Wikipedia:Village pump (technical)

 Policy Technical Proposals Idea lab WMF Miscellaneous 

• Table of contents
• First discussion
• End of page
• New post

The technical section of the village pump is used to discuss technical issues about Wikipedia. Bugs and feature requests should be made at Bugzilla (see how to report a bug). Bugs with security implications should be reported to securitywikimedia.org or filed under the "Security" product in Bugzilla.

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.

« Archives, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 191, 192, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213

Multiple bullet points on a line

Bullet point behavior on Wikipedia is unintuitive, and can become an intricate procedure when used during discussions. Responding well to a bullet point comment means placing your comment on the very next line with no break in-between, and with, at most, one more asterisk than the line above.

• Hello, how are you?
  • I am fine, thank you.
    • You are most welcome.
      • Have you guys read WP:NOTFORUM?

One must take care not to stray from these rules, lest one look like an idiot.

• Hello, how are you?

• • I appear to be a moron, thank you. left a blank line!
    • • • Wikipedia can help you pick out brain medicines, as it has for me. too many asterisks!

I have literally never come across an instance where placement of multiple bullet points on a line before a statement was intended, nor could I fathom when that might be beneficial. Can we fix this behavior and make bullet points easier to use, by doing away with these silly requirements? Let's just have multiple asterisks produce a single bullet with the requisite number of indents before it, in all cases. PS. I realize some might say bullets should just not be used in discussions, but they will be regardless. PPS. I also realize this may be some unintended technical issue, but if so I'm hoping we can agree it might be worth fixing? equazcion → 04:43, 24 Oct 2013 (UTC)

• I believe this has come up before, and I know there is a workaround to avoid this behavior.

• Hello, how are you?

• I appear to be a moron, thank you. left a blank line!

• Wikipedia can help you pick out brain medicines, as it has for me. too many asterisks!

Replacing all of the leading asterisks with colons avoids the issue. Whether or not this should be fixed in the core, is up to the developers to decide (and I will almost bet there have been tickets submitted on Bugzilla about this behavior), and you may want to ping them or post a ticket on Bugzilla... Technical 13 (talk) 04:54, 24 October 2013 (UTC)

Line breaks between list items (especially those involving bullets, like the last two examples) cause problems for screen reader users like myself. Indenting using a mixture of colons and asterisks (especially when the colons come first) also creates some hideous HTML, which is presented literally by screen readers. Graham87 08:38, 24 October 2013 (UTC)

Were it up to me, I'd have a character that served the same function as : or * that would be active only in talk pages but that didn't create html lists to implement a discussion. Very often, lengthy discussions become wall-of-text-like when one writer follows immediately after another without a line break. This I think is an encumbrance on understanding.

I don't know how to implement such a mechanism because I haven't given it any more thought than what I've described. If such a mechanism is possible, it would be a solution to both Editor Graham87's and Editor Equazcion's issues. And, it would keep other editors from harping on me for adding line breaks to my contributions to talk pages.

—Trappist the monk (talk) 14:32, 24 October 2013 (UTC)

Quick guide to the "right" way to reply to talk posts: 1. Never put a gap between posts. 2. Copy the asterisks (*), colons (:) and number marks (#) from the post to which you are replying, then add one symbol of your own.

<p style="margin-left: 1.5em">
Indented without list markup
</p>

I agree with Trappist regarding the need for better markup; the way to indent text without abusing list markup (shown right) is too painful for use in discussions. Another wiki I edit uses tildes for this purpose, but MediaWiki already uses tildes for signing posts. The developers are unlikely to add such markup because the upcoming Flow will automatically indent posts.

Regarding the wall-of-text problem, I have the following in my common.css to resolve this:

dd { margin-bottom: 0.5em; }
dd:last-child { margin-bottom: 0; }

This also adds extra space to definition lists in articles, but I like the extra spacing there too.

Part of the problem of multiple bullets is down to MediaWiki not requiring editors to mark the start and end of a list, to simplify markup. MediaWiki tries to figure out where lists start and end by seeing if an asterisk is matched at the same position on the next and previous lines. If there are multiple asterisks not matched on the previous line, MediaWiki takes it to be the start of multiple bulleted lists nested inside each other, hence multiple bullets appear. If the previous line is blank, all of the asterisks start a new list, so all cause a bullet to show. – PartTimeGnome (talk | contribs) 20:28, 24 October 2013 (UTC)

• Multi-indents and need for relative ":+" indentation: We should have relative-indentation (of the form ":+" as one more tab stop than the above msg), but there is a common case where a message is followed by a double-indented response, when interjecting a 2nd response above a prior (often lengthy) response:

           : This is the original message.
           ::: This is the 2nd, but interjected, response.
           :: This is the first response.

Such cases, of double-indented lines (2 extra colons "::"), are common when someone thinks they have a more-direct response than the immediate reply had stated. In a relative style, it could be ":++" to indent by 2 levels more than the prior msg. For relative indentation as ":+" then the parser would need to be smart about remembering the indentation level of the prior text, and when too complex, just show as relative to the margin (so then ":+++" would be same as ":::"). Of course, if templates had the parser extension for global variables, then we could write a smart template which kept a counter and indented by "{{in+}}" as a template which increments the global variable for indentation in the talk-page. Unfortunately, {:+} will transclude page "+" into a discussion, so that would be even more confusing. At least we have {od}, which pretends to be smart enough to know wherever the last word was displayed, to handle the reverse outdent cases:

So, we have made some progress over the years. -Wikid77 (talk) 20:09, 24 October 2013 (UTC)

{{od}} is not that smart, nor does it pretend to be (what you describe cannot be done in a template). Using it without any parameters produces an outdent marker with a length equivalent to 10 colons of indent. If you want a different length, specify it as a parameter. E.g. {{od|::::}} will make an outdent marker equivalent to 4 colons of indent. – PartTimeGnome (talk | contribs) 20:36, 24 October 2013 (UTC)

Any solution to this is going to have to be local; the WMF devs are working on Flow, for talk pages, and that will eliminate this problem. -- John Broughton (♫♫) 17:37, 25 October 2013 (UTC)

That is probably the case, unfortunately. I do hope Flow is the miracle it's intended to be, but for the assuredly long time before it's actually implemented, we probably won't see any effort put into any development requests regarding talk pages. equazcion → 17:11, 29 Oct 2013 (UTC)

Drop-down option for saved edit summaries/subject headers

The drop-down option for saved edit summaries/subject headers etc. has vanished for the second time (previously fixed by unclicking 'Always use a secure connection when logged in' on 'Preferences'), and I can't seem to fix it. Any help welcome. GiantSnowman 11:44, 24 October 2013 (UTC)

Anyone? GiantSnowman 11:22, 29 October 2013 (UTC)

GiantSnowman Seems you're referring to the gadget (first one in the "Editing" section). I tried enabling it and it works for me. You may want to try disabling it, saving your preference, re-enabling it, and saving your preferences again. Also bypass your cache. equazcion → 07:01, 3 Nov 2013 (UTC)

Font color of article text

Has the font color of text in the body of Wikipedia pages changed? I think it used to be black, but now all pages appear to have grey text. Is this the case for anyone else? -- Toshio Yamaguchi 19:53, 24 October 2013 (UTC)

To expand a bit, I know that the font color used to be the same as the color of my signature (which is black). Now there is a noticeable contrast (at least for me). -- Toshio Yamaguchi 20:06, 24 October 2013 (UTC)

It's definitely black for me - MonoBook and Vector. --Redrose64 (talk) 20:46, 24 October 2013 (UTC)

Still grey for me (vector skin + Safari). -- Toshio Yamaguchi 11:15, 25 October 2013 (UTC)

Seems to be defined as #252525 in the style sheet. The printable versions seems to have changed as well. Not a fan. — Whisternefet (t · c) 20:45, 25 October 2013 (UTC)

I see it too. What is weird is that I can't find it defined anywhere in the source! Where the hell does it come from? — Edokter (talk) — 10:19, 26 October 2013 (UTC)

It comes from Vector's skin styles, and was changed – I presume accidentally – in https://gerrit.wikimedia.org/r/#/c/79948/ . I'll submit a patch to restore the previous value. — Preceding unsigned comment added by Matma Rex (talk • contribs) 11:00, 26 October 2013 (UTC)

[edit conflict] It comes from https://bits.wikimedia.org/en.wikipedia.org/load.php?debug=true&modules=skins.vector&only=styles&skin=vector, which is generated by LESS. It seems the new color was introduced by Jon (WMF) in gerrit:79948, but I believe it was supposed to be an opt-in beta feature. Helder 11:04, 26 October 2013 (UTC)

https://gerrit.wikimedia.org/r/#/c/92065/ Matma Rex talk 11:07, 26 October 2013 (UTC)

The color #252525 is present in the image of this document: mw:Wikimedia Foundation Design/Color usage#Color Coding/ What Colors Represent. Helder 11:17, 26 October 2013 (UTC)

The patch has been accepted and will be deployed on November 7 per mw:MediaWiki 1.23/Roadmap. The hack in MediaWiki:Vector.css should be removed then. Matma Rex talk 14:27, 31 October 2013 (UTC)

That is not a hack! That is perfectly valid CSS!! Shame!!! — Edokter (talk) — 11:24, 3 November 2013 (UTC)

Okay, okay, the workaround. ;) Matma Rex talk 11:29, 3 November 2013 (UTC)

DigiCert?

Everytime i come to wikimedia, i run a background sniffer, it detected that this site connects to DigiCert on ip (198.35.26.106) which isn't actually a wikimediaIP range (208.80.154.xxx), is this necessary?. It started recently, about a week or 2 back. I'm asking cause it uses around 252kb of data everytime I refresh the Watchlist or go to any page. I don't care about any of my pages being "safe" but I don't like unnecessary usage of Data as i'm on a Limited data plan and on wikipedia, you have to refresh your watchlist a LOT...Is there a way to disable this? Was this set in place after the recent "passwords compromised" debacle a few weeks back?--Stemoc (talk) 00:12, 26 October 2013 (UTC)

I notice that Wikipedia's SSL certificate is from DigiCert. Could it be your browser that is making this HTTP request? — This, that and the other (talk) 00:30, 26 October 2013 (UTC)

I'm on Firefox 24. If true, is there a way to disable it? cause when i do a search for the IP, it tells me its owned by Wikimedia Foundation Inc. it does eat a lot of unnecessary data and ignores the cache which means if i go to a page, it will get cached right, so when i close that tab and open it again, it should not eat the same amount of data it used in the first time since its cached but it does, which probably means that now wikipedia is running on https instead of http...--Stemoc (talk) 00:46, 26 October 2013 (UTC)

Open Tools -> Options or Edit -> Preferences (depending on your build), then go to Advanced -> Certificates -> Validation. If the first option about OCSP is on, then it's connecting to DigiCert to verify the certificate hasn't been revoked since it was issued. You can disable this if you want, but it will probably reduce your security and barely save any data at all. Jackmcbarn (talk) 03:42, 26 October 2013 (UTC)

Why do you say it is connecting to "DigiCert"? When I check whois it tells me that the IP range 198.35.26.0/23 is registered to Wikimedia, and doing a PTR lookup for the IP tells me the host name is bits-lb.ulsfo.wikimedia.org; in other words, it's the load balancer for bits.wikimedia.org at the new ulsfo caching datacenter. The 252kb is going to be the CSS and JavaScript, although your browser should be caching it so it isn't re-fetched on every page view. Anomie⚔ 01:38, 26 October 2013 (UTC)

Not necessarily connecting, but that IP is new to me and yes its registered to wikimedia, i'm only familiar with the (208.80.154.xxx) and the toolserver Ip ranges. Yes my sniffer tells me its for the 'bits-lb.ulsfo.wikimedia.org' caching datacenter but what i haven't noticed until today is that my url now has a https instead of http, It wasn't https a few days ago..the new server not only makes wikipedia run through a secured server but also prevents the browser from caching it....Is that really necessary? How do i opt-put of the "secured" option?...it is set to cache in my browser so that isn't the problem..--Stemoc (talk) 03:12, 26 October 2013 (UTC)

Hi (and welcome back I was under the impression that you left. Sorry.) Cometstyles. I think you can use http by going to Special:Preferences and deselecting "Always use a secure connection while logged in", but I would generally recommend that you use https. If caching is the issue, you should file a report on bugzilla: (I can do so if you don't want to). πr² (t • c) 04:12, 26 October 2013 (UTC)

Why did i not see that option? a slight reduction (from 272 to 140kbs), thanks...I don't know who you are lol, hehe..yeah i didn't completely leave, i created a lot of articles and i had to come around from time to time to update and fix them so i decided to create an account to look after them..thanks for the option..I'm not worried about "security" I only use wikipedia from home from my PC which is hard to hack..no need to file a bug...Thanks though :) ...--Stemoc (talk) 04:53, 26 October 2013 (UTC)

Your computer being hard to hack isn't relevant. HTTPS prevents people on other computers intercepting the login token your browser sends to the Wikimedia servers, by encrypting everything sent on the connection. The login token can be used to impersonate you on Wikimedia wikis until you log out, so it's a bad thing if someone else can discover it.

The risk is particularly high if you use Wikipedia over Wi-Fi. Even without Wi-Fi, you don't know how secure the other systems used to connect to Wikipedia are. (Using traceroute, I see my requests to Wikipedia pass through four computers before reaching Wikimedia. This will probably be different for you, depending on your ISP.) – PartTimeGnome (talk | contribs) 20:39, 26 October 2013 (UTC)

I ran a trace route on mine, went through 21 computers and yet i still feel secured. We were all on an unsecured method till last week or so when they changed the url to a secured server. WiFi is not common in my country and those that are available are protected. My IP hopped between 2 other countries before reaching USA...I think others need to be notified of the new changes, though most in developed countries wouldn't care but those like me in a very much 3rd world country with limited data would..--Stemoc (talk) 01:38, 27 October 2013 (UTC)

Some of us have already been using HTTPS for years. ;) πr² (t • c) 03:05, 2 November 2013 (UTC)

• When I download Main Page over HTTPS, Firefox 24 makes two POST requests to http://ocsp.digicert.com/ and receives two 200 OK answers. Both the requests and the answers have Content-Lenght < 500 bytes. When I tested this, I first cleared all browser data (cookies, history, cache et cetera) and then typed in the full address manually. Are you doing something differently, and if so, what are you doing differently? --Stefan2 (talk) 16:37, 26 October 2013 (UTC)

WMF notification email marked as spam by Yahoo

Tracked in Phabricator
Task T54915

An email to me with the subject "You have a new notification at Wikipedia" (without quotation marks) sent October 23 or 24, 2013 (seen today), from wikipedia.org at IP 208.80.152.133 (involving terbium.eqiad.wmnet and mchenry.wikimedia.org) was treated by Yahoo as spam. (I have omitted sender and intermediate email addresses in here per a WP policy or guideline but can supply them if desired.) That may mean that the sender's mailing list administrator is not deleting bounced addresses after a second bounce, if that's still the current standard, or it may mean something else. It's not a problem for me, but it would be for many other recipients, because Yahoo sends such email into a spam folder instead of the regular inbox, and many users may not check the spam folder. Nick Levinson (talk) 20:03, 26 October 2013 (UTC)

• @Nick Levinson: I was going to suggest posting your complaint on the Y-Mail Yahoo Group, but when I went to get the URL for you, I saw the on-going thread Damn Yahoo Now Blocking Legit. Emails (5) right at the top. (which reminds me I should check to see if Wikipedia has removed the block it has put on the Yahoo Groups feedback forum ) XOttawahitech (talk) 15:45, 28 October 2013 (UTC)
• It would be useful to know if it was an HTML notification and the full headers of the email, or at least the lines containing mention of SPF, DKIM and (maybe) spam. --Nemo 14:17, 29 October 2013 (UTC)

I'm not sure how to tell if it was an HTML notification, but my Preferences > Notifications > Email format is set to Plain text, not HTML, and I haven't changed it recently.

In the full headers, the string "spam" does not appear but the string of interest is "bulk" (without quotation marks).

The full headers follow, except that in my email address I replaced my username with an ellipsis and a bracketed phrase, I replaced each leading space with the Wikipedia/HTML code for a nonbreaking space (to prevent stripping (that code otherwise did not appear in the headers)), and I replaced each leading tab with eight of the same code for nonbreaking space (there weren't that many leading spaces in any line of the headers). Nick Levinson (talk) 15:43, 30 October 2013 (UTC)

From Wikipedia Wed Oct 23 17:00:04 2013

X-Apparently-To: ...[my username in lower case]@yahoo.com via 98.137.13.238; Wed, 23 Oct 2013 17:00:06 -0700

Return-Path: <wiki@wikimedia.org>

X-YahooFilteredBulk: 208.80.152.133

Received-SPF: pass (domain of wikimedia.org designates 208.80.152.133 as permitted sender)

X-YMailISG: F47yZzkWLDtlR_AriCvzoE.WTWTPeM5e5YC71mQBjkzJqZ5r

 1FAHJYvs3bo2LXSr2.S_CSmahNgC1b5_Pcb5bWVxUID.meGYITRDR9l7woBc

 DEnm5Hr6zIOKxZ2EvEKHlgEModtMGKKHWrOrKZAHAsGtH7r66OJCyrjERfhl

 jEdPh488czYgJgikgyXuA4lD66fSThMeQpSMgcvavU5fyxAuFnVCCsbFumlU

 1QIjBC30csrKqIooS5i56SMd43tkIzdKHQ6wMD7uGKWG4kdXULB8D66piH_Z

 .7tjyYQ8Zx8_W7tTj27PwzMhWgOqVoLcE0_olyuip4UbUSoC2rem_qOcj7Gx

 usud1_CDQM7dRo7auOaAHhRFj8nJM2p8l_G1KTbYoZUDpHld2Xj58GW4lIMe

 YI8boWZgGzxmrUoaASSyMnF8z1bOoiC4VTRUhC17BCuZYa6KWZ92MrBaq4kV

 YT_iKXL4j1TJa4QNW7PUSZcT_mJ85Yx90_nROe3rKsrQeggFcOmuHotLLyCE

 aoJrHgyy6ODW9Q6b0j3Ap08X4ZfK6FP9JehuX0M0OGbgYSJinLB3NR3ttwux

 sujSCpieV2Ul.l_Px9x5583.2QopXKO8qmFcIW2cGuEr3frPBRY5GPnZIDXl

 Ts24fMozrVqbxiEEkd01qq_q6GPg2vupOlAl.OPE5bZ6vziMDDCyxMNMnuhZ

 rtPA8gJzaAl8XIiSEhfjUiDGzlWNRHVu41VUi6YGwYe2WYxOReNApxUgD_iG

 LVBakKmK05k3STAR_mLpfl1ufrh1zT3xGrQ_zOXJhYNOqd30Lst20J2so_qz

 vm23AetJkXTb9goXUUhzEpwympikO8Ab90NDfzh7Pc_1tjjxQB6TLmMz2Nm3

 cXp8qkJtCHFbdMI0Msrl2ZNGINLgw9g6iw_EgURU6CJ.Zi9Zs0sLAuFqj9Ej

 x6xlB1MrdadEAFJvHtbhei6IzaxOuz1XQBB_9KIZ6nkjlXhrBBvTrlsbsWKp

 07t2u7rCC7oONTKsovBXZiKzusbg_jO8ciFQw5QuHMRRuoszLmMTrryky5kz

 avT7YRpepLTTRyFklmLXml7KWNGL8OVZAzMiM.KN3KUJH8hRwXxl7kuSsXWW

 nZAHLJqZE8op7VCuYbM4s_A_ukKgiYs84tz0OgfGuuyNs15OQTw-

X-Originating-IP: [208.80.152.133]

Authentication-Results: mta1066.mail.gq1.yahoo.com from=wikipedia.org; domainkeys=neutral (no sig); from=wikipedia.org; dkim=neutral (no sig)

Received: from 127.0.0.1 (EHLO wiki-mail.wikimedia.org) (208.80.152.133)

  by mta1066.mail.gq1.yahoo.com with SMTP; Wed, 23 Oct 2013 17:00:05 -0700

Received: from terbium.eqiad.wmnet ([10.64.32.13]:42340)

        by mchenry.wikimedia.org with esmtp (Exim 4.69)

        (envelope-from <wiki@wikimedia.org>)

        id 1VZ8LM-0005so-Cw

        for ...[my username in my casing]@yahoo.com; Thu, 24 Oct 2013 00:00:04 +0000

Received: from apache by terbium.eqiad.wmnet with local (Exim 4.76)

        id 1VZ8LM-0001QQ-8U

        for ...[my username in my casing]@yahoo.com; Thu, 24 Oct 2013 00:00:04 +0000

To: Nick Levinson <...[my username in my casing]@yahoo.com>

Subject: You have a new notification at Wikipedia

From: Wikipedia <no-reply-notifications@wikipedia.org>

Reply-To: No Reply <no-reply-notifications@wikipedia.org>

Date: Thu, 24 Oct 2013 00:00:04 +0000

Message-ID: <enwiki.526863043e2435.77588364@en.wikipedia.org>

X-Mailer: MediaWiki mailer

MIME-Version: 1.0

Content-type: text/plain; charset=UTF-8

Content-transfer-encoding: 8bit

Content-Length: 464

Thanks! If you don't see HTML, the email is not in HTML. :) "Content-type: text/plain" is what tells you. Your headers give me a couple things to investigate on, I'll add more to the bug linked above. --Nemo 07:36, 31 October 2013 (UTC)

Requests to dezoomify an image

I was asked to upload PD [1] (direct link to zoomified image: [2]). image to Commons, but it uses the evil zoomify javascript. I tried to use the tools listed on commons:Help:Zoomable images, but they either require skills I don't have (ex. Python, Ruby + ImageMagick, etc.) or are broken ([3] by User:kolossos gives 404, gee, toolserver tool that is not maintained, what a surprise; [4] was loading an image, then gave me a black screen; upon a second run it froze at 92%). As it appears using those tools requires skills (or luck) I don't have, perhaps someone here will have a better luck. --_{Piotr Konieczny aka Prokonsul Piotrus| reply here} 02:46, 28 October 2013 (UTC)

I've managed to get the image successfully. Can you post what I should put on its description page when I upload it? Jackmcbarn (talk) 03:21, 28 October 2013 (UTC)

@User:Jackmcbarn: Description: {{de|Nationalitätenkarte der östlichen Provinzen des Deutschen Reiches nach dem Ergebnissen der amtlichen Volkszählung vom Jahre 1910 entworfen von Ing. Jakob Spett}} (I can't do more with that than a Google Translate myself; but ping User:MyMoloboaccount; he requested it so hopefully he will also translate the description). Author: Spett, Jakob; publisher Gotha Justus Perthes. Year: 1910. Category:Maps in German, Category:Ethnographic maps of Poland (I think). Thanks! --_{Piotr Konieczny aka Prokonsul Piotrus| reply here} 05:00, 28 October 2013 (UTC)

"National map of eastern provinces of German Reich based on official census of 1910" would be ok. As to licence I believe it would be PD, but I am no expert, as to reason, to demonstrate the national census of 1910 in German Empire and its demographics.--MyMoloboaccount (talk) 22:08, 29 October 2013 (UTC)

I need a PD reason to upload it, and I still need a filename. Jackmcbarn (talk) 03:43, 3 November 2013 (UTC)

Jakob Spett died in 1942 according to [5], p. 77. (He was born in 1863 or 1862 according to the same source, pp. 74-75: they state his wife was 93 years old in 1968, and her husband was 12 years older.) The map is a derivative work; it is based on "Vogels Karte des Deutschen Reiches und der Alpenländer", published also by Justus Perthes in Gotha in 1915. That earlier map was created 1891-93 by cartographer Carl Vogel (1828-1897) as "Karte des Deutschen Reiches". It was revised and extended to cover the "Alpenländer" (alpine countries) 1913-15 by cartographer Paul Langhans.[6] Langhans lived 1867-1952 and thus the map is not yet in the public domain in Germany, and neither is Spett's derivative. Therefore not suited for uploading to the Commons. If you want to upload it locally at the English Wikipedia, that would work; it's {{PD-1923-abroad}}. (Spett's map was used in the negotiations at Versailles after WWI, so it must have been published by then.) Lupo 13:04, 3 November 2013 (UTC)

File:National map of eastern provinces of German Reich based on official census of 1910.jpg If you see any improvements that can be made to its description page, please make them. Jackmcbarn (talk) 15:39, 3 November 2013 (UTC)

New file tag

Per this discussion, I've created {{esoteric file}}. Realizing full well that we have {{do not move to Commons}}, I feel that one doesn't generally get used for this purpose, and is more for situations where some technical reason exists.

Commons is already so cluttered with completely useless crap that it's difficult to find quality files there. I think we should start making it general practice to use something like this, rather than copy every single copyright-free image to Commons regardless of their conceivable value for the general public. FYI, This refers mostly to our endless library of screesnshots of Wikipedia behavior for use in bug and proposal discussions (or in my case, script documentation), but I worded the template more generally, just in case others are applicable. equazcion → 15:14, 28 Oct 2013 (UTC)

I've made a couple of tweaks. You may want to elaborate a little more in the documentation the purpose. Obviously there is more than 1 Wikipedia. Images used in Wikipedia articles should be moved to Commons if possible even if the article doesn't yet exist in any other languages. Mr.Z-man 15:25, 28 October 2013 (UTC)

Thanks for pointing that out. I expanded it a bit. Let me know if anything is still missing. equazcion → 15:34, 28 Oct 2013 (UTC)

• Thanks for this. I hope it stems the tide somewhat. — Scott • talk 14:45, 30 October 2013 (UTC)

Making links with quotes work

I've frequently run into the problem of wanting to cite a search (usually on talk pages; I realize of course it isn't typically a valid citation for articles) and failing:

"photo+credit%3A+Wikipedia"+-site%3Awikipedia.org search

The reason for this turns out to be that the standard resolution of a link doesn't handle the quotation mark. Whereas %22 works just fine:

search

One option is to put in a bug report over it - I imagine someone has - but I don't have the patience to chase that squirrel. So I'm thinking to seize the old Template:Link, which currently redirects to Template:Ill. This would involve changing a hundred or so redirects to direct links from the articles that still use it, then repurposing it into a small, simple template with the intended usages:

{{link|1=https://www.google.com/search?q="photo+credit%3A+Wikipedia"+-site%3Awikipedia.org|search}} and/or <nowiki>[{{link|1=https://www.google.com/search?q="photo+credit%3A+Wikipedia"+-site%3Awikipedia.org}} search]

i.e. so that you can use it to produce the plain text to link if a second parameter is omitted, or to produce a functioning link if it is given.

Does this seem like a good idea to you? Wnt (talk) 22:15, 28 October 2013 (UTC)

It does seem like a good idea, but can you make it so that the wikilinking is handled automatically and the text to be displayed is an optional second parameter:

{{link|http://www.google.com/search?q="foobar"}} could return [https://www.google.com/search?q=%22foobar%22]

and

{{link|http://www.google.com/search?q="foobar"|foosearch}} could return [https://www.google.com/search?q=%22foobar%22 foosearch]

I think that would be easier to use... Livit⇑^Eh?/_What? 20:12, 29 October 2013 (UTC)

I think you're saying what I meant in the first case I listed above: {{link|1=https://www.google.com/search?q="photo+credit%3A+Wikipedia"+-site%3Awikipedia.org|search}} ought to return [https://www.google.com/search?q=%22photo+credit%3A+Wikipedia%22+-site%3Awikipedia.org search] Wnt (talk) 21:28, 29 October 2013 (UTC)

You are starting with a partially encoded string, why not use a fully encoded string ? It's better to use a fully unencoded string, and feed that to the encoder OR use a fully url encoded query string to begin with. {{urlencode:"photo credit: Wikipedia" -site:wikipedia.org|QUERY}} https://www.google.com/search?q=%22photo+credit%3A+Wikipedia%22+-site%3Awikipedia.org ... —TheDJ (talk • contribs) 10:09, 30 October 2013 (UTC)

You're right - better not to reinvent the wheel (I had some notion somehow you could improve readability in some strange situation by not uuencoding everything, but this is almost certainly asking for trouble!). However, it is still nice to have one template with two parameters rather than the more unusual magic word format inside a bracketed section, and "link" seems like a really desirable synonym for "urlencode:" even when only one parameter is used. Wnt (talk) 15:00, 30 October 2013 (UTC)

Well, this wasn't much of a mandate, but I've gone ahead with Phase I and altered all of the incoming links to use Template:ill directly. The only "what links here" entries for Template:Link are user .js modules, and by doing sequential deletions I've determined these are set off by a comment line, as shown at User:Wnt/ArticleTranslator.js. While I don't know for sure that the javascript doesn't rely on the template (I don't know if the what links here is meaningful at all) I don't think {{Template|Link}} ({{Link}}) is meant literally in that usage, and if it is, I'd expect it to simply display a link to the template rather than relying on its results.

I'll hold off a bit before continuing though, to see if there are people still out there actively adding {{Link}} in the course of their editing. Wnt (talk) 00:36, 3 November 2013 (UTC)

Pool queue?

Just got an interesting error while searching from a redlinked link (Pisa): "An error has occurred while searching: Pool queue is full". I've never seen that one before... - The Bushranger _{One ping only} 18:22, 29 October 2013 (UTC)

Hehe. Web developers are known to insert jokes into error messages. A famous POP3 error apologized to users saying "Sorry it didn't work out". Mozilla had one on an XUL error page that read "There is no data, there is only XUL!" (Ghostbusters reference, as "XUL" is pronounced "zool"). I'm pretty sure we've seen this particular MediaWiki one before though. equazcion → 18:33, 29 Oct 2013 (UTC)

So the pool queue exists so the Apache servers don't overload backend services like page parsing or searches. When you see this error (which you shouldn't very often), it means the pool counter is working as designed (we could be a bit nicer on the error, though). No need to worry, you can generally just try whatever you're doing again and it should go through. ^demon^{[omg plz]} 20:06, 29 October 2013 (UTC)

At least it didn't say Pool's closed... Wnt (talk) 21:34, 29 October 2013 (UTC)

Yes, we've seen this before. See Wikipedia:Village pump (technical)/Archive 113#Pool queue is full (from June this year). – PartTimeGnome (talk | contribs) 22:25, 29 October 2013 (UTC)

And also at the Help Desk not long ago. When I saw that request for help, it was the first time I could remember encountering this problem. Nyttend (talk) 02:11, 1 November 2013 (UTC)

Maybe we should link the error to Pool (computer science) to avoid confusion (or just avoid jargon altogether, but the link lets users learn a new word ;). Bawolff (talk) 22:45, 29 October 2013 (UTC)

Technical query

from RF's talk page Script: [7]

is giving timeouts - any idea why? The script url is enough to give me timeouts and wonder if something is fundamentally wrong here. Rich Farmbrough, 21:42, 17 August 2013 (UTC).

The script is 626 kBytes. For a site where 64 kByte of content generates a "long page" warning this seems an unacceptable overhead. Can someone please copy this to VP:T. Rich Farmbrough, 09:42, 30 October 2013 (UTC).

It's a combination of all your gadgets that you have enabled and the Mediawiki core scripts. You can cut down to at least 2/3 the size by disabling all the gadgets that you use. Second, we no longer have a longpage warning, I suspect you are using User:Dr_pda/prosesize. As detailed on Wikipedia:Article_size there is a technical limit on articles of 2048 KB and there are limits on the complexity of pages as detailed here. So Long, and Thanks for All the Fish. As to why you get timeout, I cannot find an explanation, it works just fine for me. —TheDJ (talk • contribs) 10:32, 30 October 2013 (UTC)

Thank button on page histories

Tracked in Phabricator
Task T57648

Is there anyway of getting rid of or hiding this waste of space? Thanks. Lugnuts ^{Dick Laurent is dead} 14:20, 30 October 2013 (UTC)

Only via personal CSS or JS. See Wikipedia talk:Notifications/Thanks#Hide thank from History? and the immediately preceding thread above it about preferences. Steven Walling (WMF) • talk 23:09, 30 October 2013 (UTC)

I made a script for that you might like: User:Equazcion/DynaThank equazcion → 02:54, 31 Oct 2013 (UTC)

Current speed skating event .svg file

Hello.

It would be great if it was possible for someone to create a "Speed skating current event" file similar to the ones for other sports? They other ones can be found in {{Current sport}}.

Thanks in advance.

HandsomeFella (talk) 16:25, 30 October 2013 (UTC)

I suspect that what is needed is to take a pictogram like File:Speed skating pictogram.svg, superimpose File:Current event template.svg to be like the others in commons:Category:Sport current event, save it as File:Speed skating current event.svg and add a test within the {{#switch:}} of {{current sport}}. --Redrose64 (talk) 19:34, 30 October 2013 (UTC)

I've created the image now - what do you think? --Redrose64 (talk) 22:28, 30 October 2013 (UTC)

Great. Thank you very much! Is it ready to use in Template:Current sport? HandsomeFella (talk) 08:29, 31 October 2013 (UTC)

It's now ready for use. --Redrose64 (talk) 10:25, 31 October 2013 (UTC)

Brilliant! Many thanks. HandsomeFella (talk) 12:24, 31 October 2013 (UTC)

I am not signed in

But I am signed in. I never signed out, and I recently had to sign back in, and I assume I checked the box saying to keep me signed in for 30 days. I went to the login screen and was told I was already signed in as vchimpanzee, though I could sign in with a different name. I signed in again. After that, I clicked on "online" at the top of the screen and was told I was not signed in.— Vchimpanzee · talk · contributions · 19:23, 30 October 2013 (UTC)

I used the page history and undo to say I was online, and that worked fine.— Vchimpanzee · talk · contributions · 19:57, 30 October 2013 (UTC)

There are quite a lot of related threads in the archives, mostly from Wikipedia:Village pump (technical)/Archive 114 onwards. If the circumstances suggest that Wikipedia doesn't think that I'm logged in, I use Ctrl+F5 and that normally works for me. --Redrose64 (talk) 20:01, 30 October 2013 (UTC)

It didn't work. And I'm wondering if Wikipedia will think I'm not logged in next time. I did check the box. I'm about to leave anyway and change to offline. I may edit tomorrow form a library, but of course I won't already be signed in there.— Vchimpanzee · talk · contributions · 21:28, 30 October 2013 (UTC)

It happened for clicking on "Offline" and CTRL-F5 didn't help.— Vchimpanzee · talk · contributions · 21:38, 30 October 2013 (UTC)

If you do have Javascript enabled and are still seeing no notice of the login, please check for Javascript errors in the browser's error console. Please also try to update your browser cache by https://en.wikipedia.org/wiki/Wikipedia:BYPASS. --AKlapper (WMF) (talk) 13:25, 31 October 2013 (UTC)

I'm not at home but I'll try these steps when I get there. I got a message to update Java, which I am aware is not Javascript, but I'm not sure whether I actually did. I also got a message telling me to update my virus protection but it didn't say how, and I spent time in a live chat with someone connected with the virus protection company. This is when the problems started happening.— Vchimpanzee · talk · contributions · 14:13, 31 October 2013 (UTC)

Whoops. Are you sure that the "virus protection company" was legit? I get phone calls 2-3 times a week from somebody claiming to be from Windows support and asking me to visit their website for a free online check. I never do it. --Redrose64 (talk) 17:52, 31 October 2013 (UTC)

Does this only happen when clicking the online/offline links? I think there might be a bug in the online/offline script you are using. The code contains hard-coded HTTP URLs, but Wikipedia now uses HTTPS by default for logged-in users. For the links to work on HTTP and HTTPS alike, the URLs should be protocol-relative (i.e. "//en.wikipedia.org/w/...") or server-relative (i.e. "/w/index.php?..."). I suggest contacting the script's author. I've already fixed the same issue in the template associated with the script. – PartTimeGnome (talk | contribs) 22:16, 31 October 2013 (UTC)

Good catch, I'v made the links protocol relative. They probably should be server relative as well, but I leave that up to the author. —TheDJ (talk • contribs) 09:52, 1 November 2013 (UTC)

Thanks guys. Didn't realise Wikipedia had changed to a secure transmission protocol. Cheers anyway. CJ ^{Drop me a line! • Contribs} 15:25, 1 November 2013 (UTC)

To PartTime Gnome: I had the problem only when I first went to the site each day, and when clicking on the online/offline links. I go to the site using HTTP links. I saved them in an email to myself which I update as I work my way through the Help Desk archives.

To Redrose: I clicked on the symbol for the virus protection company (McAfee, actually) that appears on my screen automatically. From there I can go to support. I hardly think a scam artist could reproduce the entire web site or redirect me to him from the legitimate site.

I forgot to click "keep me signed in" so I signed out and signed in again. I also deleted history. I don't have the problem now. Thanks.— Vchimpanzee · talk · contributions · 19:13, 1 November 2013 (UTC)

Didn't work. I clicked on "back" as many times as necessary to get back to where I started, and the first page still looked like I wasn't signed in. I even changed one number in the URL and got taken to another page that looked like it. I signed in, forgot to say I wanted to stay signed in, signed out, and signed in again. Now all is well. I hope.— Vchimpanzee · talk · contributions · 19:34, 1 November 2013 (UTC)

After clicking "back" sufficient times, you may still need to force your browser to reload the page (in Firefox, F5 may be sufficient, otherwise try Ctrl+F5) - otherwise it may display a cached copy, one which still suggests that you're not signed in. --Redrose64 (talk) 22:10, 1 November 2013 (UTC)

Interlanguage link

Please take a look at Koji Uehara. The link to ja/日本語 brings you to this template. The correct link is ja:上原浩治. And from the ja page, the link to en brings you to Template:2013 Boston Red Sox. But clicking "Edit links", it looks OK. [8] I just don't know how to correct the links. It would be grateful if someone correct the links. Thank you. Oda Mari (talk) 17:02, 31 October 2013 (UTC)

• The templates contain interwiki links linking the templates to each other, but a </noinclude> was misplaced, so the interwiki links in the templates overruled Wikidata content for the articles. Fixed on both projects. --Stefan2 (talk) 18:52, 31 October 2013 (UTC)
  • Good catch! Jason Quinn (talk) 19:00, 31 October 2013 (UTC)

Edits from 127.0.0.1

There appears to be something odd going on -- there are a series of edits dated today from Special:Contributions/127.0.0.1, which I would have said was impossible. Anyone have an idea what might be causing it? There's been at least one less-than-constructive edit from this IP, so I'm concerned as to what to do if vandalism edits come through this way. —Darkwind (talk) 23:04, 31 October 2013 (UTC)

I wouldn't say it's impossible that it's someone editing from the en.wikipedia server computer. I also don't see any problems with blocking it. Just don't hardblock it (xff problems). Ginsuloft (talk) 23:20, 31 October 2013 (UTC)

There's apparently multiple issues going on right now. I've placed a short block on it since it seems to be inconveniencing a couple of users. Elockid^(Boo!) 23:34, 31 October 2013 (UTC)

Oh, so it is a bug after all. Apologies to the Wikimedia sysadmins for assuming bad faith. Did you do a CU on it? Ginsuloft (talk) 23:39, 31 October 2013 (UTC)

No, I monitor the Requests for unblock and there are multiple users who have reported being autoblocked, both logged in and logged out. Elockid^(Boo!) 00:36, 1 November 2013 (UTC)

On a completely unrelated note, can someone trout Elockid for having an overly large signature? KonveyorBelt 18:06, 1 November 2013 (UTC)

Roan is looking into this now. --MZMcBride (talk) 00:58, 1 November 2013 (UTC)

tl;dr version: this should be fixed now, although it might take a while for things to settle down completely.

Today (Oct 31) at 14:52 UTC, we started sending page view requests from Oceania to ulsfo, our new caching center in San Francisco. We were already sending other kinds of traffic there (like uploaded images and static assets), but this was the first time the main traffic stream was going through ulsfo. However, there's an issue with the way our servers are set up in ulsfo which meant that if an anonymous edit came in via HTTPS or IPv6, it would be reported by MediaWiki as coming from 127.0.0.1. At 15:34 UTC, two of these edits came in, but other than that there weren't any, presumably because it was the middle of the night in that part of the world at the time. Starting at 19:37 UTC there was a slow trickle of 127.0.0.1 edits, gradually increasing in frequency as the morning started in Oceania, and around 23:00 UTC it started really taking off, going from a couple of edits per hour to dozens of edits per hour. At 00:21 UTC (on Nov 1), User:Master of Puppets reported the issue on IRC, and User:Reedy and I started investigating. After spending some time chasing a red herring, we figured out that these edits were all HTTPS or IPv6 requests from ulsfo with 127.0.0.1 appearing in the XFF header. At 01:25 UTC, User:LeslieCarr (WMF) undid the 14:52 UTC change, sending page view traffic from Oceania back to eqiad (our primary data center in Virginia). As of the time of writing, there have been no 127.0.0.1 edits since 01:28 UTC, but since these traffic switches aren't instant so a few more might get through for a little while.

My apologies for this glitch, and thanks for reporting this and bearing with us while we figured it out. There might be some residual autoblock damage as described above, but other than that everything should be fine now. I put in logging so we can track down issues like these more easily in the future, and we'll be sure to watch out for 127.0.0.1 edits next time we move traffic over to ulsfo. --Catrope (talk) 01:59, 1 November 2013 (UTC)

Update: It turns out I misdiagnosed the issue. The bug wasn't caused by HTTPS/IPv6 termination, but by strange behavior when the backend cache a request was mapped to happened to be on the same box as the frontend cache that received the request. Because there are 6 Varnish boxes in ulsfo, each with one frontend and one backend, this would have happened for about 1/6th of all anonymous edits going through ulsfo, independent of whether HTTPS or IPv6 was used. I was led to believe that HTTPS/IPv6 termination was to blame because almost all edits in the slice of the log I was looking at were HTTPS and the one HTTP edit in that slice was IPv6, but that was mostly a coincidence. Either way, the underlying issue is believed to be fixed now, and our networking people are planning to switch Oceania page view traffic back to ulsfo soon. When they do that, they'll be on the lookout for 127.0.0.1 issues. --Catrope (talk) 17:54, 1 November 2013 (UTC)

Thanks; I've added some text referencing this message at the IP's user page and its talk page. Graham87 05:53, 1 November 2013 (UTC)

Fix needed for Template:Alphanumeric TOC

Hi, when using Template:Alphanumeric TOC, it is adding an unwanted section of ==Contents== to each article wherever it is transcluded. See: 2000s in film and note in the TOC that there are 2 sections labelled "Contents". The usual Template:Compact ToC does not do this. Would someone be able to fix the template for us? Thanks much Funandtrvl (talk) 00:05, 1 November 2013 (UTC)

I think the problem is in Template:TOC top, which inserts the offending H2: <h2>{{{title|{{MediaWiki:Toc}} }}}</h2>. I'd prefer if someone more familiar with TOC templates could look at this though. The templates involved are also all still full-protected. equazcion → 00:25, 1 Nov 2013 (UTC)

(edit conflict) It's caused by the use of <h2>...</h2> in [9] by User:TheDJ. h2 was removed earlier after discussion at Template talk:TOC top#Edit request but it has gone in and out. I think it should be out. PrimeHunter (talk) 00:27, 1 November 2013 (UTC)

I doubt this will be all that controversial. Just get rid of it. equazcion → 00:30, 1 Nov 2013 (UTC)

In reading the Talk page referenced above, it looks like the "Content" section link/listing in the TOC of an article is usually undesirable. I vote for taking it out, since when the alphanumeric toc is used multiple times in an article, like 2000s in film, it looks very weird to have multiple Content sections in the TOC. The template is protected, and I don't know enough code to be able to fix it. Do I need to make an edit request on the template's talk page? Funandtrvl (talk) 01:14, 1 November 2013 (UTC)

It seems like User:TheDJ inadvertently re-inserted the H2 tags recently in an unrelated edit. The consensus is to leave them out. If an admin downgrades the protection level to template-protect I can probably do this but I would really rather User:PrimeHunter or User:TheDJ, or someone else who knows TOC templates, to be the one. equazcion → 01:38, 1 Nov 2013 (UTC)

You could make an edit request at that talk page, but if you don't have a specific code replacement to provide I don't think it would happen. equazcion → 01:44, 1 Nov 2013 (UTC)

There was little inadvertent about it. It was quite on purpose, to make it mirror the ACTUAL Table of Contents structure. As the Template:TOC top documentation mentions, use "primary=false" when a TOC is used in addition to a proper TOC. The reason to use h2 is for accessibility reasons. And remember that Template:Alphanumeric TOC is not the only user of Template:TOC top. —TheDJ (talk • contribs) 09:32, 1 November 2013 (UTC)

The actual TOC does use <h2>...</h2> but it doesn't produce a TOC entry saying "Contents". PrimeHunter (talk) 11:12, 1 November 2013 (UTC)

The primary TOC mentioning Contents is a side effect of editors using more than one TOC on a page. Most uses of {{TOC top}} REPLACE a proper TOC and then there is no problem whatsoever in using H2, and you will have consistent and expected behavior for pages. It's when you add a secondary TOC, when you start to create trouble. So make that distinction by using the 'primary=false' option and you won't have a problem. —TheDJ (talk • contribs) 11:59, 1 November 2013 (UTC)

I think this edit to 2000s in film] illustrates what DJ means. Without the option, the templates cause an extra heading to appear in the template. In my experience it is more common for the template to be use as replacement for the TOC. older ≠ wiser 12:12, 1 November 2013 (UTC)

• The protection of Template:Alphanumeric TOC has been lowered to TE if there is a consensus for a change, which there seems to be. Happy editing. Technical 13 (talk) 11:42, 1 November 2013 (UTC)

Template:TOC top is still full-protected, and that's where the change would need to be made. User:TheDJ, you just told us a bunch of things that were already mentioned -- but didn't mention why you felt your change back to H2 was a benefit. Is there any reason it wouldn't be better to make multiple TOCs not produce this unintended behavior by default, rather than requiring a parameter to fix it? Edit: I see you mentioned accessibility reasons, but I'm not sure what those could be -- and most other TOC templates don't seem to produce an H2, at least not by default. Any reason it particularly belongs in templates that use Template:TOC top? equazcion → 13:48, 1 Nov 2013 (UTC)

Now dropped to TE protection.--User:Salix alba (talk): 14:07, 1 November 2013 (UTC)

Thanks Salix :) PS. For the time being I added mention of this "primary" parameter to the documentation at Template:Alphanumeric TOC. I doubt people using this template would've reasonably known to go looking at TOC top to find that information. equazcion → 14:18, 1 Nov 2013 (UTC)

1. Regarding accessibility reasons, if you don't use h2, a screenreader will not identify the header of the Article ToC in it's own Document ToC (which includes much more than just the article headers) as it does for other pages. The official TOC and most fake ToCs do actually use h2 (I did a rewrite of almost all of them not so long ago).
2. "multiple TOCs not produce this unintended behavior". That would indeed be nice, but that is unfortunately not possible since we are faking our own ToC here. There is no 'knowledge' that we can use to determine if a table is an additional fake next to a proper TOC, or if it supposed to be replacing the proper TOC.
3. Again the problem here is not the h2 itself, it's that you use are using two elements that are designed to present a ToC for an article (combined with the fact that the core software only knows about the it's own primary ToC). Another example that shows that this is a somewhat faulty usage for 2000s_in_film: if I put a heading named G in between Events and Highest-grossing films, and then press the G of the Template:Alphanumeric TOC in that article, it will take me to this newly added G instead of the G in the list. It has been visually added to the section of the page, but that doesn't magically make it specific to the section. —TheDJ (talk • contribs) 14:26, 1 November 2013 (UTC)

These are all valid things to consider. I nevertheless have to say it seems haphazard to change the default behavior and add a parameter required to get it back to the currently expected behavior, when that parameter would of course not be in any invocation yet, no one would know about it, the talk page consensus was against it, and without making any documentation changes to the affected parent templates (which are the only places any users would realistically be looking). And yes, regardless of your opinion on whether multiple TOCs should actually be used on a page. I've now changed the documentation for the particular template that brought the issue to VPT, but there are still another 30 or so to go, and adding this parameter to the affected uses is probably a bigger job. equazcion → 14:51, 1 Nov 2013 (UTC)

It's true, there is a lot of cleanup. And it is true that when I made the original changes, I did not think about the fact that about 2-5% of the uses would be 'incorrect' to begin with. But this one problem is not more important than the problems solved. In general we should move forward and become more correct, not keep crap in simply because we have crap that depends on crap.

If you ask me, we should replace every list or section section usage with a "Alphanumeric section TOC" that has primary set to false, and preferably uses prefixes for the header IDs. —TheDJ (talk • contribs) 15:14, 1 November 2013 (UTC)

Thanks for adding the primary option to {{Alphanumeric TOC}} in [10]. But many editors will fail to set primary=false when it's required. Is there no way to make it accessible in screenreaders without causing the "Contents" entry in cases where the page also has a normal TOC? PrimeHunter (talk) 15:29, 1 November 2013 (UTC)

Stop forcing https!

The netbooks at my school are extremely dodgy, and https pages either do not work on come in plain view on them. How can I stop it forcing https? Changing the bar from https to http doesn't work,not even forcing port 80 does. -- ^t numbermaniac ^c 02:02, 1 November 2013 (UTC)

Go to Preferences and uncheck "Always use a secure connection when logged in". Yes, it'd be nice if this was default unchecked... - The Bushranger _{One ping only} 02:39, 1 November 2013 (UTC)

Okay then. Thanks! :) -- ^t numbermaniac ^c 05:35, 1 November 2013 (UTC)

Heading and image overlap

A reader reported, via OTRS, that in Hinduism in Tamil Nadu, the heading "Bhairavar" overlaps the image. I see the problem, in Mozilla Firefox, but not in Chrome. Can anyone see why?--S Philbrick(Talk) 13:45, 1 November 2013 (UTC)

Depends on screen resolution and default fonts and font sizes on the operating system that you use. More information welcome (I cannot reproduce with 1440px screenwidth on Firefox 24 on a Fedora 19 system). --AKlapper (WMF) (talk) 15:26, 1 November 2013 (UTC)

Help regarding titles of the articles

Hi, Surya again. Tamil wiki admin. I want to know how the titles of the articles are italicized? Is there any template used for it? I did browse something. But, couldn't get anything. Please tell me how to italicize the title of an article? (ex: The Mary Tyler Moore Show) Thanks -- SuryaPrakash  ^Talk... 17:03, 1 November 2013 (UTC)

It's usually done with {{Italic title}}, often called from an infobox. The manual method not relying on templates or modules is for example {{DISPLAYTITLE:''The Mary Tyler Moore Show''}}. PrimeHunter (talk) 17:11, 1 November 2013 (UTC)

Users not listed as confirmed on Commons

Why is commons:User:Bùi Thụy Đào Nguyên not listed as confirmed? According to commons:Commons:Autoconfirmed users, this user should be automatically confirmed after editing for 4 days. Magog the Ogre (t • c) 18:26, 1 November 2013 (UTC)

Autoconfirmed status is checked automatically and not shown in logs, but the users themselves can see it at Special:Preferences. Manually confirmed users are shown in logs, for example Special:ListUsers/confirmed. PrimeHunter (talk) 18:37, 1 November 2013 (UTC)

Contains external links

Tracked in Phabricator
Task T58485

When I make edits like [11], it asks me to do a CAPTCHA, with the explanation you can read at MediaWiki:Fancycaptcha-addurl. The explanation "Your edit contains new external links" is false in this case (the links are internal - unless the external link is buried in the template, which they shouldn't expect us to know). Others such as this example have made the same complaint, but no one has fixed or explained it. Shouldn't we at least change the false explanation? 24.22.173.74 (talk) 18:31, 1 November 2013 (UTC)

"protection log" made by the template is technically an external link. It's even more confusing when the external link is caused by a change to a template used elsewhere on the page and not a template added by the user. Maybe that happened in your other example [12] but it's hard to determine now. Our software cannot say why the saved page contains an external link that wasn't there before, so the best we could do is modify MediaWiki:Fancycaptcha-addurl with a general message explaining the external link is sometimes caused indirectly by the user's edit but they have to enter a CAPTCHA anyway. Such a message would be unnecessary and potentially confusing in cases where the user actually did enter an external link. This is probably the large majority of cases. PrimeHunter (talk) 18:55, 1 November 2013 (UTC)

Actually, that protection log link is whitelisted. The problem in that edit was a hidden link to "urn:x-wp-editsemiprotected:Jennifer_Lopez" which is generated in {{edit semi-protected}} so various tools don't have to parse template invocations to find out what pages have edits requested for them.

Normally you could whitelist problematic links at MediaWiki:Captcha-addurl-whitelist, but that only works for http and https protocol links. I've filed Template:Bug to track the problem. BJorsch (WMF) (talk) 19:30, 1 November 2013 (UTC)

Interlanguage links in different font

Tracked in Phabricator
Task T58346

The interlanguage links have changed their font-family. They used to be in the same font-family as the rest of the page (Arial), but now they are something called "Autonym" which is apparently pulled from

@font-face {
  font-family: "Autonym";
  font-style: normal;
  src: local("Autonym"), url("//bits.wikimedia.org/static-1.23wmf1/extensions/UniversalLanguageSelector/lib/jquery.uls/css/font/Autonym.woff?2013-10-24T17:33:20Z") format("woff"), url("font/Autonym.ttf") format("truetype");
}

It appears to be a result of <div id="p-lang" class="portlet" role="navigation">...</div>. The trouble with that font is that it's indistinct (Windows XP, Firefox 24, MonoBook) - it's blurred, with blue and red fringing, particularly on tall thin letters like lowercase i and l - see screenshot at right. Arial, by contrast, is sharp-edged without fringing. This isn't just English Wikipedia; it's others too e.g. French. --Redrose64 (talk) 21:13, 1 November 2013 (UTC)

More specifically:

#p-lang ul {
    font-family: 'Autonym',sans-serif;
}

So, I should be able to override it with

#p-lang ul {
    font-family: 'Arial',sans-serif;
}

in my Special:MyPage/skin.css (as indeed I can) - but when and why did the change come in, and where was it announced? --Redrose64 (talk) 21:28, 1 November 2013 (UTC)

Gag, this looks horrible, especially as letters have "gaps" in them. Hopefully this can be quickly fixed... - The Bushranger _{One ping only} 21:20, 1 November 2013 (UTC)

See https://blog.wikimedia.org/2013/10/28/the-autonym-font-for-language-names/. The aim is to have all language names in all scripts visible, even where a user doesn't have the relevant fonts installed on their computer. It's a good idea, but I agree that the font itself could use some work. the wub "?!" 21:23, 1 November 2013 (UTC)

#p-lang ul { font-family: inherit; } should reset the font its original value. I agree the font does not render well at all on Windows. — Edokter (talk) — 22:48, 1 November 2013 (UTC)

...I find it hard to believe that there are very many computers out there that don't have Arial... thanks Redrose64 for the fix script (and btw, it works if you have it all on a single line in the skin.css as well). - The Bushranger _{One ping only} 23:56, 1 November 2013 (UTC)

It's not a script, it's a CSS rule. CSS is very tolerant of whitespace - generally speaking, if a newline is permitted, a space is permitted instead - and most spaces may be removed. In fact, in that particular CSS rule, only one of the spaces is mandatory - the one before ul - so you can put

#p-lang ul{font-family:'Arial',sans-serif;}

and it works equally well. --Redrose64 (talk) 13:03, 2 November 2013 (UTC)

"inherit" is better because you don't need to specify any font; it resolves to the parent elements font, which is whatever the rest of the page uses. — Edokter (talk) — 14:54, 2 November 2013 (UTC)

I know; but I was replying to The Bushranger, who was replying to me, so I repeated my example changing only the whitespace. I could have used

#p-lang ul{font-family:inherit;}

but this would not have illustrated the whitespace elimination quite so well, because there were other changes not related to whitespace. --Redrose64 (talk) 19:20, 2 November 2013 (UTC)

I was also replying to The Bushranger :) — Edokter (talk) — 19:26, 2 November 2013 (UTC)

I like and use User:Equazcion/SidebarTranslate which solves the original missing-fonts problem (and a few others) entirely.

I liked the idea of using the native-language name, but in practice it made things so much harder for me to find (as a monolingual reader who occasionally wants to check out various other language examples) - I'd often have to mouse-over each of the links, looking for the 2letter prefix that looked familiar or correct. –Quiddity (talk) 20:08, 2 November 2013 (UTC)

Uncommanded unchecking of preferences options...

I just discovered that somehow "Add pages I create and files I upload to my watchlist" had been unchecked in my preferences without my having done so. Whats up? — Preceding unsigned comment added by The Bushranger (talk • contribs) 23:53, 1 November 2013 (UTC)

I can confirm that. In fact, if I go to Preferences → Watchlist, enable "Add pages I create and files I upload to my watchlist" and save, it immediately disables itself again. --Redrose64 (talk) 13:08, 2 November 2013 (UTC)

I tested it here and it still works for me. Interestingly, although Add pages I create and files I upload to my watchlist is unchecked, this page I just created was still added to my watchlist. -- Toshio Yamaguchi 13:18, 2 November 2013 (UTC)

I have been able to check that option just now and it stayed checked. Matma Rex talk 13:28, 2 November 2013 (UTC)

Crappppp, this is probably my fault, I've suggested some cleanup-only changes in the config which I needed for something else (they basically make this option off by default), they were accepted, and we all missed the fact that this was set to true elsewhere… sorry, you'll have to live with this for a few hours/days until somebody with appropriate permissions is around to fix it D: (But you should be able to reenable this for yourself right now perfectly well. It's just the default that was accidentally changed.) Matma Rex talk 13:36, 2 November 2013 (UTC)

The preferences should be restored for everyone now. Matma Rex talk 16:41, 2 November 2013 (UTC)

Failed action=purge

I see a strange error message at [13]:

“ A database query error has occurred. This may indicate a bug in the software.

• Function: Title::invalidateCache
• Error: 1205 Lock wait timeout exceeded; try restarting transaction (10.64.32.26)

”

Any idea what's going on? --Stefan2 (talk) 00:47, 2 November 2013 (UTC)

I was getting a similar message while trying to perform the following actions:

• Deleting File:04 - Pero Me Acuerdo De Tí.jpg
• Uploading a new version of a file at File:44adelaidelogo.png
• Modifying file description of File:44adelaidelogo.png.

The problem occurred both in Chrome and Firefox and seems to be resolved now -- Diannaa (talk) 01:08, 2 November 2013 (UTC)

Weird. I got a watchlist notification that File:Agents of S.H.I.E.L.D intertitle.png had been deleted by User:Mark Arsten, but the file was still there, so I tried to purge the page, which failed. I see that you later deleted it. --Stefan2 (talk) 01:24, 2 November 2013 (UTC)

I just had a bunch of database query errors myself while trying to delete some images. Very frustrating. Mark Arsten (talk) 01:26, 2 November 2013 (UTC)

I deleted it shortly after this bug resolved itself. Things have been kinda sketchy on the deletions for the last day of two, where I get the "Wikimedia Error" page. Then I would go back and see if the file had actually gotten deleted, and sometimes it was and sometimes not :/ -- Diannaa (talk) 01:29, 2 November 2013 (UTC)

Right now it seems like I can't delete the last four files in Category:Orphaned non-free use Wikipedia files as of 25 October 2013. Any luck there? Mark Arsten (talk) 01:44, 2 November 2013 (UTC)

It took a lot longer than it should have, but File:Cfnews13.jpg deleted successfully on the first try. - The Bushranger _{One ping only} 09:36, 2 November 2013 (UTC)

crashing

I am using IE and wikipedia keeps crashing on me intermitently, it is the only site that is doing this. — Preceding unsigned comment added by 65.95.145.147 (talk) 01:20, 2 November 2013 (UTC)

Can you give more detailed info? Even a guess could help (e.g., "It seems to happen more often when I do XYZ"). What version of IE are you using? Does it provide any error messages? Does this happen on other Wikimedia wikis as well (e.g., try it on Wiktionary or French Wikipedia). πr² (t • c) 03:09, 2 November 2013 (UTC)

jQuery UI CSS may no longer load by default

See http://lists.wikimedia.org/pipermail/wikitech-l/2013-October/072779.html. As of next week, jQuery UI may not be loaded by default on some wikis. I'm not sure if this is true here - we might have something enabled for all visitors that includes jQuery UI - but if we do lose this CSS, templates like {{clickable button 2}} and {{help pages header}} will suffer. (What's worse, many logged-in users won't notice, since gadgets like Twinkle include jQuery UI.) — This, that and the other (talk) 01:59, 2 November 2013 (UTC)

Hmm... {{clickable button 2}} just makes a plain link for me (see my test). {{help pages header}} is just a horizontal list of links with bold formatting. Have the styles already been disabled? Does this depend on skin? (I'm using Monobook.) Does it require some gadget to be enabled? (I've disabled most gadgets, including many of the defaults.)

The e-mail you link to mentions editing common.js to re-enable the styling, not common.css as one would expect for CSS. Is this in fact a JavaScript effect and not CSS? (I have JS disabled.) – PartTimeGnome (talk | contribs) 16:11, 2 November 2013 (UTC)

Since jQuery UI is a javascript library, its accompanying CSS is indeed only loaded through javascript. Instead of relying on jQuery UI, we migh consider using MediaWiki.UI for pretty buttons instead, or have all buttons styled pretty by default. — Edokter (talk) — 16:27, 2 November 2013 (UTC)

Is 'mediawiki.ui' module loaded by default? Ruslik_Zero 19:19, 2 November 2013 (UTC)

No, but there have been suggestions on making it so (and a patch pending[14]). Hard to say if it will be, in the end, there was some opposition. You can always just make common.js load it here. Matma Rex talk 21:12, 2 November 2013 (UTC)

Might as well just load jquery UI from common.js here then, since we already have templates relying on that. Plus there's the standards thing. equazcion → 21:16, 2 Nov 2013 (UTC)

Search algorithm

A reader tried to search, using "cobalt II chlorate" in the search box.

The reader wondered why it didn't immediately find Cobalt(II) chlorate

I note that it does suggest this in a drop-down box, but it doesn't think it is a good enough match, so it doesn't bring you directly to the article.

I note that the search term differs from the title in three respects:

1. Case
2. The parentheses around II
3. The space after cobalt

The case is not an issue, as the search will find the right term even is the case doesn't match.

One question I am posing is whether we think that the lack of parentheses should be a problem. Is it reasonable to configure the search algorithm to create a match when the search terms and target differ by parentheses?

I'm sure it could be done. One question is whether it is supposed to work, and doesn't because of a bug, versus something that might be nice, but has never been requested. Assuming it isn't handled, the question is whether we think articles with parentheses are common enough to make a request for a change in the algorithm. I'm personally on the fence, and would be supportive if the developer say it was easy, and would not push for it if the developer says it is tricky.

The second is whether the existence of a space should prevent the search from getting a hit. I see that if I search for "hollydolly" it asks if I mean "holly dolly". If I search for "slip stream" I get to Slipstream, but only because someone created a redirect (I think).

I think it would be nice if searching could give you the right article even if the search ans target differed by a space—even if you knew to add the parentheses, I don't think it is obvious that a searcher should know to enter "cobalt(II) chlorate" rather than "cobalt (II) chlorate".--S Philbrick(Talk) 15:27, 2 November 2013 (UTC)

Cobalt(II) chlorate is the first hit for me on cobalt II chlorate. Do you mean when the search includes quotation marks as in "cobalt II chlorate"? That doesn't work for me but people should know to try without quotation marks when they are not sure of the exact title. The whole point of using quotation marks is to ask for exact hits only. I wouldn't expect it to match when there are differences in spacing. PrimeHunter (talk) 16:35, 2 November 2013 (UTC)

I think what Sphilbrick is saying is that typing "cobalt II chlorate" into the search box and hitting enter should immediately take you to Cobalt(II) chlorate, much like how typing "ApPLe" into the search box and hitting enter immediately takes you to Apple (since it's case insensitive, I presume). Theopolisme (talk) 21:35, 2 November 2013 (UTC)

It's not due to case (in)sensitivity, but punctuation and spacing. Cobalt(II) chlorate has a pair of parentheses and one space; "cobalt II chlorate" has no punctuation but two spaces. If I type that into the search box, it suggests "Cobalt(II) chlorate" in the list, but unless I pick that entry before hitting ↵ Enter, it will go to Special:Search. Now, if Cobalt II chlorate were created as a redirect (which is permissible because it's a plausible search term, and so could fall within either or both of "Alternative names" and "Alternative ... punctuation") you would simply need to type that in and ↵ Enter without needing to select from the list. --Redrose64 (talk) 22:55, 2 November 2013 (UTC)

I asked the question, and what Redrose64 said is exactly what I meant. If the user is searching for cobalt II chlorate, I think it's obvious the users means Cobalt(II) chlorate, and therefore it seems logical to me he/she should be redirected. Besides that, you don't often use parentheses when typing search-terms, so I assume there are more users than only me searching for cobalt II chlorate instead. Of course, if this 'change in the search algorithm' would be accepted, it would have to affect all chemical formula-searching. IsaiahvH (talk) 13:45, 3 November 2013 (UTC)

It's obvious to a chemist that "cobalt II chlorate" and Cobalt(II) chlorate are one and the same. But computers aren't so intelligent; they need to be given special provision about such matters as IUPAC nomenclature of inorganic chemistry. In this case, section IR-2.8.2 of Nomenclature of Inorganic Chemistry states "In names they [Roman numerals] indicate the formal oxidation state of an atom, and are enclosed in parentheses immediately following the name of the atom being qualified." At school (early 1980s), we were taught about the parentheses, but not that there should be no space before the opening parenthesis. --Redrose64 (talk) 13:59, 3 November 2013 (UTC)

Wikipedia talk:Persondata-o-matic editing rights

Hi there. I've got a problem using the above, it says that I've not got the right to edit the article in question every time I try to use it on an article and then click 'save'. I've browsed the docs for the gadget and messaged the creator with no response. Anyone know a fix or what I'm doing wrong? --S.G.^(GH) _ping! 20:14, 2 November 2013 (UTC)

Infobox criminal is turning programmers into murderers

Robert Tappan Morris created the first computer virus, though Template:Infobox criminal seems to be forcibly adding "Killings" to his page. There doesn't appear to be anything on his page's syntax that is causing this. Is this a mistake, or is the template revealing a darker truth? ▫ JohnnyMrNinja 23:46, 2 November 2013 (UTC)

In this edit today the template was merged by @Underlying lk: with Template:Infobox murderer.Blethering Scot 23:58, 2 November 2013 (UTC)

Fixed; n.b. I've not checked for other parameters that might trigger that header even though they are displayed higher up. --Redrose64 (talk) 00:02, 3 November 2013 (UTC)

Thanks for the quick fix! ▫ JohnnyMrNinja 03:57, 3 November 2013 (UTC)

Tylenol

Tylenol (edit | talk | history | protect | delete | links | watch | logs | views)

There's something up on the page. The "Advertisement" header shows up in the TOC, but it's blank in the text itself, and the link it broken. I've examined the Wikimarkup and can't figure out what's wrong. All the ref tags are closed, so what is causing this problem? hbdragon88 (talk) 07:52, 3 November 2013 (UTC)

Slight correction: the header was ==Advertisements== (plural). I've confirmed your observation. After I disabled AdBlock Plus for Wikipedia pages, however, the header, and the correct TOC link operation, reappeared. As a workaround, I've changed the header to ==Marketing==. --Lexein (talk) 08:09, 3 November 2013 (UTC)

Revert oversights edit?

There was an edit to Post that was reverted, and it seems that the edit was oversighted out of existance. Help:User contributions doesn't explain what a struck-out edit means, and Help:Reverting doesn't mention that this could happen. User:Eeekster is not a sysop or 'crat; I can't tell if they're an oversighter. If I remember correctly, oversighting actions are logged as such anyways - and there's nothing in the logs for the appropriate time. So what happened? Josh Parris 09:49, 3 November 2013 (UTC)

The suppression log is also private. --Rschen7754 09:55, 3 November 2013 (UTC)

So, suppression is undocumented (in the help pages) and privately logged. Orwellian. Josh Parris 10:14, 3 November 2013 (UTC)

Help:User contributions#Deletion says: "In some limited circumstances individual contributions (that is, specific edits) may be removed from public view by administrators using Revision Deletion; such edits remain visible to administrators. In even more limited circumstances edits may be oversighted, remaining visible only to the handful of users with the Oversight permission." The link on "Revision Deletion" shows how it looks. There is no secrecy about the existence of the feature. It's just a question of how much overlap to have between help pages. PrimeHunter (talk) 10:58, 3 November 2013 (UTC)

The only real fix to the problem is to have live updated mirrors that aren't under the same administration. I'd like to tell them not to do it, but you know full well that they may feel they have no choice under governments with a quite imperfect notion of what 'free expression' is about, and once they start doing it at all it is just so enticing to keep going further, gradually growing and expanding their godlike powers. Wnt (talk) 15:53, 3 November 2013 (UTC)

Superfluous line-break in template

Hello.

Could someone help identify where there's an unnecessary line-break in {{Year in Norway}}. The line-break is transcluded, which makes the articles that use the template look a bit awkward.

Thanks.

HandsomeFella (talk) 11:35, 3 November 2013 (UTC)

I removed some line breaks between categories that seemed extraneous, which fixes the issue in tests, but the whitespace is still showing up when this template is used just after {{Use dmy dates}} (as seems to be the case with most of its uses). equazcion → 11:59, 3 Nov 2013 (UTC)

This seems to be caused by the line break in the code before the wikitable begins. Since that wikitable code needs to be at the beginning of its own line, I can't find a way to get rid of the whitespace, aside from chopping off everything that comes before it (which does effectively remove the whitespace in these articles, FYI). This seems like it might be a common issue for templates so maybe someone knows of a solution I'm unaware of. equazcion → 12:07, 3 Nov 2013 (UTC)

I think there is at least some improvement. Thanks for trying. HandsomeFella (talk) 14:29, 3 November 2013 (UTC)

I've managed to fix it. That was a tricky one. A single line break before the table is fine, the problem was caused by having two consecutive line breaks before the table. The first line break is not in the template itself, but in the articles that use it (between the {{Use dmy dates}} and {{Year in Norway}} templates). The fix was to put something between those line breaks to stop them being consecutive, but that doesn't cause a visible gap itself. An empty <nowiki/> tag did the trick. – PartTimeGnome (talk | contribs) 16:33, 3 November 2013 (UTC)

importScript execution delayed

I created a Javascript function, addTopLink, to insert an arbitrary convenience link into the horizontal menu at the top right of the page. I had the function inside my common.js, followed by a call to insert one link, and it worked.

I thought I'd make the function generally available, so I moved it to User:Largoplazo/toplinks.js. In my common.js I replaced the function definition with importScript('User:Largoplazo/toplinks.js');, and refreshed the cache. It didn't work: my custom link stopped appearing in the top menu. So I did some troubleshooting, the state of which you can see in the two files. At this moment the bottom three lines of my common.js are commented out, but when they weren't, what happened each time a page loaded is that:

1. A dialog box reading "undefined" appeared, which I dismissed.
2. A dialog box reading "Just reached the end of toplinks.js" appeared, which I dismissed.

The first dialog should display the contents of window.teststring, which is set inside toplinks.js. Since it was appearing as "undefined", that means it hadn't been set yet. The second dialog comes from an alert call directly inside toplinks.js. So the calls were coming in the wrong order.

Finally, I opened Firebug and went to the console. I could see that window.teststring had its assigned value and the function addTopLink existed. So toplinks.js was being fully executed, just not in the expected sequence.

What am I missing? —Largo Plazo (talk) 13:07, 3 November 2013 (UTC)

Scripts loaded via importScript() are executed asynchronously – that is, the browser sees the importScript, starts loading that script in the background, and continues down to alert(window.teststring) and addTopLink, which don't work like you expect because the script might not be loaded yet.

You could use something like this, but I'm not sure if it's the best practice. It definitely works, though.

importScript('User:Largoplazo/toplinks.js').onload = function() {
  alert(window.teststring);
  addTopLink("pt-newpagesfeed", "NewPagesFeed", "/wiki/Special:NewPagesFeed");
}

Matma Rex talk 13:51, 3 November 2013 (UTC)

Oh, also – there's a default function called mw.util.addPortletLink which does pretty much the same thing as your addTopLink and some more: mw:ResourceLoader/Default_modules#addPortletLink. Matma Rex talk 13:56, 3 November 2013 (UTC)

Since WP pages are HTML 5, and I see the script element is defined to have an onload event in HTML 5, this is great! Though I'll check out addPortletLink as well. Thanks. —Largo Plazo (talk) 15:34, 3 November 2013 (UTC)