more. --Carnildo (talk) 04:35, 16 June 2009 (UTC)[reply]
Ouch, I have code that may break on that too. Thanks for the heads up. Dragons flight (talk) 05:08, 16 June 2009 (UTC)[reply]
It's rev:50041 in response to bugzilla:18617. It occurs whenever a text node is given as an element's content. haz (talk) 07:54, 16 June 2009 (UTC)[reply]
My apologies, I had no idea that XML parsers would differ between the absence and presence of xml:space="preserve" in such a brain-damaged way. The attribute was added for the benefit of people using XML parsers with the very annoying (but still standards-compliant) behavior of folding whitespace in elements' contents when xml:space="preserve" was absent. Next time an API format changes in such a small way, I won't consider it too small to announce again.
To answer Carnildo's question: no breaking changes were announced since the last scap, and no unannounced breaking changes were introduced. --Catrope (talk) 08:33, 16 June 2009 (UTC)[reply]

Sarcasm is rarely appreciated in reporting problems, especially since the fault in this case is with your parser, and not with the MediaWiki API. — Werdna • talk 21:22, 17 June 2009 (UTC)[reply]

Central buttons

I really don't like having the buttons and fields centralized on Special:RecentChangesLinked and the &action=delete pages... especially when other pages are the same as they have been. Should a bug report be filed, is this fixable in MediaWiki-space, or is there really consensus for this change? It seems much less usable (less intuitive), and looks fairly ugly, IMO. –Drilnoth (T • C • L) 01:47, 16 June 2009 (UTC)[reply]

Actually, I guess it's been that way on the delete page... but on RecentChangesLinked it is definitely new. –Drilnoth (T • C • L) 02:34, 16 June 2009 (UTC)[reply]
It looks even worse in the User Profile tab in Monobooks' Special:Preferences: The 30%/70% column widths causes a very different layout in the four fieldsets since the tables themselves are of very different width, depending on the content. The labels in the two top fieldsets have a width of 159px and 115px here, and the labels in the bottom fieldsets 492px. Looks extremely bad. Amalthea 10:00, 16 June 2009 (UTC)[reply]

How do you use a variable to insert a page's name uncapitalized?

I'm trying to insert the pagename into part of the title of another page (Outline of x, where x is the pagename). Unfortunately, it turns out like Outline of Construction instead of Outline of construction.

Is there any way to return the pagename in small case, without the capitalization?

The Transhumanist    02:08, 16 June 2009 (UTC)[reply]

{{lc:BOB}} = bob
{{lcfirst:BOB}} = bOB
Dragons flight (talk) 02:10, 16 June 2009 (UTC)[reply]
how would that work with:
{{main|Outline of <includeonly>{{subst:</includeonly>PAGENAME}}}}
I look forward to your reply. The Transhumanist    02:54, 16 June 2009 (UTC)
Hum, not sure this is what you want to do but try this??
{{main|Outline of <includeonly>{{subst:</includeonly>{{lc:PAGENAME}}}}}}
--Stefan talk 04:57, 16 June 2009 (UTC)[reply]
{{<includeonly>subst:</includeonly>lc:{{<includeonly>subst:</includeonly>PAGENAME}}}}. --Splarka (rant) 07:50, 16 June 2009 (UTC)[reply]
Having had a look at Special:PrefixIndex/Outline of, this would be better:
{{main|Outline of {{<includeonly>subst:</includeonly>#ifexist:Outline of {{lcfirst:{{PAGENAME}}}}|{{<includeonly>subst:</includeonly>lcfirst:{{<includeonly>subst:</includeonly>PAGENAME}}}}|{{<includeonly>subst:</includeonly>PAGENAME}}}}}}
Works for both Outline of construction and Outline of Albania. Just make sure you subst the template. haz (talk) 15:12, 16 June 2009 (UTC)[reply]
Thank you. This works better, but it doesn't catch the outlines with "the" in them, like Outline of the United States, Outline of the Cayman Islands, or Outline of the Internet. There are quite a few of those. And it doesn't catch outlines that use the subject's plural form, such as Outline of ants and Outline of sharks. The Transhumanist    21:28, 16 June 2009 (UTC)[reply]
OK... <takes a deep breath...>

{{main|Outline of {{<includeonly>subst:</includeonly>#ifexist:Outline of {{lcfirst:{{PAGENAME}}}}s|{{<includeonly>subst:</includeonly>lcfirst:{{<includeonly>subst:</includeonly>PAGENAME}}}}s|{{<includeonly>subst:</includeonly>#ifexist:Outline of the {{PAGENAME}}|the {{<includeonly>subst:</includeonly>PAGENAME}}|{{<includeonly>subst:</includeonly>#ifexist:Outline of {{lcfirst:{{PAGENAME}}}}|{{<includeonly>subst:</includeonly>lcfirst:{{<includeonly>subst:</includeonly>PAGENAME}}}}|{{<includeonly>subst:</includeonly>PAGENAME}}}}}}}}}}

Certainly works with all those examples. I dread to think what that would have looked like back in the days of {{qif}}... haz (talk) 10:44, 17 June 2009 (UTC)[reply]
Thank you. The Transhumanist    17:56, 18 June 2009 (UTC)[reply]

edit-tools not functioning in Safari for the Mac

Sorry if I've used the wrong terms. For at least a week, they've been black, not blue, and won't work. Even the blockquote syntax. I notice also that the "Cancel" button to cancel an edit has been black, although it turned blue again yesterday and remains functional. Is it just my computer? I use Safari for the Mac. Tony (talk) 08:14, 16 June 2009 (UTC)[reply]

This was caused by the recent software update, and has been fixed. --Catrope (talk) 09:06, 16 June 2009 (UTC)[reply]
It is called MediaWiki:Edittools. ---— Gadget850 (Ed) talk 12:49, 16 June 2009 (UTC)[reply]
Actually, I'm still having the same problem that Tony outlined above with the Edittools — I'm also on Safari for the Mac. Mlaffs (talk) 16:46, 16 June 2009 (UTC)[reply]
Yup, still all dead black. Tony (talk) 07:42, 17 June 2009 (UTC)[reply]
I Just updated a few minutes ago from Safari 4. to Safari 4.0.1 -- and they still do not work. I am a little puzzled, because the update talked about updating both Safari and JavaScript, & I figured that would fix it. FWIW , it works in Firefox 3.0.11. DGG (talk) 06:45, 18 June 2009 (UTC)[reply]
Confirming that there is still a problem and that it seems to be en:wp specific; no issue on en:Wikisource, Commons, or Meta. Cheers, Jack Merridew 09:26, 18 June 2009 (UTC) (who mostly uses Firefox;)[reply]
  • The edit tools here depend on the site javascript to function. If there's something in there that isn't compatible with with your browser, it may just abort the script, so later things stop working. Can someone who sees the problem check their javascript console to see if there's an error message in there which tells where it stopped? -Steve Sanbeg (talk) 20:18, 19 June 2009 (UTC)[reply]

(od) This appears to be a problem with MediaWiki/en serving to WebKit based browsers. I have tested on browsers with the following UA strings:

Google Chrome/WinXP Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US) AppleWebKit/531.0 (KHTML, like Gecko) Chrome/3.0.190.0 Safari/531.0

Safari 4/Mac OS X 10.5.7 Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10_5_7; en-us) AppleWebKit/530.17 (KHTML, like Gecko) Version/4.0 Safari/530.17

Fluid.app 0.9.6/Mac OS X 10.5.7 Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10_5_2; en-us) AppleWebKit/525.13 (KHTML, like Gecko) Version/3.1 Safari/525.13

Opera, Firefox, Internet Explorer all not effected. The edit tools beneath the edit box on the WebKit browsers all have the same issues: no dropdown menu to choose which set of tools to display, all tools display in a single box labeled "Copy and Paste" instead of the dropdown menu which begins "Insert" followed by "Wiki markup", "Symbols" etc. on the correctly functioning browsers. The problem is exclusive to WebKit browsers and en.wikipedia.org, other language versions (tested de, fr, pl) work correctly in WebKit. Sswonk (talk) 02:44, 20 June 2009 (UTC)[reply]

This is definitely a problem involving the MediaWiki/en.wikipedia.org interpretation of the User Agent string. To fix the problem in Safari 4, simply go to "Preferences" and under "Advanced" enable "Show Develop menu in menu bar". Close "Preferences", go to the newly active "Develop" menu and click on "User Agent", choose to assert you're any browser other than Safari (recommend "Firefox 3.0.10 - Mac" for the time being) and the edit tools start to work again. Sswonk (talk) 03:01, 20 June 2009 (UTC)[reply]

No, that doesn't seem to have done it, at least not for me. Followed those steps exactly - even restarted and did them again just in case - but no change. Mlaffs (talk) 03:17, 20 June 2009 (UTC)[reply]
Try several different UAs to see what happens. I am getting results all over the map but it seems if I go to "Internet Explorer 7.0" it will work. Try that one, then go to http://useragent.org to confirm "Mozilla/4.0 (compatible; MSIE 7.0; Windows NT 6.0)", then go back to a random Wikipedia page and try to edit. Caching may be a problem as well. If I go to "Default (Automatically Chosen)" it goes back to a "Copy and Paste" all black list, even after changing to FF or Opera later, but I seem to be able to kick it back to correct behavior if I go to IE7, then into anything but Default or Safari, which breaks it all over again. Sswonk (talk) 04:10, 20 June 2009 (UTC)[reply]
Man, this sucks. Still no go - I've tried as above, as well as other versions of IE, Firefox, and Opera. Still not getting a working set of tools any more than once about every 15 or 20 times. Mlaffs (talk) 04:26, 20 June 2009 (UTC)[reply]
Final report for the evening, it's getting late. I now am experiencing random drops in the implementation of the edit tools when using different user agents chosen by the Safari "Develop menu method" I talk about above. Sometimes choosing "Default" or "Safari" will work as well. The only thing I can say with a high degree of certainty is that it is a WebKit/en.wikipedia.org-only problem. Testing other browsers and other language sites fails to show the problem, but English Wikipedia under Safari will still fail to render the edit tools from time to time and Google Chrome on the Mac (still in developer only release) will always fail to render them. I believe it is time to file a bugzilla report, and I leave that to anyone reading this who is efficient at doing that as I have not yet tried one of those for MediaWiki. Sswonk (talk) 05:38, 20 June 2009 (UTC)[reply]
Probably made a hash of it, but I've filed a report at Bugzilla #19306. Mlaffs (talk) 12:14, 20 June 2009 (UTC)[reply]

I replicated it on Safari in Windows. I'll have a look at it. Ale_Jrbtalk 12:35, 20 June 2009 (UTC)[reply]

Bugzilla report has been closed as "RESOLVED INVALID", with the following explanation via email: It means it's closed because it's not a bug in the MediaWiki software. Per comment #1 on that bug, it's a bug in the JavaScript that the English Wikipedia community added locally. In less technical terms: the enwiki community added a customization, so if that customization is broken, they're the ones that can and should fix it, not 'us' MediaWiki developers. So, what now? Mlaffs (talk) 16:17, 20 June 2009 (UTC)[reply]

I've applied a fix. Please ensure you've cleared your cache, and check again. Ale_Jrbtalk 16:31, 20 June 2009 (UTC)[reply]
Yep, that's got it — EditTools are working great for me now. I'm still getting dialog boxes that I have to dismiss when I use HotCat, which also started happening following the Safari upgrade, but I assume that's a separate problem. Thanks much. Mlaffs (talk) 17:04, 20 June 2009 (UTC)[reply]
No worries. You'll have to talk to the developer of HotCat if you are having problems with it. It's funny though - I didn't think it was possible to hate Apple software any more than I did before. Ha - wrong! :) Ale_Jrbtalk 17:11, 20 June 2009 (UTC)[reply]
This fixes Google Chrome as well. WebKit is open source, not Apple software, and did nothing do deserve hate escalation since it was an en.wikipedia.org bug. Nevertheless, thanks for the good work. Sswonk (talk) 18:30, 20 June 2009 (UTC)[reply]
Google Chrome worked fine in my testing before the fix - so think what you will. In fact, the WebKit engine simply refused to attach the imported script to the page onload, for some reason unknown to me. Firefox, Opera and IE all worked as expected. It wasn't a Wikipedia bug, it was incorrect execution of valid code by WebKit. Which was rather annoying to fix, to say the least. Thus, while the iPod may be a work of genius, I now do hate Safari more than I did before (which is quite a lot :P). Ale_Jrbtalk 20:10, 20 June 2009 (UTC)[reply]

Date issues

It seems linked dates that are formatted without a space are not showing up properly: April 12001 ([[April 1]][[2001]]); April 1 2001 ([[April 1]] [[2001]]). I know we don't link dates anymore, but all the existing ones that lack the space look really odd in articles. Aboutmovies (talk) 10:50, 16 June 2009 (UTC)[reply]

In Aboutmovies' second example, a comma is being inserted between "April 1" and "2001", so the date autoformating is still functioning. What makes you believe date autoformatting ever inserted a comma and space in the first example? --Jc3s5h (talk) 18:26, 16 June 2009 (UTC)[reply]
I posted at WP:RFAR/DDL/Workshop and WT:MOSNUM to try to find someone that knew about this. When this had been discussed at WT:DATEPOLL#Tcncv's table, this appeared to be a recognized format. Also, when I've been editing articles, a common date format I've seen Advisor.js recommend fixing is the format that is now giving errors (by inserting a comma), though I hadn't noticed this problem until this post. —Ost (talk) 19:27, 16 June 2009 (UTC)[reply]
Besides my memory? Take a look at this version of one I corrected yesterday. Do you think an odd looking date would last for 6 months and a dozen edits? Or you can read Help:Date formatting and linking where it says "putting a comma and/or a space between the links, or starting month names with a capital gives the same result". Aboutmovies (talk) 19:31, 16 June 2009 (UTC)[reply]

You are correct. The software would previously add a space if the links are adjacent, plus add a comma if missing from the "U.S." date format, plus remove it if present in the other (day-first) format. I have no idea why it has ceased to do this, even if you turn your date prefs on. I'll note that a lot of surprising software changes recently took effect at the same time. — CharlotteWebb 21:47, 16 June 2009 (UTC)[reply]

These dates may eventually get unlinked or formatted differently, but I filed bugzilla:19258 for the problem. —Ost (talk) 13:37, 17 June 2009 (UTC)[reply]

Finding all edits a user made to a certain page

(cross-posted from the help desk) Hi there. I think I remember that there is a script somewhere that gives you the diffs for all edits a specific user made to a selected page but I cannot find it again. Does someone know of that script and where it's located (if I'm not making this up of course)? Regards SoWhy 15:59, 16 June 2009 (UTC)[reply]

I've certainly never found one; I'm currently coding that exact tool right now, as it happens! I'll let you know when it's done... haz (talk) 19:51, 16 June 2009 (UTC)[reply]

I already have one of these. Hold on, let me find it. — CharlotteWebb 20:36, 16 June 2009 (UTC)[reply]

See User:CharlotteWebb/history filter. Very much a work in progress, though I've made no progress lately. — CharlotteWebb 21:23, 16 June 2009 (UTC)[reply]

You were probably thinking of the Per-page contributions tool. Graham87 00:54, 17 June 2009 (UTC)[reply]
Yes, thank you Graham, that is it :-) Regards SoWhy 07:21, 17 June 2009 (UTC)[reply]
This is also possible with the API: rvuser=Splarka. I believe this requires an index to be built, that exists on en.wp but not very many other projects. --Splarka (rant) 07:47, 17 June 2009 (UTC)[reply]

What I've been wanting is some way to get the combined contributions to the same article by multiple users. I realize I could do adapt my script to do that by just making a separate query for each user, then sorting them by time-stamp. I'm sure I could figure out how to do that but it would be more complicated. I asked about this possibility on the mailing list here but got no response. — CharlotteWebb 15:33, 17 June 2009 (UTC)[reply]

Sorry for my tardiness, I replied on the mailing list. Splarka's wrong about the index: rvuser works on all projects, it's rc'user (in list=recentchanges) that doesn't; Tim is supposed to have added the required index for that recently, though, so I'll look into finally enabling rcuser. --Catrope (talk) 07:53, 19 June 2009 (UTC)[reply]

Wikipedia and caching

Hello everyone. I would like to know how wikipedia's server-side caching works for HTML Pages (not client side caching, or caching of non html pages). Wikipedia:Purge seems to indicate that HTML server-side caching is done (optionally) for registered users too. I was wondering how that was possible since every page is generated differently for every registered user (for example, there is the user name in the top right corner which is of course different from user to user).

So, does it "cache" only "some parts" of the page..? How can I do that? KoperTest (talk) 17:37, 16 June 2009 (UTC)[reply]

Yes, for logged in users, the rendered WikiText is cached, which excludes the user skin. The cached HTML for this page is available at Wikipedia:Village pump (technical)?action=render.
I'm not quite sure who assembles the page though, in particular since even logged out users will still get the "you have new messages" banner. There has to be some database interaction happening somewhere. Amalthea 18:18, 16 June 2009 (UTC)[reply]
mw:Manual:Cache explains the different layers of caching in MediaWiki. --brion (talk) 15:59, 17 June 2009 (UTC)[reply]

Regex question

(I have the regex gadget installed above the edit window).

Below is a watchlist for use with Related changes. How would I use regex to add the corresponding talk page to the end of every entry on the list?

Wikipedia:WikiProject Outline of knowledge/Watchlist using Related changes

I look forward to your reply.

The Transhumanist    21:24, 16 June 2009 (UTC)[reply]

Not sure how that works exactly but you'd want to do something that has the effect of this, where txt is the content of the edit-window:

txt = txt.replace(/\n\*\s*\[\[([^\]]+)\]\]/g, "\n*[[$1]] ([[Talk:$1|talk]])");

// or better yet if you want a bunch of other links use a template
txt = txt.replace(/\n\*\s*\[\[([^\]]+)\]\]/g, "\n*{{article|$1}}");

That would work for the article pages anyway. The other stuff would be more complicated. — CharlotteWebb 21:38, 16 June 2009 (UTC)[reply]

Hold on a sec... I think that I can do this. I've done it with watchlists before, thanks to the handy {{swl}} template. –Drilnoth (T • C • L) 22:05, 16 June 2009 (UTC)[reply]
 Done; just changed it to use the template with two simple regular expressions. –Drilnoth (T • C • L) 22:08, 16 June 2009 (UTC)[reply]
It looks like it's now done, but if you wanted to do it with the Regex-tab script, you could have replaced
\* \[\[([^\]\[]*)\]\]
with
* {{swl|$1}}
which is effectively what Drilnoth did on the page in question. The '\[' means a literal bracket character, the '\*' is a literal asterisk, the '[^\]\[]' means any character other than brackets, and the parentheses saves the match so it can be referenced later as '$1'. I hope this makes sense. Plastikspork (talk) 00:04, 17 June 2009 (UTC)[reply]
Thank you. That helps a lot! The Transhumanist    19:06, 18 June 2009 (UTC)[reply]

WhatLinksHere?

I'm having an issue with the WhatLinksHere page. Clicking on the link to go to the next 50 pages works the first time, but subsequent clicks just reload the same 50 pages (#51-100). The "previous 50" and "next 50" links are showing the same &from and &back numbers. I've confirmed this while logged in and logged out on two computers now. Can other people reproduce the problem? Does some new change to MediaWiki need to be reverted? Dekimasuよ! 02:32, 17 June 2009 (UTC)[reply]

Confirmed. A temporary substitute would be to hard-code the numbers in the URL or to expand the # of results shown. –Drilnoth (T • C • L) 02:47, 17 June 2009 (UTC)[reply]
I can't find any MediaWiki edits that could have done this. Hmm. Dekimasuよ! 03:20, 17 June 2009 (UTC)[reply]
Already being worked on; mostly fixed in SVN. [7] Ale_Jrbtalk 13:11, 17 June 2009 (UTC)[reply]
Thanks for the link and update. Dekimasuよ! 13:41, 17 June 2009 (UTC)[reply]

Need to find the name of a particular tool...

Can anyone help me with the name and location of the tool/program I described here? I saw it used a couple of weeks ago in the case of someone with a long history of uploading copyvio content and I believe that it may be useful in dealing with the current AnyBot situation. Thanks. --Kurt Shaped Box (talk) 04:27, 17 June 2009 (UTC)[reply]

Gee, I'm really sorry, but I have absolutely no idea what tool you're referring to. I did a search all over the site, and couldn't find anything. Are you sure you're not mixing this up with something else, and it doesn't really exist the way you described it? Beyond that, is it possible to simply check your browser history and see if it is in there? If not, then I don't know how to help you. Whatever it is, however, it's probably something on the toolserver. I'll get back to you if I find anything, though. The Earwig (Talk | Editor review) 04:33, 17 June 2009 (UTC)[reply]

API on the secure server

Resolved

Why can't the api.php on the secure server be accessed anymore, e.g. https://secure.wikimedia.org/wikipedia/en/w/api.php? Bug or feature? This obviously makes NAVPOP, Twinkle, and probably other scripts fail there, or will force them to break encryption.
From a glance, I don't see what the problem is anyway. Amalthea 09:39, 17 June 2009 (UTC)[reply]

I fixed it. — Werdna • talk 21:18, 17 June 2009 (UTC)[reply]

Thanks! Amalthea 22:54, 17 June 2009 (UTC)[reply]

Vector skin

I saw a report in the Signpost at the beginning of the month about a new skin called "Vector" which was supposed to be available as of revision 51094 of MediaWiki. We're now up to 1.43.0-wmf.17 (21482f8), but I still don't see it in "my preferences". Does anyone know what happened to this skin?-gadfium 06:38, 18 June 2009 (UTC)[reply]

It is still in development. It works partially with useskin=vector. See The Usability Wiki. --Splarka (rant) 07:05, 18 June 2009 (UTC)[reply]
Okay, thanks.-gadfium 07:07, 18 June 2009 (UTC)[reply]

Styling tags

Now that MediaWiki has been updated past r52071, Tags are now wrapped in a span which allows us to identify them. There is now an open discussion on whether we should style tags when they appear in RecentChanges, Watchlist, etc. All commens welcome! Happymelon 09:54, 18 June 2009 (UTC)[reply]

Did you mean...?

I was searching around for the Wikimedia-wide Terms of Use, regarding the switch to cc-by-sa, and I ran this search. I was kind of surprised to see what the "Did you mean...?" suggested. Is this, er, expected behavior? It does play into the stereotypes about Wikipedia pretty well. – Quadell (talk) 13:57, 18 June 2009 (UTC)[reply]

The 'Did you mean:' function is powered by the current Lucene search version (extension). I believe the suggestion is based on a rather complicated algorithm that finds the most similar word(s) that is(are) regularly searched for - or something along those lines. Perhaps an exception list would be a good idea though - you probably wouldn't see Google search make that sort of mistake... Perhaps Bugzilla it? Ale_Jrbtalk 14:07, 18 June 2009 (UTC)[reply]

Er... what is the point? ╟─TreasuryTaghemicycle─╢ 18:32, 18 June 2009 (UTC)[reply]

You are not the first to wonder. At Wikipedia:Help desk/Archives/2009 May 14#Just wondering... Algebraist wrote: According to the release notes it is 'for benchmarking, etc.' PrimeHunter (talk) 19:10, 18 June 2009 (UTC)[reply]
There's a point to it, involving what a page looks like / is without content. It's for the devs rather than the visitors, naturally. It came up a few weeks back I think. - Jarry1250 (t, c, rfa) 19:51, 18 June 2009 (UTC)[reply]
Thankfully, there's an encyclopedia that has an article that explains it in part. ;) EVula // talk // // 19:54, 18 June 2009 (UTC)[reply]

I think it's just a joke. There is no bug for it, and the commit logs don't explain it. --- RockMFR 20:06, 18 June 2009 (UTC)[reply]

If you want to benchmark the performance of your site/server, having a single special page lets you easily filter the tests out of your logs. More importantly, if you want to debug without having your target do anything fancy that might interfere, now you can[8][9] (view the source). I guess Domas used to use Special:Version for this purpose.[10] :-) —Emufarmers(T/C) 07:56, 19 June 2009 (UTC)[reply]

Author's name in the cite book template

Please see Hi-Fi Murders. The link to Gary Kinder used in the cite book template goes to the wrong article. How does disambiguation work in that template? Just add the disambiguation phrase in the template? Who then was a gentleman? (talk) 20:56, 18 June 2009 (UTC)[reply]

Just change the authorlink param to any value you like, really. - Jarry1250 (t, c, rfa) 21:03, 18 June 2009 (UTC)[reply]
That worked, thanks. Who then was a gentleman? (talk) 21:08, 18 June 2009 (UTC)[reply]

anyone available to write a simple history merge script?

For the links below, look instead at Wikipedia talk:New histmerge list.

it seems we've got a metric shit-tonne of cut&paste moves to prepare. We're going to get a bot on it, but there will still be ones that need human eyes on 'em.

Can someone write a simple (or complicated, up to you) history merge script? At the bare minimum it should move the page over the target location deleting it, then restore the deleted edits.

I thought this would be a nice addition to Twinkle, but Amalthea unfortunately has RL aggro. I'm sure you could build it with the Twinkle framework.

If you want to get fancy, you can check for already-deleted edits so nothing uncanny comes back in the restore. If this is too complicated, we'll just put a disclaimer to check the target for deleted edits first. Thanks in advance, ye code wizards. –xenotalk 21:48, 18 June 2009 (UTC)[reply]

I use this method for history merging, which is less risky and kinder on the servers. I've made some more comments at [[User talk:Anthony Appleyard/New histmerge list. Graham87 09:55, 19 June 2009 (UTC)[reply]
^^ I'm fine if the script uses that method instead as it's lighter on the servers. Though, MZMcBride also raised a good point at the BOTREQ:

I don't usually like history merges in general. If there's any overlap between the two pages, the page history gets mangled, diffs become confusing and unreadable, etc. There's a core feature (disabled by default) that was written by one of Wikimedia's contractors to deal with split histories safely (see "#$wgGroupPermissions['sysop']['mergehistory'] = true;" in DefaultSettings.php). If possible, I'd much rather seem that activated than an ad hoc bot solution. For what it's worth. Perhaps can someone can convince me this is a good idea in a bot request given appropriate safety measures.... --MZMcBride (talk) 02:46, 19 June 2009 (UTC)

Any reason this isn't activeated? –xenotalk 12:47, 19 June 2009 (UTC)[reply]
I don't believe it's been extensively tested. ^demon[omg plz] 13:15, 19 June 2009 (UTC)[reply]

"undo" for move entries in history

Why do history entries for article moves have an undo link when that link doesn't actually revert the move? dramatic (talk) 00:52, 19 June 2009 (UTC)[reply]

Guess it is like the move tab in filespace— it might work someday. ---— Gadget850 (Ed) talk 11:36, 19 June 2009 (UTC)[reply]
There isn't a move tab in filespace, since that feature was deactivated. –Drilnoth (T • C • L) 18:26, 19 June 2009 (UTC)[reply]
Ah- I have the "Add page and user options to drop-down menus on the toolbar." gadget checked. I have Page → Move page on all but Commons images. ---— Gadget850 (Ed) talk 18:35, 19 June 2009 (UTC)[reply]
Gotcha; no that isn't set by default. –Drilnoth (T • C • L) 18:41, 19 June 2009 (UTC)[reply]

Permission

I'm sorry, but please tell me, where can I find the text of permission to be written on a webpage with free pictures that may be used for Wikipedia under the licence CreativeCommons Attribution ShareAlike? My colleague Sergey Zagraevsky is about to give such permission for any photo from his medieval architecture photo collection and I promised to send him the text of such permission. --Ozolina (talk) 18:58, 19 June 2009 (UTC)[reply]

Go to http://creativecommons.org/license/ and make sure you select 'Yes' to commercial uses and 'Yes, as long as others share alike' to modifications. Algebraist 19:12, 19 June 2009 (UTC)[reply]
Thank you! I've sent to Dr. Zagraevsky this information and hope that he will soon place in at his site so we shall be able to take necessary photos to Commons. --Ozolina (talk) 19:41, 19 June 2009 (UTC)[reply]

Scary behaviour

Something really weird just happened: I started up my computer from hibernation. Some Wikipedia page was still open in a browser window. After clicking on "my watchlist", I found myself logged in as User:Malafaya! The window displayed his watchlist, on which he has 10 pages, under "my contributions" I saw Special:Contributions/Malafaya. I guess I would have been able to change his password and usurp his account if I wanted to ... but I didn't and logged out. Is this some know bug? It's certainly not a behaviour one would expect ... -- Momotaro (talk) 20:31, 19 June 2009 (UTC)[reply]

It happens occasionally, when there is a "session collision": purely by coincidence, the cookies you got when you logged in (which contain random number strings) happened to match exactly the cookies Malafaya got when he logged in; when you give the cookies back to the software when doing later edits, there is a chance that the software will get confused and swap you over. I do believe, however, that it is occuring more frequently in recent time than it has in the past. Maybe the cookies need to be upgraded to contain more random data to reduce the probability of collisions. Happymelon 20:49, 19 June 2009 (UTC)[reply]
I have a hard time believing this is true. Is this behavior documented anywhere? --MZMcBride (talk) 20:54, 19 June 2009 (UTC)[reply]
I remember at least two other people in the last couple months reporting incidents just like this one here. Accidental session id collisions should really be nigh impossible though. Anyone knows a bug id on this? Amalthea 21:30, 19 June 2009 (UTC)[reply]
I was thinking the same thing. Too many people have reported it for nothing to be happening, but I also don't think that the session ids are colliding - I suspect another issue of some kind, personally. Ale_Jrbtalk 21:41, 19 June 2009 (UTC)[reply]
There is of course the birthday “paradox” which, more or less, says that collisions are more likely to happen than one would expect by naïve intuition. But still … I also find it hard to believe this happens, and even occasionally, as it seems. Does anyone know how long the random numbers in the cookies are? -- Momotaro (talk) 22:17, 19 June 2009 (UTC)[reply]

(undent) They are 128 bit MD5 hashes, for example for Momotaro they might be something like:

centralauth_Token=acbd18db4cc2f85cedef654fccc4a4d8;
centralauth_User=Momotaro;
enwiki_session=37b51d194a7513e45b56f6524f2d51f2;

This does seem highly extremely unlikely to be a token collision. More likely they are just getting a session for the wrong user from some other bug...? --Splarka (rant) 00:28, 20 June 2009 (UTC)[reply]

Happened to me back in February: [11]. I'm at least glad to see that I'm not crazy; after a few weeks, I really started to question my recollection of what happened. --Floquenbeam (talk) 00:44, 20 June 2009 (UTC)[reply]
This is bug 19158. Mr.Z-man 01:03, 20 June 2009 (UTC)[reply]

They're not md5, they're randomly-generated. — Werdna • talk 10:34, 20 June 2009 (UTC)[reply]

Thanks for all your answers. There are 32 digits, which means indeed 128 bit. If I calculate correctly ([12]?), assuming there are about 100 000 active sessions (or even 10 billion), the chance of a random collision should indeed be tiny, so maybe the collision happens somewhere else in the software. However, according to the bug report, a fix seems to be coming. -- Momotaro (talk) 17:48, 20 June 2009 (UTC)[reply]
It's probably just the usual problem of "random" data not being as random as it should be. E.g. with a random number generator that has a relatively short loop, or that is reset to the same seed value each time the server restarts. Hans Adler 18:47, 20 June 2009 (UTC)[reply]

This has happened to me as well. On 2 June, switching tabs, I found myself logged into Kroshkmd.  Roger Davies talk 19:33, 20 June 2009 (UTC)[reply]

Page layout errors

The layout of Wikipedia pages has recently changed to that every link on a page now appears by itself on its own line, which looks awful. The "printable" version of a page looks the same as always, however. Have you begun using some advanced HTML function that is not supported by my old browser (Microsoft Internet Explorer Version 5.00.2314.110031C, running under Windows 95)? —Preceding unsigned comment added by 207.69.248.250 (talk) 21:42, 19 June 2009 (UTC)[reply]

Confirmed with browsershots.org - IE 5.01 / Windows 2000. IE 5.5 works though - [13]. (IE 4 is broken in an even more interesting way). Mr.Z-man 23:11, 19 June 2009 (UTC)[reply]
rev:51807 added some more "child combinator" selectors to shared.css. IE5 has a horrible regressive state (since child combinators didn't exist even in draft) whereby, rather than ignoring this CSS2 property, it parses it as a comma.
#p-logo a {display:none;}
This would parse in both IE5 and lets say "good" browsers, as 'Hide any link descendants of the logo".
#p-logo > a {display:none;}
In good browsers, this would just hide links that were direct children. In IE5 however, it basically sees:
#p-logo, a {display:none;}
Which is "hide any object with id="p-logo", and any links."
This means per 51807 all LI are being applied the styles float: right;, and links display: block;. This could either be fixed in IE50Fixes.css, or by convincing Trevor Parscal to give the objects better classes and not refer to descendants. Note: I too have IE5 but I never use it. --Splarka (rant) 00:11, 20 June 2009 (UTC)[reply]

Italic titles

Why aren't some ways of italicising titles working at the moment? Articles like Methicillin-resistant Staphylococcus aureus with: <span id="RealTitle" style="display:none">Methicillin-resistant ''Staphylococcus areus'' </span> and articles that have {{taxobox}} with the "name" parameter (Paracerceis sculpta) aren't working at the moment. {{italictitle}} used to italicise the Homo in Homo (genus), but doesn't anymore too. What is wrong? Smartse (talk) 22:30, 19 June 2009 (UTC)[reply]

See request for template edit at Template talk:Taxobox title, and the recent repair to {italic title}. Celefin (talk) 22:46, 19 June 2009 (UTC)[reply]
Just to clarify, this recent change [14] removed the RealTitle system. For flexible title italicisation, it is now necessary to use the magic word DISPLAYTITLE, for example:
{{DISPLAYTITLE:Methicillin-resistant ''Staphylococcus aureus''}} Celefin (talk) 22:54, 19 June 2009 (UTC)[reply]
I think we need guidelines for use of DISPLAYTITLE. There is some discussion at MediaWiki talk:Common.js#Technical restrictions code ready to be removed about implementing it as a template. ---— Gadget850 (Ed) talk 17:05, 20 June 2009 (UTC)[reply]

Collection/create-a-book extension tweaks

Just a quick note y'all -- we've implemented bugzilla:18902 so we can limit saving of collections to autoconfirmed users and above. Will be peeking at other issues with it over the next weeks. --brion (talk) 22:50, 19 June 2009 (UTC)[reply]

Announcement of pending Checkuser and oversight elections

The Arbitration Committee has determined that there is a need for further oversighters and checkusers to improve workload distribution and ensure complete, timely response to requests. Beginning June 20, 2009, experienced editors are invited to apply for either or both of Oversight or CheckUser permissions. Current holders of either permission are also invited to apply. Voting will begin on July 28, 2009 and close on August 10, 2009. Further information, including instructions for application and a complete timeline, are available at Wikipedia:Arbitration Committee/CheckUser and Oversight elections/August 2009. For further information about the Committee's relevant resolutions, please see the Arbitration Committee noticeboard.

For the Arbitration Committee,
Risker (talk) 02:53, 20 June 2009 (UTC)[reply]

Customise editing-toolbar or special-characters-list

Is it possible for me to remove some of the buttons that I don't often use (such as ) and replace them with buttons to paste in my own custom text, something that I'd find useful?

Or could I do something equivalent with the – — … ‘ “ ’ ” ° ″ ′ ≈ ≠ ≤ ≥ ± − × ÷ ← → · § toolbar beneath the edit-window? Thanks! ╟─TreasuryTagprorogation─╢ 18:59, 20 June 2009 (UTC)[reply]

Could probably be scripted without too much difficulty; I'll look into trying to write something, because I also think that that would be more useful. –Drilnoth (T • C • L) 19:13, 20 June 2009 (UTC)[reply]

Preferences changed by them selves

My settings for Image size limit and Thumbnail size (under Appearance) both changed to the smallest value, but I didn't change them. Any idea what happened, and what are the default values? --Apoc2400 (talk) 20:27, 20 June 2009 (UTC)[reply]