Wikipedia:Village pump (technical)/Archive 147

From Wikipedia, the free encyclopedia
Jump to navigation Jump to search


Rollback function has been changed

I've started experiencing a new behavior when using rollback (on one Win XP computer + newest Firefox, I'll check newer Windows platforms later). The most annoying one is that "blind" rollback (rollback without viewing the diff) now requires confirmation - an extra page pops up asking 'do you want to revert this edit' OK? Does anyone know where this implementation was discussed and how to bypass this feature? I've lost a quick rollback function for dealing with obvious vandals. Materialscientist (talk) 22:21, 26 May 2016 (UTC)

I'm on the other side of that fence. I found it annoying when a stray click rolled back stuff I didn't intend to roll back. I gave up the rollback right after about the fourth time that happened to me. It's simply poor design to do something like that without a confirmation click, which only takes another second or two. But I don't know whether it was discussed. ―Mandruss  22:29, 26 May 2016 (UTC)
I'm with Materialscientist, it's also screwed up Twinkle's automatic completion of the linked article field when warning users. I've just returned to anti-vandal work after a short break, and don't feel inclined to continue unless I can do so efficiently.  —SMALLJIM  22:44, 26 May 2016 (UTC)
See phab:T49782 about that. Helder 16:03, 27 May 2016 (UTC)
I just rolled back two edits on this page - neither required a confirmation. — xaosflux Talk 22:57, 26 May 2016 (UTC)
Also with my testing account using all defaults - I do get the little spinning circle and the upper right corner has a popout confirmation box, but I am not getting a confirmation prompt at all. Xaosflux ep (talk) 22:59, 26 May 2016 (UTC)
Can you verify you are actually using rollback? (Hover the link and ensure it has &action=rollback). What is the exact text of the message you are getting? — xaosflux Talk 23:01, 26 May 2016 (UTC)
I'm thinking the enhanced "rollback" with WP:Twinkle is what you are seeing broken - I'm having a problem with it as well. — xaosflux Talk 23:04, 26 May 2016 (UTC)
This was intentionally changed due to phab:T88044. —TheDJ (talkcontribs) 23:09, 26 May 2016 (UTC)
(edit conflict)Something has changed, but I did not get a confirmation dialog when I just now rolled back. There was a (JQuery?) box that briefly appeared in the top right corner indicating that that the roll back was successful.- MrX 23:11, 26 May 2016 (UTC)
Let me rephrase my original comment. If I preview the diff (of the edit I want to revert) then rollback doesn't ask to confirm it. If I try revert without seeing the diff then I'm asked to confirm it. I want to avoid this confirmation dialogue. I also want to see what I've rolled, particularly when rolling multiple edits (now I don't see it). I do not use Twinkle, just plain rollback. Materialscientist (talk) 23:16, 26 May 2016 (UTC)
I used the rollback link from history with no confirmation dialog appearing.- MrX 23:25, 26 May 2016 (UTC)
I used rollback from RecentChanges - no confirmation either - @Materialscientist: Can you copy the url you are using (copy before clicking) and paste here? — xaosflux Talk 23:29, 26 May 2016 (UTC)
  • OK, reproduced the problem: If using standard rollback, in a new tab/window it is no longer instantly working-ending in a confirmation. If you just click the links they process in-page now, instead of going to a new page. — xaosflux Talk 23:36, 26 May 2016 (UTC)
  • Should we be able to have our cake and eat it too? (Basically just remove the Revert edits to this page? [OK] prompt?) — xaosflux Talk 23:41, 26 May 2016 (UTC)
    @Krinkle: - any hints? — xaosflux Talk 23:44, 26 May 2016 (UTC)
    (ec) Xaosflux is absolutely correct in that the confirmation dialogue appears only when I open the rollback in a new tab, which I always do, as I want to keep the starting page (be it my pre-designed watchlist or recent changes log). It doesn't appear if I roll in the same tab, but then I (1) don't see what I've rolled, and (2) can't revert several edits from the same user/IP. Annoying. Materialscientist (talk) 23:47, 26 May 2016 (UTC)
  • WP:LAVT seems broken for me. I also don't like that I click rollback, and the page refreshes suggesting that I haven't rolled back when I indeed have. (it still says "Rollback 2 edits") — Andy W. (talk ·ctb) 23:53, 26 May 2016 (UTC)
    Admin rollback is definitely broken for Lupin's Anti-vandal tool. (Non-admin javascript is okay) The HTML that gets returned seems different, and the tool thinks that someone else has edited since the vandalism event. — Andy W. (talk ·ctb) 00:00, 27 May 2016 (UTC)
  • I've created phab:T136375 asking for a roll back - not sure if that will happen - please feel free to subscribe. — xaosflux Talk 00:12, 27 May 2016 (UTC)
  • WP:Stiki was broken with me too (since two hours ago). I've confirmed that my regular rollback issues are independent of the PC platform (WinXP, Win10) and Wikipedia skin (MonoBook, Vector). Materialscientist (talk) 00:16, 27 May 2016 (UTC)
  • Added watchlist notice. — xaosflux Talk 00:27, 27 May 2016 (UTC)
  • The problem with Twinkle hasn't happened to me, even though I use Twinkle all the time. One somewhat annoying Twinkle-related thing I have experienced is that if you revert an edit by clicking on "Rollback AGF", it opens the user whose edit you're reverting's talk page and the Twinkle menu's warn option. However, this really isn't an issue for me since typically it's in everyone's best interest not to WP:BITE good-faith newcomers who make honest mistakes, and instead it is better to explain what they did wrong. However, I am unclear on whether this thread is for discussing a Twinkle error or, as Materialscientist's comment above would suggest, a WP:ROLLBACK error. Everymorning (talk) 00:38, 27 May 2016 (UTC)
    We think the unexpected Twinkle behavior is tied to this change. — xaosflux Talk 00:41, 27 May 2016 (UTC)
  • Please see my comments posted at WP:ANI at Wikipedia:Administrators' noticeboard/Incidents#Malfunction_with_automated_rollback. This recent change has effectively broken rollback while using Lupin's Anti-vandal tool and is producing inaccurate results when clicking (rollback) directly from the page history as well. Oftentimes it will incorrectly state that "rollback failed" when in fact it succeeded, and in other cases it fails to rollback entirely, on what appears to be a completely inconsistent basis. Is there an option in preferences which would allow us to revert back to the original rollback features before this change was implemented? Regards, Yamaguchi先生 (talk) 00:42, 27 May 2016 (UTC)
  • It's only saying that rollback failed if you're clicking the link multiple times before it has the time to process it (at least, that's what I've found from some limited testing). Same thing if you open the rollback link in a new tab 50 times; it will say that it has failed for the 49 times after the initial processing of the action. Obviously this change should have been better communicated to the maintainers of the anti-vandalism scripts, so they could update them before the change. There is no way of changing back on a per-person basis. Ajraddatz (talk) 00:51, 27 May 2016 (UTC)
    No, we are not sharing the same experience in that case. I am seeing failures based upon a single click on an intermittent basis. A window box temporarily appears in the upper right hand corner of my screen notifying that rollback has failed, while a check of the page history reveals that the rollback actually succeeded. Regards, Yamaguchi先生 (talk) 01:04, 27 May 2016 (UTC)
Oh, this sounds familiar — it wasn't with rollback, but I noticed one of those "blah blah has failed blah blah" popups when doing something or other, even though it was still working on whatever it was (the popup went away once the action finished)... Idk if that's related though. Goldenshimmer (talk) 01:45, 27 May 2016 (UTC)
  • Any Idea if this is effecting Huggle too? --Cameron11598 (Talk) 00:53, 27 May 2016 (UTC)
    I just fired up Huggle and made a revert OK. — xaosflux Talk 01:01, 27 May 2016 (UTC)
  • Additional error As a test, I used rollback on the latest Sandbox edit, opening it in a new tab; it asked if I wanted to use rollback, and when I clicked yes, it went through properly. I then clicked rollback (no new screen) on my own edit, and again it went through properly. I then clicked rollback on my second edit (the result should have been the removal of the text in question), and I get a failure message:

    Rollback failed [line break] Cannot roll back edit to Wikipedia:Sandbox by Nyttend (talk · block · contribs) because someone else has edited the page. [line break] The last revision was by Nyttend (talk · contribs).

    When I open it in a new tab, it produces the confirmation page properly, but okaying the action produces the same error message. As of 01:00, 27 May 2016 (UTC), nobody has edited the page since my latest edit; it's not like an intervening edit got in the way. This is distinctly not good: the software's thinking that I've tried to roll back my first edit to the page, not my latest. Rollback should always attempt to revert the latest edit, doing any consecutive previous ones along the way, but here it's trying to do something different. Nyttend (talk) 01:00, 27 May 2016 (UTC)
    PS, the sandbox-maintenance bot just removed stuff, so I rolled it back, rolled back my rollback, and attempted to rollback my rollback rollbacking, but once again, the third edit failed. I understand that self-reverting twice is an exceptionally rare situation, but that doesn't mean that we should ignore the tool's misanalysis of what link was clicked. Nyttend (talk) 01:02, 27 May 2016 (UTC)
    @Nyttend: It's behaving as I would expect.
    • Your first rollback reverts all consecutive edits made by the most recent editor (Epizode), producing revision 722273317 of 00:54, 27 May 2016, identical to revision 722270423 of 00:31, 27 May 2016 by Everymorning;
    • your second rollback reverts all consecutive edits made by the most recent editor (Nyttend), producing revision 722273325 of 00:55, 27 May 2016, identical to revision 722271034 of 00:36, 27 May 2016 by Epizode;
    • your third attempt to rollback reverts all consecutive edits made by the most recent editor (Nyttend). This would produce a version identical to revision 722271034 of 00:36, 27 May 2016 by Epizode but as noted, that is identical to revision 722273325 of 00:55, 27 May 2016 by Nyttend, which is the current version; so since it's a no-change edit, you get the "Rollback failed" error message.
    It has worked like that since I was given WP:ROLLBACK in June 2010. Rollback doesn't toggle between versions, it reverts to the most recent version that was made by any person who is not the most recent editor. --Redrose64 (talk) 07:10, 27 May 2016 (UTC)
Ok, this is kind of off topic, but I'm wondering, what exactly is the big deal over having rollback? Can't you do the same thing with a couple extra clicks by going to the page revision right before the offending editor, clicking "edit", and clicking "save"? This has puzzled me for a while. Goldenshimmer (talk) 01:43, 27 May 2016 (UTC)
Replied on your talk. — xaosflux Talk 01:47, 27 May 2016 (UTC)
  • Not sure if I really like this change. At the very least, there should be an option in Special:Preferences to revert it. Now I have to go through additional clicks in order to get to the user's talk page and be able to warn them with Twinkle with the "linked page" field automatcially filled in. I can at least confirm the option to leave a custom edit summary with a rollback as described at Wikipedia:Rollback#Additional_tools still works. —k6ka 🍁 (Talk · Contributions) 02:35, 27 May 2016 (UTC)

There's an incongruency with this change and the user's settings. At Preferences -> Browsing, I have "After rolling back a user's edit, automatically open their contributions page" enabled. Enabling this gives a rollback failed message which may or may not be correct (it seems non-deterministic).

After I disable "After rolling back a user's edit, automatically open their contributions page", rollback indeed asks for a confirmation. — Andy W. (talk ·ctb) 03:33, 27 May 2016 (UTC)

That would be mw:MediaWiki:Gadget-modrollback.js. Helder 16:57, 27 May 2016 (UTC)
  • This change has absolutely neutered massRollback, which now requires me to go through each rollback page that pops up and individually confirm that I mean to roll it back. This makes it much more time-consuming to keep track of and revert the few prolific sockpuppeteers that I would consider myself an expert on identifying. What problem was trying to be fixed when this was implemented? This seems to have created more issues than it solved. ~ RobTalk 04:31, 27 May 2016 (UTC)
Apparently the problem was the use of GET where the software should use POST. Helder 16:57, 27 May 2016 (UTC)

We've elevated this to a talk page notice? At minimum, we should have a temporary roll back the change so developers like myself (STiki) have time to accommodate the change. Why are we breaking legacy functionality? "Because it would be nice..." for HTTP POST/GET semantics to be normalized? This is much more understandable for security issues, but script/tool editors should have gotten some lead time on this one. What trouble does a temporary withdrawal cause? West.andrew.g (talk) 04:35, 27 May 2016 (UTC)

  • I can see no benefit in the new design. It now requires a lot of extra work when mass-reverting changes made by a single user. . . Mean as custard (talk) 08:30, 27 May 2016 (UTC)
  • It's as if Preferences → Appearance → Don't show diff after performing a rollback is permanently enabled. You can still get to the diff showing the rollback that you made, it's in the little box that appears at upper right, but it's an extra click following a (potentially large) mouse move to that box, which you might not complete before it fades. --Redrose64 (talk) 08:36, 27 May 2016 (UTC)
  • Please could someone explain to me, preferably in words of one syllable: (a) when and where script developers were notified that this change was happening; and (b) how it affects scripts that do rollback using AJAX, by querying the rollback token and then POSTing it. Thanks. Philip Trueman (talk) 11:08, 27 May 2016 (UTC)
    (a) This was added to the Tech News. Helder 16:57, 27 May 2016 (UTC)
    • Granted. But this says nothing about scripts that do rollback via AJAX. No window is opened. Philip Trueman (talk) 18:37, 27 May 2016 (UTC)
  • WTF is this? Change rollback back. Was there even any community discussion or consultation before this change was made? Absolutely textbook case of a solution looking for a problem, and not only are anti-vandal tools like STiki now broken, the "warn" function of Twinkle is dead as well since the "linked article" field no longer autocompletes as rollback doesn't open in a new page anymore. This is some VisualEditor level incompetence and whoever instigated this deserves a good WP:TROUTing (and then some). Again, change this shit back. Satellizer el Bridget (Talk) 11:46, 27 May 2016 (UTC)
  • @West.andrew.g: Not sure what you mean by "talk page notice"? I added this to the watchlist notice because I thought it would reach the right audience without being too intrusive to people that don't use rollback. — xaosflux Talk 13:00, 27 May 2016 (UTC)
  • This is really annoying. The whole point of admin rollback is to be able to do a quick rollback. WTF? --jpgordon𝄢𝄆 𝄐𝄇 13:27, 27 May 2016 (UTC)
  • Comment: I was wondering why Twinkle suddenly stopped working. Is this going to be fixed anytime soon? Why would a change like this take place without discussion and testing? GenQuest "Talk to Me" 13:42, 27 May 2016 (UTC)
    The place to complain about "why was this done" or "roll this back" is phab:T136375 - if it gets enough noise perhaps the developers will revert. — xaosflux Talk 13:51, 27 May 2016 (UTC)
With respect, Xaosflux, no it isn't. The developers should be monitoring this page on a regular basis especially after any major change is implemented, because it's the first place that reports about such changes are likely to be raised. A pretty obvious way of helping to keep their customers happy, I'd have thought.  —SMALLJIM  14:21, 27 May 2016 (UTC)
@Smalljim: I agree they should be, and I did link this discussion to the phab ticket that I opened asking for a revert. From what I can tell this is change was wide in scope, so reaches far beyond the English Wikipedia. I suggested the phab page so it could get more "heat". — xaosflux Talk 14:50, 27 May 2016 (UTC)
As a tool developer I am monitoring this page, and its the first place I come when something breaks. If a mistake is made and something is broken, developers fix it. This is something else entirely. This is a change with seemingly inconsequential benefit that breaks a whole host of tools (CBNG, Lupin, Twinkle, and STiki). I can't imagine how many man-hours are going down the drain, and how much more vandalism is surviving longer than it should. It takes me time to code up a fix, test it, and push a new version. Responsible development finds the tools effected by a change and gives them lead time to make a change (and ideally, doesn't break legacy functionality without a darn good reason). Look to the recent "authentication manager" changes for a good example on how this is done. I haven't seen any mention made of harm incurred for a short term withdrawal of the change. West.andrew.g (talk) 15:01, 27 May 2016 (UTC)
Yes, thanks for your efforts over this, Xaosflux. The apparent lack of any awareness (here and at phab) by the devs of the mess that this has caused is disappointing. I should think it's something their managers could usefully look into correcting - these are paid staff, aren't they?  —SMALLJIM  15:10, 27 May 2016 (UTC)

At the risk of looking like I'm just piling on here, I'd like to add my voice to those saying that these recent changes are entirely detrimental to the use of rollback (at least the way I use it). Reverting vandalism and other unconstructive edits is simply more difficult and time consuming now. Some of the problems I now encounter include: 1) Repeatedly being asked to confirm each rollback when rolling back multiple edits by one user, 2) after using rollback Twinkle no longer opens a user talk page with the "linked page" field automatically filled in, and 3) inaccurate "rollback failed" messages. Deli nk (talk) 14:00, 27 May 2016 (UTC)

And to add to the cheer, Cluebot NG seems to have been AWOL since yesterday evening. Now, I wonder how that could be? Philip Trueman (talk) 14:02, 27 May 2016 (UTC)

I'll add an issue I've observed: when rolling back with the option that allows one to enter an edit summary, which I almost always use, my summary text is not saved. DonIago (talk) 14:03, 27 May 2016 (UTC)

  • As Materialscientist said, normal rollback displays a loading gif, opening in a new tab needs confirmation. What pisses me off is that after the rollback is complete, it does not reload the page (previously it led to a new page with affirmation). Yet to see the behaviour with multiple edits. Nice to know the devs give zero fucks about the tons of scripts and users who were about to be affected by this. --QEDK (T C) 15:08, 27 May 2016 (UTC)
    • There is a difference between not giving a fuck and misjudging the impact of a change. —TheDJ (talkcontribs) 15:13, 27 May 2016 (UTC)
      • Whatever you say. Just rubs me a bit on the wrong side when all we try to is help the project and then we take a step back. Not a big deal, guess I should be more calm. --QEDK (T C) 15:21, 27 May 2016 (UTC)
        • Speed is probably the WORST metric there is when it comes to development. 17:53, 27 May 2016 (UTC) — Preceding unsigned comment added by TheDJ (talkcontribs)
          • Back when I was a programmer in IBM, it was impressed upon us that "timeliness is part of quality". More seriously, I agree with you on the general point, but here we are discussing the particular point of the speed with which the developers admit their mistakes. By my standards, they are not doing well. Philip Trueman (talk) 18:29, 27 May 2016 (UTC)
      • I can't defend QEDK's language, but we won't be able to tell which of the two it is until we see how quickly the developers fix things. Philip Trueman (talk) 15:27, 27 May 2016 (UTC)
Argh, same for Huggle actually. I can't revert on Huggle - at first I thought it might be down to my name change and HG not recognising my new name but nope, just saw this. I am therefore outlining that this has happened on Huggle too. Regards --PatientZero talk 15:13, 27 May 2016 (UTC)
@Patient Zero: Huggle is working just fine for me. —MRD2014 (formerly Qpalzmmzlapq) T C 17:10, 27 May 2016 (UTC)
MRD2014 - I'll try again now. You changed your name too, didn't you? I was Chesnaught555. --PatientZero talk 17:21, 27 May 2016 (UTC)

As noticed above, a developer is on this, and nothing should be broken anymore very soon. Thanks for your patience. --Elitre (WMF) (talk) 15:51, 27 May 2016 (UTC)

Thank you Elitre (WMF). --PatientZero talk 15:54, 27 May 2016 (UTC)
Why are WMF ruining everything on this site ? ... As if VE, Flow and the green-background update thing isn't bad enough They now mess around with (and ruin) Rollback ...., WOuld be nice if they actually left things alone ......., 'Don't fix what isn't broken springs to mind. –Davey2010Talk 16:56, 27 May 2016 (UTC)
I agree Davey2010. Rollback was working fine until all this happened. --PatientZero talk 17:22, 27 May 2016 (UTC)
Good to hear that everyone knows best about things they don't know jack about. —TheDJ (talkcontribs) 17:53, 27 May 2016 (UTC)
I've never said I know best and I'll admit I don't know best inregards to rollback however when you fuck around with a tool that's used by a lot of people on a daily basis and don't bother saying anything then you should know shit's going to hit the fan - As I said above if you don't fix what isn't broken then there's not going to be a problem, Everything gets updated and improved that's life but this wasn't an improvement ... it was an inconvenience as demonstrated in this tread alone. –Davey2010Talk 13:56, 28 May 2016 (UTC)
Huggle now works for me, by the way. That's that sorted. --PatientZero talk 17:24, 27 May 2016 (UTC)
  • Rollback works for me, but I don't like the new way it works. Before it would rollback the edit(s) and show me a new page with the edits removed, similar to an undo. Now, I get one of those little bubbles in the upper right-hand corner of the screen saying the rollback was successful, but it stays on the page, making me feel like it didn't work. I don't mind the bubble (they're now all over the place, e.g., "edit saved"), but I'd like to see the "proof" without going to history.--Bbb23 (talk) 18:35, 27 May 2016 (UTC)
  • Somehow I don't think it has affected me. Whether I click on the rollback button on the same tab or opened a new tab with rollback, I get the same result (action complete with diff). Is this only a problem for Twinkle / Huggle / other similar scripts that piggyback on rollback for editors that have the rollback button? (But if you have Twinkle why do you need rollback so desperately? Just handle it at software level!) Deryck C. 22:18, 27 May 2016 (UTC)
@Deryck Chan: The changes were reversed (albeit temporarily) when you tested it out. —k6ka 🍁 (Talk · Contributions) 22:30, 27 May 2016 (UTC)

Reversion is queued

A reversion of the breaking portion of the change is currently queued for updates and should be deployed in a few hours. The new change is not a 100% reversion - so keep an eye out for anything unexpected. — xaosflux Talk 19:15, 27 May 2016 (UTC)
Rollback is back to normal here now – straight to a page showing the diff of my edit so I can confirm I reverted the vandalism that I intended to. Just as it should be. And then on moving to the vandal's user talk page, Twinkle picks up the article title so I can warn without copy-pasting the title. Just as it should be :)  —SMALLJIM  20:15, 27 May 2016 (UTC)
So that works, but ClueBot NG is still down. —MRD2014 (formerly Qpalzmmzlapq) T C 21:12, 27 May 2016 (UTC)
(edit conflict)Rollback looks good. I confirmed that the generic WP:LAVT seems good. User-derived independent versions might still have issues (out of scope). Any confirmation about STiki or ClueBot NG? (I'm a bit worried about the bot) — Andy W. (talk ·ctb) 21:15, 27 May 2016 (UTC)
Message left for operator User_talk:ClueBot_Commons#Your_bot_may_need_to_be_restarted - this is something the operator will need to address. — xaosflux Talk 21:24, 27 May 2016 (UTC)
ClueBot NG has been given a restart and should hopefully be back reverting - RichT|C|E-Mail 22:07, 27 May 2016 (UTC)
  • I'm seeing a random "1" at the bottom-left of the diff page after a rollback. Noticed it on the Simple English Wikipedia too. —k6ka 🍁 (Talk · Contributions) 22:20, 27 May 2016 (UTC)
    @K6ka: Thanks for the note - looks like that is a bug that got rolled in to the release - I've notified the developer. — xaosflux Talk 22:44, 27 May 2016 (UTC)
  • @K6ka: - the spurious "1" problem has been resolved. — xaosflux Talk 23:18, 27 May 2016 (UTC)

I think it's still broken

I use PILT - my own modified version of AVT. One big difference is that PILT does MediaWiki rollback using AJAX rather than by opening a new window. That has stopped working and is still broken. Philip Trueman (talk) 09:50, 28 May 2016 (UTC)

A bit more thorough description than: "broken" is usually helpful. Unfortunately i don't have the rollback permission, so I can't really test this. —TheDJ (talkcontribs) 19:54, 28 May 2016 (UTC)
"Doesn't work"? More seriously, I think there is an onus on the developers to tell the users and script writers precisely what they have changed. Something has changed, and not been reverted, in the way Wikipedia responds to a script that queries the rollback token for a diff and then issues an AJAX POST with an option of action=rollback, passing the token. Maybe there's an additional required option now - I don't know. Maybe that action is no longer accepted, in which case they should justify the reduction in functionality. If there's a way for a script to query why such a request was not actioned I should very much like to learn it, but I'm not aware of one. Philip Trueman (talk) 03:25, 29 May 2016 (UTC)
An update: The relevant Phabricator request seems to have been removed by Aklapper. Frankly, I'm not impressed. Please can it be said boldly: This project is suffering from a total disconnect between the developers of the MediaWiki software and the users thereof. I want, very much, for that disconnect to be healed. If it is not, the project is heading for the garbage can. It is disappointing that the developers simply do not seem to care. Philip Trueman (talk) 20:05, 29 May 2016 (UTC)
Developer console allows you to use the debugger to step through any code and to inspect the requests and responses of any XHR (or normal for that matter) request. This might sound a bit snappy, but things change. If you can't debug this code now, then that means that you can't maintain it, which means that it by definition is going to break. If you write code you are a developer, not a user who needs to be told which 'importScript' to change. There is onus on MediaWiki developers to tell fellow developers about deprecations and removals yes, but there is no onus on them to figure out all problems. If you can't inspect a request, when other tools seem to have recovered just fine, then I feel very demotivated to start reading every character that was changed in the php code to figure out what is the reason for the 'breakage'. —TheDJ (talkcontribs) 20:29, 29 May 2016 (UTC)
You make a fair point but I do not completely agree with you. I want to start by saying that in real life I am a professional software developer. I am used to writing code to documented APIs. Often, those APIs are not documented as clearly or completely as I as a developer might wish - DWARF has been a particular challenge, to take a recent example. The plain fact is, the MediaWiki API has changed recently, and I am not aware of any relevant change to the documentation thereof. I have seen indications that the developers are changing things in the area of rollback and AJAX - beyond that, I am ignorant. I very much want to know exactly what it is they have changed (that is still changed), and what they have changed it to. My code uses the API in a way different from that used by Twinkle, AVT, Cluebot NG and the like. So? I don't see that that makes me a second-class user - I want to know what the relevant change is. My code had not changed. I regard the failure of the developers to be more forthcoming as arrogant, high-handed, and worthy of condemnation. I will, as you suggest, apply what tools I can to determine where things have changed, but some constructive input from the developers would be welcome. Philip Trueman (talk) 21:03, 29 May 2016 (UTC)
@Philip Trueman: Hi, could you explain which specific Phabricator request I "removed"? Phabricator tasks cannot be removed, however the associated projects of a task can be changed (if you referred to that). If you have specific questions why I did something, please elaborate and I'm happy to explain. Thanks, --AKlapper (WMF) (talk) 08:40, 30 May 2016 (UTC)
I apologise - I should have said 'closed'. As I understand it, you closed a request where a user (me) had made it clear that the fix seemed not to be complete, yes? Philip Trueman (talk) 08:43, 31 May 2016 (UTC)
@Philip Trueman: Thanks for clarifying. Hmm, the only activity of your Phabricator account seems to be in phab:T136375 which was closed by a developer in phab:T136375#2335144 (though more followup fixes took place after that action). My first activity in that task was removing unrelated projects from that task. Seems like some misunderstanding here. :) --AKlapper (WMF) (talk) 09:56, 31 May 2016 (UTC)

@Philip Trueman: I notice PILT uses the (since 1.24 deprecated) rvtoken parameter of the revisions module. Reading the diffs of the code, I suspect that this deprecated accessor was not converted along with the other tokens and might have broken in this latest changes. I'm verifying with other developers. —TheDJ (talkcontribs) 14:48, 30 May 2016 (UTC)

It appears this is also be reverted, still under (closed task) phab:T136375. — xaosflux Talk 16:57, 30 May 2016 (UTC)
Yes. PILT is working again. Thanks to those who listened. Philip Trueman (talk) 11:09, 1 June 2016 (UTC)
TheDJ: Thanks for this. I need to find the time to fix this, and any other deprecated usages, but it may take a few days. Philip Trueman (talk) 08:43, 31 May 2016 (UTC)

Rollback past SineBot?

One minor annoyance I've encountered when using rollback (both Twinkle and the tool) is that when the editor being rolled back does not sign their initial post, SineBot will usually sign it for them. This cases a break in the string of edits and rollback only goes back to SineBot's edit. Is there a way we could have rollbacks ignore SineBot and include them in the rollback? A similar issue occurs when references are rescued by AnomieBOT. See the edit history on Talk:Racism from May 28 for an example of what I'm referring to. EvergreenFir (talk) Please {{re}} 18:34, 28 May 2016 (UTC)

Twinkle has a button to restore a particular revision. Σσς(Sigma) 19:01, 28 May 2016 (UTC)
@Σ: I do use that when I realize there's an intervening edit. Problem is when I don't. Takes more time too (on my slow ass laptop with 40+ tabs open and whatnot). EvergreenFir (talk) Please {{re}} 04:41, 29 May 2016 (UTC)
True WP:ROLLBACK (as opposed to Twinkle's rollback) is a MediaWiki feature, which "rollbacks the last edits made by the last editor of the page; in other words, it restores the last version of the article whose editor is not the last one". It has no way of knowing that SineBot has a special task to perform: as far as the MediaWiki software is concerned, SineBot is just another user, which may well be the editor that is not the last editor. It could find out that SineBot is a bot, but would have no way of distinguishing it from e.g. Cyberbot II (talk · contribs) which also edits Talk: pages.
To get the behaviour of true rollback changed would need a phab: request. --Redrose64 (talk) 15:58, 1 June 2016 (UTC)

Tony Dean (racing driver)

Hello, I wonder if anyone can shed any light on why the above article I moved into mainspace on 30 April does not appear in my list of articles created? Two others I moved at the same time did so immediately. It's only a small thing, I know, but it was my 50th article and these tiny milestones are the only rewards we can hope for. :) Thanks. Eagleash (talk) 15:46, 1 June 2016 (UTC)

I see it in these search results. Where do you not see it? Fred Gandt · talk · contribs 16:00, 1 June 2016 (UTC)
It does not appear when clicking on 'contributions' (top of any page) and then 'articles created' at the bottom of the page. I.e. here and then here. Thanks. Eagleash (talk) 16:11, 1 June 2016 (UTC)
@Cyberpower678 and MusikAnimal: because I think you're likely to be most useful here (xtools). Fred Gandt · talk · contribs 16:56, 1 June 2016 (UTC)

Edit notice not highlighted

I added a 1RR edit notice to Flag of Northern Ireland but it isn't highlighted and I doubt it will be noticed by many editors. Can this be fixed? Thanks. Doug Weller talk 16:07, 1 June 2016 (UTC)

@Doug Weller: It shows for me. What are you expecting to see, but are not? --Redrose64 (talk) 17:24, 1 June 2016 (UTC)
On Template:Editnotices/Page/Flag of Northern Ireland you could add some styles or colors, see Template:Editnotice/doc for some examples. — xaosflux Talk 17:37, 1 June 2016 (UTC)
Thanks. For some reason I expected to see some color. I've fixed it now so it's more obvious. Doug Weller talk 18:33, 1 June 2016 (UTC)

Blocked for 1 century and 18 hours

I just blocked ThisIsaTest for 100 years, but the block log says that it will be effective for 1 century and 18 hours. How does 100 years end up with eighteen extra hours? I remember raising this kind of question before (but the other way: blocking for a year's worth of seconds ended up being several minutes short), but I couldn't find it with a search of the WP:VP archives. Nyttend (talk) 01:21, 2 June 2016 (UTC)

@Nyttend: see Wikipedia:Village_pump_(technical)/Archive_137#Weird_blocking_time. Basically it is a rounding error. — xaosflux Talk 01:44, 2 June 2016 (UTC)
Note the magic 43 minutes and 12 seconds times too - 99.9% of 30 days. — xaosflux Talk 01:47, 2 June 2016 (UTC)

Visual editor problems

Extended content

Expected vs observed behavior: "<small>" should be either in the "Style text" menu or should function as expected when typed (like "{{" or "[["); ideally both. If not possible/practical, it should trigger the "Wikitext markup detected" message. Also, the text of that bubble refers to a nonexistent user interface feature (a "Switch to source editing" menu option in the dropdown to the left of the "Save page" button); rather, that capability is located in the "Switch to source editing" button (icon looks like "[[ ]]"). The current reading is:

You are using the visual editor - wikitext does not work here. To switch to source editing at any time without losing your changes, open the dropdown menu next to "Save page" and select "Switch to source editing?".

I propose that it be changed to read:

You are using the visual editor — wikitext does not work here. To switch to source editing at any time without losing your changes, tap the "Switch to source editing" button next to the "Save page" button.


You are using the visual editor — wikitext does not work here. To switch to source editing at any time without losing your changes, tap the "Switch to source editing" button (reads "[[ ]]") next to the "Save page" button.

or something to that effect.

Problem encountered at: (see the bottom of the diff) (next parenthetical is tangential) (the <nowiki>''</nowiki> in the diff is user error attributable to trying to get the "Wiki markup detected" box to show up again so I could read the instructions for using the "Switch to source editing" feature without realizing that it disappears before the offending markup is deleted (I mistakenly assumed that its disappearing indicated that the markup was gone because I had happened to delete my added markup manually at pretty much the same time as the notification disappeared the first time I saw it); I wasn't being particularly attentive so I neglected to notice the continued presence of the '') (if anyone actually understands this tangential, "<small>"-ified wall of text, they probably deserve a fuckin award for language comprehension skills; I bet I'd have gotten lost and/or glazed over way before getting to the end of it).

Browser: FirefoxDeveloperEdition

Browser version: 48.0a2 (2016-05-31) Goldenshimmer (talk) 04:11, 1 June 2016 (UTC)

Thanks for the feedback! mw:VisualEditor/Feedback is the best place to make sure it is seen. :) --AKlapper (WMF) (talk) 09:47, 1 June 2016 (UTC)
Thanks, I've posted it there: mw:Topic:T539q7on6cc8l30b! :) —{{u|Goldenshimmer}}|✝️|ze/zer|😹|T/C|☮️|John15:12|🍂 02:16, 2 June 2016 (UTC)

Find sources


Could someone change Template:Notability so that the "find sources" external link creates a search query that excludes parenthetical disambiguators.--Prisencolin (talk) 03:53, 1 June 2016 (UTC)

Discussion at Template talk:Notability#Find sources. PrimeHunter (talk) 10:42, 1 June 2016 (UTC)
This is now implemented. — Mr. Stradivarius ♪ talk ♪ 13:38, 2 June 2016 (UTC)

Finding contributions of a editor who changed names

An editor was told at The Teahouse she was required to change her name. Sinebot signed her name with a link to contributions. Fortunately she replied with her new name and I was able to find her contributions with that name. But shouldn't a link to contributions with the old name somehow give us a clue as to how to find contributions with the new name?

She was asking for more help but in the articles she linked to neither name had contributed. And the other article she needed help with had been deleted.— Vchimpanzee • talk • contributions • 19:40, 1 June 2016 (UTC)

The talk pages redirect because the renaming occurred so I don't know what more do you need. The Recordings username has no contributions which I think is because they were moved. The new on has all the edits. It's the same as for Wikipedia:Courtesy vanishing. -- Ricky81682 (talk) 20:53, 1 June 2016 (UTC)
You can also see the rename in the log. — xaosflux Talk 01:17, 2 June 2016 (UTC)
When I go back to the contributions under the old name and click on "Log", I get "No matching items".— Vchimpanzee • talk • contributions • 18:23, 2 June 2016 (UTC)

Trying again on tech question

I asked about this on another page (got some helpful info but not quite the clarity I was after)..perhaps there are some very technical types here...FOR EXAMPLE: this very page/function/space I'm on right now must not have existed at some point in time...someone had the idea to create it and it was then was it created? Did, perhaps, editors suggest it be created and then somebody at the Wikipedia Foundation had to code it/create it? Or are tech-savvy editors with certain permissions able to create it on their own?? Nobody seemed to exactly know when I posted this question at "idea labs" Thanks.. (talk) 14:17, 31 May 2016 (UTC)

Hi IP, perhaps you were looking for the help desk? Either way, any editor can edit and create* articles on Wikipedia - if you check out this guide you can find out how (* unregistered editors (like yourself) cannot create new articles outside of this process)-- samtar talk or stalk 14:25, 31 May 2016 (UTC)
I'm not sure you're understanding my question..I'm talking about an entirely new function/space not an article... (talk) 14:31, 31 May 2016 (UTC)
Ah! You mean (for example) Wikipedia:Village pump (carrots)? I believe any editor could propose such a creation and, if the idea gained consensus, create it. -- samtar talk or stalk 14:48, 31 May 2016 (UTC)
Anyone can create it without consensus, it would take consensus for it to stay though. -- GB fan 15:00, 31 May 2016 (UTC)
Any logged in editor can create a page in any space including a page similar to this one. It does not take any special permissions other than having a named account. Whether that page will remain or not may require a consensus discussion to take place. If someone believes that the new page is not helpful/does not comply with policy then they can take it to a deletion process. It is usually easier if the discussion takes place before hand rather than building the page and hoping it survives. If you are thinking about a bot to do the task you are talking about elsewhere, that will require approval before it can operate, see Wikipedia:Bot policy. -- GB fan 14:38, 31 May 2016 (UTC)
so this entire village pump space/function can be created by just normal editors and doesn't require anything on the end of the Foundation? what about the log of all blocks, for example..that space/function was and can be created by just anyone and placed into Wikipedia? and how sure of all this are you? if others could confirm/add anything... (talk) 14:46, 31 May 2016 (UTC)
See my response above, but in relation to the logs - these are programmatically generated by the Mediawiki software -- samtar talk or stalk 14:49, 31 May 2016 (UTC)
The entire Village pump are just pages in Wikipedia and like any page in Wikipedia they can be created by any logged in user. Logs are a different matter and that takes coding in the back end and would require the developers to take action. I am very sure of my answer. -- GB fan 14:58, 31 May 2016 (UTC)
and if a new/different log was desired what then would be the process?? you're saying normal editors couldn't do it, right? (talk) 14:54, 31 May 2016 (UTC)
Correct - It would be a 'feature request' made on our phabricator. This guide (shameless plug) helps explain the process -- samtar talk or stalk 14:57, 31 May 2016 (UTC)
Normal editors can not create logs that would show up at Special:Logs. Now if you are looking at a log like is at WP:SPI anyone can manually create it but if they want a bot to create it, then there is an approval process. -- GB fan 14:58, 31 May 2016 (UTC)
well it would be similar to that SPI and maintained automatically by a bot I wouldn't have to go through phabricator?? but where would I request the page/bot be created? (talk) 15:04, 31 May 2016 (UTC)
You can start at Wikipedia:Bot requests. -- GB fan 15:09, 31 May 2016 (UTC)

so if I wanted a new log related to unblock requests the only process would be to go through this phabricator process, is that correct? (talk) 15:01, 31 May 2016 (UTC)

It depends on what you want. If you want it to show up as a simple list in Special:Logs then yes. If you want it on a page with additional information like the log at WP:SPI then no it would not go through phabricator. If this is related to the discussion about the unblock transparency, you probably don't need to go through phabricator. -- GB fan 15:04, 31 May 2016 (UTC)
Thought I recognised this train of thought - IP, it's likely what you're after could be solved with a bot as GB fan suggests -- samtar talk or stalk 15:07, 31 May 2016 (UTC)
okay, so it wouldn't require a petition to phabricator, right? the page (log) could be created and then maintained by a bot (and bot only) where would I a. petition someone who knows how to do this...and then b. it would have to be approved..before or after I request it be made?? (talk) 15:11, 31 May 2016 (UTC)
The bot would have to be approved after it was made, before it started making any edits. If you didn't see it above you can start the bot process at Wikipedia:Bot requests. -- GB fan 15:14, 31 May 2016 (UTC)

Okay, I guess that clarifies, thanks. (talk) 15:16, 31 May 2016 (UTC)

To clarify, the reason that a bot would be used here rather than a feature in Mediawiki (i.e. a phabricator request) is because unblock request (AFAIK) are not a feature of Mediawiki: they are handled basically just by putting the user talk page into Category:Requests for unblock. Things in Special:logs tend to be special actions, rather than normal edits. —  crh 23  (Talk) 15:48, 31 May 2016 (UTC)
okay, thanks. (talk) 15:52, 31 May 2016 (UTC)

Wikidata alert is sticky

I'm getting an alert that I "edited" Wikidata by moving a page on English Wikipedia. How do I make the alert go away. "Mark as read" is checked, and clicking on it doesn't appear to do anything. I don't really care to receive such alerts that I "edited" Wikidata as a side effect of my English Wikipedia actvities; can I turn those off? wbm1058 (talk) 11:53, 2 June 2016 (UTC)

I see, clicking the "X" removed it from the drop-down; that's not standard, and expected, behavior. Now the notification is gone completely; seems odd that's the only way to "mark it as read". wbm1058 (talk) 11:56, 2 June 2016 (UTC)
I agree that the "X" is not obvious. We knew about this problem, so we've been looking at alternative designs. We now have a new design that we like, which we'll implement in the next few weeks. --Roan Kattouw (WMF) (talk) 19:43, 2 June 2016 (UTC)
@Wbm1058: In Special:Preferences#mw-prefsection-echo You can turn off "show notifications from other wikis". — xaosflux Talk 17:43, 2 June 2016 (UTC)
Thanks. Now, what if I do want to see all notifications from other wikis except Wikidata? wbm1058 (talk) 17:48, 2 June 2016 (UTC)
That is not currently available. See mw:Help:Notifications/Cross-wiki. The programming teams are looking for feedback and you can make suggestions here: mw:Help talk:Notifications. — xaosflux Talk 17:56, 2 June 2016 (UTC)
Thanks, maybe this open bug: Cross-wiki mark as read doesn't work is the problem? wbm1058 (talk) 18:04, 2 June 2016 (UTC)
@Wbm1058: Do you use a browser extension such as PrivacyBadger, Ghostery, AdBlock, or similar?
If yes: Those are the current usual/suspected causes of not being able to "mark as read" successfully (and what that bug you linked as now tracking), because they can prevent cross-domain API requests (IIUC). There's ongoing work in PrivacyBadger to fix the general problem. Other browser extensions can usually be fixed by whitelisting the other Wikimedia domains.
If no: I've filed phab:T136878 for the "mark as read" link in the "..." menu not working. Please let me know if that also doesn't cover the issue you're experiencing.
(Note: the "mark all as read" button which is shown at the top of the flyout, iff you have un-read local notifications, will not clear the cross-wiki bundle. This is to reduce confusion, as the cross-wiki notifications are not left in the local flyout, once read. Cross-wiki notifications which have been read, will eventually be shown in special:notifications - they're working on that at phab:T115316.)
Re: cross-wiki preferences, Xasoflux added your request (thanks!), and there are some followup questions for you there. Thanks, HTH. Quiddity (WMF) (talk) 20:04, 2 June 2016 (UTC)
I'm not aware that I'm running any of those browser extensions, but who knows what gets installed behind my back. I generally don't mess with browser extensions.
Yes, phab:T136878 describes what I was getting. I could not clear the notice count from 1 back to 0. When I clicked the "X", not only did it drop the count back to 0, it also removed the Wikidata alert from my alert list entirely. So I can't recall or tell you what the particular alert was now, I forgot which particular page move it was notifying me about. wbm1058 (talk) 20:31, 2 June 2016 (UTC)

Odd error when using efn

A recent minor edit to Mark XIV bomb sight has caused a large error message to appear due to an empty REF tag. There is no ref tag at that point, it's an EFN. There are other EFNs on the page that are working fine. Can someone spot what's going on here? Maury Markowitz (talk) 01:38, 29 May 2016 (UTC)

It looks like User:Kind Tennis Fan already fixed this. — xaosflux Talk 02:10, 29 May 2016 (UTC)
@Maury Markowitz and Kind Tennis Fan: As AustralianRupert noticed, the problem was the equals sign inside the URL; but the change to using {{cite web}} added undesirable quote marks. All that was necessary was to explicitly number the parameter, i.e. use {{efn|1=An image instead of {{efn|An image. --Redrose64 (talk) 17:22, 1 June 2016 (UTC)
Thanks RR. So any tag with an equals inside will be something to look for? I'm not 100% I understand the cause here. Maury Markowitz (talk) 01:18, 2 June 2016 (UTC)
@Maury Markowitz: When you have a construct like
{{efn|An image of the Mk. XI is available at [ this page].}}
that is seen by the MediaWiki parser as a template {{efn}} with one parameter. When a parameter has an equals sign anywhere in it (outside of templates that are entirely enclosed by that parameter), the first equals sign is taken as the separator for a name/value pair. So, there is a parameter named An image of the Mk. XI is available at [ and its value is detail&db=object&pageSize=1&id=125598 this page]. The {{efn}} template is not set up to have a parameter of that name, so it is ignored. By inserting 1= at the start, you are altering the position of the name/value pair separator, so now we have a parameter named 1 and its value is An image of the Mk. XI is available at [ this page]. A parameter whose name is a pure integer is taken to be an override for the similarly-numbered positional parameter, so we gain the original intention at the loss of an extra two bytes. --Redrose64 (talk) 18:46, 2 June 2016 (UTC)

Facilitating! Excellent explanation as always RR! Maury Markowitz (talk) 23:42, 2 June 2016 (UTC)

Hide one user's edits from watchlist

Is there any way to hide a single user's edits from your watchlist? I know it is possible to hide my own edits and to hide all minor, bot, etc., edits, but I want to hide just edits from one user. There is a guy making a bunch of minor category edits to tons of articles on my watchlist, and they are drowning out the important edits I want to see. I can't find any option; does anyone know if this is possible? Thanks!--Dudemanfellabra (talk) 04:20, 2 June 2016 (UTC)

This is not a native option, but you could try User:UncleDouggie/smart watchlist.js - it has many options. — xaosflux Talk 04:29, 2 June 2016 (UTC)
As far as "tons" of "minor" edits - if this is really a huge number, they may need to run as a bot. You can also hide minor changes, assuming they are being marked as such. — xaosflux Talk 04:30, 2 June 2016 (UTC)
Thanks for pointing me to that script, which I didn't know existed. It does appear to include an option to hide a single user's edits, as well as many other options that could be useful. I may install it later, but shortly after I posted this comment and before your response, I thought about scripting and wrote up a short script to do only what I wanted it to do. I just tested it, and it works for me. For those interested, the code is below:
var WatchListHiddenUsers=["Hmains"] // include multiple users if you wish, e.g. ["Hmains","Dudemanfellabra","user3",...]

function WatchlistHideUsers() {
    if (mw.config.get('wgPageName')!="Special:Watchlist") return; // only fires on watchlist
    if (!=='[object Array]') return; // if you haven't defined any hidden users, exits
    var edits=document.getElementsByClassName("mw-userlink") // find links to user pages
    if (edits==null) return; // if none, quit

    for (var i=0; i<WatchListHiddenUsers.length; i++) { // cycle through list of editors to be hidden
        for (var j=0; j<edits.length; j++) {
            if (edits[j][i])!=-1) edits[j]"none" // hide edits by those editors

$(window).load(WatchlistHideUsers); // run script when watchlist loads
--Dudemanfellabra (talk) 04:57, 2 June 2016 (UTC)
Thanks for the update @Dudemanfellabra: - feel free to update Wikipedia:Customizing_watchlists#Hide_pages_in_your_watchlist :D — xaosflux Talk 00:49, 3 June 2016 (UTC)


Why is the box at Special:BookSources spread over four lines instead of two, why does it have fuzzy edges? Both of these waste space, and mean that extra scrolling is now necessary where it wasn't before. It looks very much like it's Vector skin, although I use MonoBook. And what's the asterisk for? --Redrose64 (talk) 23:51, 2 June 2016 (UTC)

I don't know about the asterisk but the layout must be mw:MediaWiki 1.28/wmf.1#Core changes (installed here 10 May):
I'm not a fan of the large space use in mw:OOUI. Wikipedia:Village pump (technical)/Archive 146#WhatLinksHere broken resulted in a conversion to OOUI being reverted at phab:T135773. I use Special:ExpandTemplates a lot and also find OOUI annoying there. PrimeHunter (talk) 01:06, 3 June 2016 (UTC)
The asterisk indicates that the field is required, and an attempt to submit the form without filling in the field will be rejected. — This, that and the other (talk) 08:02, 3 June 2016 (UTC)

Geo co-ords not connecting to

When I click the geo co-ords at the top of a geo article, it fails to get a response from Eventually get: 502 Bad Gateway nginx/1.9.4. Nurg (talk) 08:55, 2 June 2016 (UTC)

I was about to report the same issue. I'm still looking for a way around it right now. ---------User:DanTD (talk) 13:15, 2 June 2016 (UTC)
It's working now. Nurg (talk) 09:45, 3 June 2016 (UTC)

MathML glitch

Display of formulas in composition algebra suddenly came up red though rendering had been successful on Preview. — Rgdboer (talk) 21:48, 31 May 2016 (UTC)

Problems with math display

I'm seeing issues with <math> ... </math> expressions. Apparently there is a separate server to process these expressions? I'm getting messages such as:

Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "/mathoid/local/v1/":)

This isn't on a page I'm editing, just one I'm viewing. (Specific example at this point in time: Deconvolution (permalink to failed version). WikiDan61ChatMe!ReadMe!! 21:39, 31 May 2016 (UTC)

A search on "MathML with SVG or PNG fallback" finds more examples. A purge has fixed all examples I have tested so it seems the problem is either gone or only affects some servers. PrimeHunter (talk) 22:12, 31 May 2016 (UTC)
This is related to T131177 which changes the default of the maths renderer form the 10 year old PNG renderer to a newer SVG system which uses server side mathjax rendering. It was made the default today. You may need to change your preferences to get things to work right, I had "MathML with SVG or PNG fallback" as my preference and I needed to change things to PNG and back again to get things to work. --Salix alba (talk): 23:02, 31 May 2016 (UTC)

Stepped outside, just returned. Formulas rendering well now. Thank you everyone for the information and efforts. — Rgdboer (talk) 01:50, 1 June 2016 (UTC)

Still a lot of messed up pages throughout all wikimedia wikis. I'm searching for Math+extension+cannot+connect+to+Restbase with 78 errors across all languages.--Salix alba (talk): 04:48, 1 June 2016 (UTC)
I did a round of manual purging, so most of these should be fixed now. There was a brief outage related to a Cassandra upgrade not long after the preference switch (which caused a lot of re-renders), which resulted in these errors. Sorry about that. -- GWicke (talk) 15:10, 1 June 2016 (UTC)
Could you please check Kessler syndrome, I see the same error. --Michelino12 (talk) 11:00, 2 June 2016 (UTC)
I did an extra round of manual purges today, including Kessler syndrome, which now renders fine. Do note, however, that there are subtle differences in formula parsing between the old and new modes, so in some cases purges won't do it. If you encounter problems of this sort, please create a task in Phabricator and link it to T136931. Marko Obrovac, Wikimedia Foundation (talk) 11:25, 3 June 2016 (UTC)

Xtools down?

I've noticed over the past few days that anything requiring Xtools is non operational. I get that little spinning thing and the word "Connecting". Eventually it goes to 502 Bad Gateway. Same thing if I try to access Xtools directly from Labs: [1] — Maile (talk) 14:00, 3 June 2016 (UTC)

@Maile66: Same for me - possibly related to the section above? T136787 related, fix seems to be asking for a restart of the tool - thoughts @MusikAnimal:? -- samtar talk or stalk 14:06, 3 June 2016 (UTC)
Hmm I restarted all of xTools since the replica db issue, so this is unrelated. xTools requires a regular restart due to memory leaks and some other gobbledygook, but that should happen automatically. I'll investigate why it didn't, but anyway, I've manually rebooted and seems to be working now. Best MusikAnimal talk 15:39, 3 June 2016 (UTC)
Everything is working now. At least, for me. — Maile (talk) 15:40, 3 June 2016 (UTC)

Cite journal filler again

Sorry to bring this up again but the cite journal filler is still not working for me--when I try to use it to fill out a template by entering the doi or pmid, in many cases the template doesn't even load at all when I click on the "cite journal" button on the cite tab on the toolbar. I really hope someone knows what I can do to fix this problem permanently. Everymorning (talk) 20:36, 1 June 2016 (UTC)

I get this too if the doi is invalid, or sometimes it is valid, but it is for a book chapter and not a journal article. It would be good to get feedback if a problem is detected, and what that problem is. Graeme Bartlett (talk) 23:57, 1 June 2016 (UTC)
If this is an ongoing problem, then maybe we should talk about whether to switch it to the mw:citoid service. Alternatively, click the pencil icon in the upper right corner of the wikitext editor's toolbar and jump into the visual editor for a moment. "Cite" works really well there with most dois and URLs to PubMed. (Paste in the whole URL from PubMed, not just the PMID number.) Whatamidoing (WMF) (talk) 22:30, 3 June 2016 (UTC)

Permanently use Desktop version on a phone or tablet

A reader contacted Wikimedia (ticket:2016052610025225), desiring to turn off the mobile mode permanently.

According to Wikipedia:Village_pump_(technical)/Archive_136#How_to_disable_mobile_interface.3F, if they have cookies enabled and click on the desktop mode it should be permanent. However, they claim this is not so, or if so then the cookies must be expiring quickly because the reader has to re-enable the desktop option every couple of days.

Can we provide any advice?--S Philbrick(Talk) 13:08, 28 May 2016 (UTC)

It's not permanent but these cookies last for 30 days (as mandated by wmf:Privacy policy). User must have cookies enabled, and prefferrably JS too. Max Semenik (talk) 00:53, 29 May 2016 (UTC)
If they are using a browser, they may be able to workaround by changing their load page to this. — xaosflux Talk 02:09, 29 May 2016 (UTC)
I know this is then just for logged-in users, but can it be made permanently through a user-.css or .js (or a gadget or preference)? --Dirk Beetstra T C 05:06, 29 May 2016 (UTC)

Sphilbrick, Beetstra: You can add = 'm.', '' );

to Special:MyPage/minerva.js. Nirmos (talk) 09:11, 29 May 2016 (UTC)

@Nirmos: Nope, that results in a constant refresh of the page (at least in the non-mobile). --Dirk Beetstra T C 10:33, 29 May 2016 (UTC)
Beetstra: You added it to Special:MyPage/common.js. It needs to be in Special:MyPage/minerva.js. Nirmos (talk) 10:54, 29 May 2016 (UTC)
Ah, thanks .. have to start keeping track of the different .js-s. At the moment it is not clear what .js does what and when (I thought this was skin-dependent ..). --Dirk Beetstra T C 10:58, 29 May 2016 (UTC)
@Nirmos: that had the same effect on mobile than what I earlier saw on a regular site: continuous refresh of the mobile site. --Dirk Beetstra T C 15:52, 29 May 2016 (UTC)
Beetstra: Does it work better if you make the assignment conditional?
if ( 'm.' ) ) { = 'm.', '' );
Nirmos (talk) 19:08, 29 May 2016 (UTC)
I'm afraid that doesn't work. As long as the desktop page redirects to the mobile site, you will get stuck in a loop. -- [[User:Edokter]] {{talk}} 19:21, 29 May 2016 (UTC)

Hello all,

I'm the user whom Stephen was assisting, and many thanks to Stephen for being patient with me. So, is there a solution for this, other than having to be logged in all the time (and having to re-log in every 30 days)? I use 4 tablets and 2 mobile phones, with various browsers installed on each of them. If the only way to avoid the mobile site is to be logged in everywhere all the time, I'm afraid it's not very user-friendly. Plus, I use Wikipedia in various languages (at least 4 or 5 different languages quite regularly), which would multiply the obligation for me to be logged in everywhere to a quite unbearable degree (each mobile device & each browser on the device & each Wikipedia language version in each browser).

Sigh. I just wish Wikipedia never introduced this (to me) awful "tablet mode". My iPad Pro 13" has a larger screen than my MacBook 12 – to encounter Wikipedia's "tablet mode" on it is quite absurd. But I find it every bit as objectionable on the iPhone – I'm perfectly fine with Wikipedia's desktop version on the iPhone as well. Worst of all, the switch to "Desktop mode" is placed at the very bottom of each page, so if you land on a long page, you first must scroll to the very bottom of it (an ordeal on touch-screen devices), and only then you can get rid of the mobile version. If at least that switch to desktop mode was placed at the top of every Wikipedia page, I wouldn't mind having to activate it manually from time to time. As it is, though, I find the situation pretty much infuriating.

Thank you to anyone who could provide a user-friendly solution for this.

Alex aka Faterson (talk) 22:26, 3 June 2016 (UTC)
Isn't this just begging for a preferences setting solution?--S Philbrick(Talk) 23:26, 3 June 2016 (UTC)

Transclusion count link not working?

At WhatLinksHere, the "Transclusion count" link isn't working for me. Is the service down? Example:

— Andy W. (talk ·ctb) 17:17, 2 June 2016 (UTC)

It looks like Jarry1250 is the maintainer for that tool on wmflabs. Jarry1250, any information on this? — xaosflux Talk 17:45, 2 June 2016 (UTC)
Is Template linking and transclusion check by Chameleon the tool you're referring to? That seems to be nonresponsive. wbm1058 (talk) 17:56, 2 June 2016 (UTC)
I'm referring to "templatecount", not "templatetransclusioncheck" (didn't realize that tool had a problem as well). But if this depends on this, that makes sense...? — Andy W. (talk ·ctb) 18:10, 2 June 2016 (UTC)
Perhaps someone from Community Tech (Tool Labs support) can help? BTW, congrats to MusikAnimalm:User:MusikAnimal (WMF) on your new job :) wbm1058 (talk) 18:26, 2 June 2016 (UTC)
Thanks! We had an issue with the replica databases yesterday (phab:T136787) and I noticed some tools such as xTools needed to be rebooted after the replicas came back online. These two tools might also need a kick MusikAnimal talk 19:09, 2 June 2016 (UTC)
I should have done this earlier, but I just logged on to #wikimedia-labs connect and asked an admin to reboot those two tools, and sure enough they're back now :) Probably a bunch of other tools that need a restart too MusikAnimal talk 22:10, 2 June 2016 (UTC)
@MusikAnimal: Edit count needs restarted. Can someone write a bot that runs on something independent of wmflabs that regularly monitors the status of these, and orders a restart when it detects that they are down? This is a too-often recurring problem. – wbm1058 (talk) 11:21, 3 June 2016 (UTC)
Actually, we had an automatic restart script but had to disable due to unrelated Labs issues. However I think those issues have been resolved. So now we can set the cron to call that script every 5 minutes, when before the service was auto-restarting every two hours. Hopefully it won't misbehave... FYI, it's been the plan for some time to rewrite the tools from the ground up, but it's been a slow process MusikAnimal talk 06:12, 4 June 2016 (UTC)
Sorry, got this confused with thread below. xTools is an edge case, most tools run on an lighttpd server which as I understand is automatically restarted on Tool Labs should they go down. Technically the lighttpd servers powering the translcusions tools didn't go down; my guess is the connection to the database was lost, and a restart was needed to re-establish it. In short, most tools shouldn't go down unless something dramatic happens (arguably infrequent), xTools on the other does go down regularly and requires a dedicated restarting mechanism -- hence why it needs a rewrite :) MusikAnimal talk 06:26, 4 June 2016 (UTC)

Firefox section "jumping" bug?

Could some other users of Firefox test these links to Idea Lab:

Firefox "jumps" for me to a section much lower down on the page rather than to the desired section. (Actually, at first it does start loading at the target section but then it jumps.) When I load the page in a new tab it sometimes does not jump (but sometimes does). I've just tried version 46.0.1 for Linux using safe-mode and a new profile and the bug still occurs. This has been a long standing issue for me but I've only now been annoyed with it enough to try to track it down. Jason Quinn (talk) 06:42, 4 June 2016 (UTC)

This behaviour happens for me too. But it doesn't happen in Safari. Sandbh (talk) 07:16, 4 June 2016 (UTC)
e/c The second of those links takes me further down the page than the first, and oddly, doesn't show as a visited link when I came back here. I'm using 46.0.01 and Win 10. I've had this issue for so long I've come to think of it as normal, through many iterations of firefox. I suspect it is because I never clear my cookies or browsing history. It happens all over the project, and I'd like it if it didn't. -Roxy the dog™ woof 07:18, 4 June 2016 (UTC)ADDING - It doesn't happen on my awful ipad in Safari though. -Roxy the dog™ woof 08:34, 4 June 2016 (UTC)
Known problem, plenty of similar threads in VPT archives. The two collapsible boxes in Wikipedia:Village pump (idea lab)/Archive 20#Extended confirmed protection policy are loading in uncollapsed state, then your browser moves to the correct heading, then the boxes collapse pulling the rest of the page up with them. --Redrose64 (talk) 08:53, 4 June 2016 (UTC)

Deletion reason gone missing

Normally when I delete an article which has a speedy tag on it, the deletion reason in the speedy tag is copied to the "other/additional reason" box on the deletion page. This useful feature has stopped working. I have tried different browsers and different machines. Bring the feature back, please. — RHaworth (talk · contribs) 15:02, 3 June 2016 (UTC)

Hmm, are you sure? I'm still seeing it in Firefox. Normally, I *don't* see it, since I usually check the article history before deleting, and that feature doesn't work from the article history. But clicking on the delete button directly from the page still seems to work for me. Writ Keeper  15:07, 3 June 2016 (UTC)
I also wrote a script to keep that function working from the history tab too (so long as you're going from the main article page to the history page, at least, which is how my workflow goes): User:Writ Keeper/Scripts/historyDeleter.js, if you're interested. Writ Keeper  15:46, 3 June 2016 (UTC)

Possible clue: when I delete an article which has a talk page, I get offered a link to delete the talk page. If I click on that link, I am provided with the appropriate "other/additional reason". — RHaworth (talk · contribs) 12:46, 4 June 2016 (UTC)

None of my gadgets are working

I've tried this on two computers and two different browsers (Google Chrome and Internet Explorer, which both assure me I'm using the most up to date version). All the gadgets I use have stopped working, the most noticeable of which are that I can no longer rollback edits other than with my Rollback rights and I now have to tag articles for deletion manually because Twinkle isn't working. They work on college computers, but there Reflinks and Dabfix don't. What am I doing wrong?--Launchballer 10:47, 3 June 2016 (UTC)

Do you have a working show/hide link in the table of contents on this page? (that requires JavaScript). Have you tried to clear your entire cache? What is your skin at Special:Preferences#mw-prefsection-rendering? Do you see a clock in the upper right corner at Do you see a clock on normal pages if you enable "Add a clock to the personal toolbar" under "Appearance" at Special:Preferences#mw-prefsection-gadgets? How about the similar UTCLiveClock at commons:Special:Preferences#mw-prefsection-gadgets when you view normal Commons pages like commons:? Gadgets don't run when you are in your preferences pages so you have to test it on other pages. PrimeHunter (talk) 12:34, 3 June 2016 (UTC)
My show/hide link works, I've just cleared it, my skin is Monobook, I see clocks in all cases.--Launchballer 19:42, 3 June 2016 (UTC)
If the clock gadget works then some gadgets are able to run. Try to disable the broken gadgets, save preferences, enable one of them again and save again. Do gadgets work in the Vector skin? User:Launchballer/monobook.js doesn't run when you are in Vector. PrimeHunter (talk) 19:59, 3 June 2016 (UTC)
HotCat works a treat having changed the skin to Vector and back. Twinkle works in Vector, but I can't stand it. I suspect it may be a problem with my Monobook.js file.--Launchballer 21:06, 3 June 2016 (UTC)
It would be prudent in that case to temporarily disable the JS in your monobook.js by commenting it all out (place a /* before it and a */ after it all). By this action, you can establish if something there is or is not the root of the issue(s). Fred Gandt · talk · contribs 21:56, 3 June 2016 (UTC)
I have disabled it, and it now works. I'd love to know what's malfunctioning in there, since - as far as I can remember anyway - the only thing I added to it was DYKcheck.--Launchballer 23:03, 3 June 2016 (UTC)
var numNotes = /\d+/.exec($("#pt-notifications").text())[0]; throws an error because there is no "#pt-notifications element. Ruslik_Zero 20:23, 4 June 2016 (UTC)

Talkback template bug or my mistake?

I've been attempting to place a {{talkback}} template on an IP's talk page and point to a non-user talk page. Talk:Princess Charlotte of Cambridge is the specific page I'm trying to point to. I'm using Google Chrome on a Mac El Capitan OS, and when I place this code: {{Talkback|Talk:<Princess Charlotte of Cambridge>|[1st birthday]|ts=~~~~}}, per the instructions, it produces a talkback template which, where the name of the talk page should be, has this: [[User talk:Talk:<Princess Charlotte of Cambridge>#[1st birthday]|User talk:Talk:<Princess Charlotte of Cambridge>]]. Am I using the template wrong, or is there a bug somewhere here? Thanks! Chickadee46 (talk) 16:30, 5 June 2016 (UTC)

@Chickadee46: {{talkback|Talk:Princess Charlotte of Cambridge|1st birthday|ts=~~~~}} seems to work fine. In general, I believe {{Reply to}} (or {{Re}}) is used more often nowadays.not IPs — Andy W. (talk ·ctb) 16:38, 5 June 2016 (UTC)

Screen rendering

Minor glitch - now resolved

Going to wp:CCI looks fine in Mozilla, but in Chrome, the screen bleeds off to the right.

Mozilla: File:CCI Mozilla.jpg

Chrome: File:CCI Chrome.jpg

Note that "skip to Instructions" should be in the middle of the page, it is in Mozilla, but way to the right in Chrome.

The change occurred with this edit

Any thoughts?--S Philbrick(Talk) 16:06, 5 June 2016 (UTC)

Actually, it was the edit before that, adding a rediculously long unbreakable string. Should be fixed now. -- [[User:Edokter]] {{talk}} 17:05, 5 June 2016 (UTC)
Thanks, I actually knew to look for a ridiculously long string, but someone, looked at the wrong edit. My guess is that Chrome takes it literally, while Mozilla decides to override it?--S Philbrick(Talk) 17:37, 5 June 2016 (UTC)

Usb → USB

Someone should stop bots doing this and this to chemical element. Maybe it is not just one article (the one I've mentioned).--Obsuser (talk) 01:31, 5 June 2016 (UTC)

Those aren't bots. Bots don't correct typos/spelling. I suggest using inline HTML comments to note that it is correct as "Usb". FYI: Obsuser and Project Osprey. — JJMC89(T·C) 02:57, 5 June 2016 (UTC)
It is AWB, actually ("semi-bot"). Other articles might get changed too, and this one is on my Watchlist so I will wait several days/weeks.--Obsuser (talk) 03:08, 5 June 2016 (UTC)
You could try wrapping it in {{Not a typo}} or {{Text}} olderwiser 03:35, 5 June 2016 (UTC)
Or the AWB-user should read what s/he is changing. When you add "Regex typo fixing" in AWB, it says "Check each edit before you make it.[...]". It is obvious it shouldn't be USB. Christian75 (talk) 08:18, 5 June 2016 (UTC)
If Usb is commonly used then the USB entry could be removed from Wikipedia:AutoWikiBrowser/Typos#New capitalisation. It appears rare and wasn't even listed at USB (disambiguation)#Other uses before I just added it. PrimeHunter (talk) 10:40, 5 June 2016 (UTC)
 Done I updated the typos page to not replace for initial first letter capital (Usb), it will still replace for all lc (usb). Feel free to revert if this causes any unintended impact. — xaosflux Talk —Preceding undated comment added 14:34, 5 June 2016 (UTC)
...though I had already added {{Not a typo}} to Extended periodic table. -- John of Reading (talk) 19:05, 5 June 2016 (UTC)

Map template help

Arising from a Help Desk query, I edited Template:National Historical Parks of the United States map to add a site to the map (following the same code as the others) but the dot does not show up. diff Is the page too long or did I do something wrong? Alanscottwalker (talk) 20:04, 5 June 2016 (UTC)

You copied the position of Adams National Historical Park. I have changed it.[2] PrimeHunter (talk) 20:37, 5 June 2016 (UTC)
Thanks so much. If I may, how did you get to that "position" (somehow I thought it was tied to the article coords but that is obviously wrong)? Or, what do you mean by "position" in this context? Alanscottwalker (talk) 22:09, 5 June 2016 (UTC)
You copied position: absolute; left: 266px; top: 46px from the following entry in the template. These are pixel positions for the image and not geographic coordinates. I saw the right location in the infobox map in Abraham Lincoln Birthplace National Historical Park. Then I found the pixel positions for nearby Kentucky locations in the template, adjusted the numbers a little, and previewed until it looked right. I may be a pixel off. PrimeHunter (talk) 22:27, 5 June 2016 (UTC)
Thanks, again! There is kind of an invisible grid there then, gah! Alanscottwalker (talk) 22:39, 5 June 2016 (UTC)

No option to keep changes when switching to VE

I have noticed that when trying to switch to VE in the middle of an edit, it's no longer possible to choose to keep my changes. That was a very useful feature, as sometimes I switch to VE for things like deleting a table row, and I want to be able to do that without saving a half-complete edit. nyuszika7h (talk) 20:00, 4 June 2016 (UTC)

I think there’s no option not to keep the changes. The “switch” button keeps the edit both ways. --Tacsipacsi (talk) 20:52, 4 June 2016 (UTC)
You can't keep your changes if you begin in the wikitext editor by opening a section (rather than the whole page). Before the devs disabled this, the net result was everything except that section getting silently blanked.
I hope that they will someday be able to make this work properly (i.e., switch, keep changes, and not blank the rest of the page). Whatamidoing (WMF) (talk) 05:29, 6 June 2016 (UTC)

Visually broken dates in Multiple issues rendering


I've seen this for as long as I can remember, and have finally decided that I've had enough; can someone (else - I'm passing the buck) add the appropriate CSS rule to the dates in whichever module or template (Message box?) outputs them, to keep them from snapping please? Fred Gandt · talk · contribs 21:23, 5 June 2016 (UTC) P.S Chrome on Win7Fred Gandt · talk · contribs 11:02, 6 June 2016 (UTC)

This seems to be a Chrome (Blink) issue, as Firefox is not affected. If it's root is in Webkit, Safari may be affected as well. Toggling the display flag in the console seems to 'fix' it temporarely. I'm afraid no CSS can fix this. -- [[User:Edokter]] {{talk}} 22:08, 5 June 2016 (UTC)
Hmm, ticking show/hide also clears the issue. Perhaps the code for mw-collapsible is to blame? -- [[User:Edokter]] {{talk}} 22:11, 5 June 2016 (UTC)
The screenshot is from Lea Salonga discography. PrimeHunter (talk) 22:17, 5 June 2016 (UTC)
If the date's <i> is given class nowrap, shouldn't that fix it? It does if added manually via the inspector/console. Fred Gandt · talk · contribs 23:05, 5 June 2016 (UTC)
.mbox-text i {
	white-space: nowrap;
to my common.css; seems to work. Fred Gandt · talk · contribs 02:20, 6 June 2016 (UTC)
OK, how does it look when the messages are long, as with Abbey? --Redrose64 (talk) 08:36, 6 June 2016 (UTC)
With my CSS, that looks fine Redrose64; with my CSS disabled, the longer one's date didn't snap, but the shorter did (table element width 1483px).Fred Gandt · talk · contribs 11:02, 6 June 2016 (UTC)
I don't think I can reproduce this Chrome issue on Mac OS X btw.. —TheDJ (talkcontribs) 09:53, 6 June 2016 (UTC)

Autotranslate bug or display error?

For some bizarre reason, the content is getting switched on c:File:Пропаганда по стените на патлейнски манастир.jpg so that part of the first template is transcluded into the parent template. Oddly, this doesn't happen when I try the same templates on c:Special:ExpandTemplates. Can anyone figure this out? I'm not good enough with #invoke to know how to debug it. Magog the Ogre (tc) 02:21, 4 June 2016 (UTC)

@Magog the Ogre: There are several templates - and other transclusions - at c:File:Пропаганда по стените на патлейнски манастир.jpg. Which specific ones do you mean by "the first template" and "the parent template"? --Redrose64 (talk) 08:50, 4 June 2016 (UTC)
Redrose64, the page looks like this, but if I copy and paste all of the wikicode to c:Special:ExpandTemplates, I see this in the preview on the special page. I did this which somehow fixed the problem, although I'm not sure why. --Stefan2 (talk) 20:31, 4 June 2016 (UTC)
<center> should be killed with fire. In this case, it may have caused HTML Tidy to throw things around to fit its own idea of 'legal' HTML. I replaced it with a div. -- [[User:Edokter]] {{talk}} 21:33, 4 June 2016 (UTC)
Well how are users supposed to view the page if they're using Internet Explorer 3??? Magog the Ogre (tc) 00:10, 5 June 2016 (UTC)
@Magog the Ogre: I honestly can't tell if that's sarcasm... we have a large number of users come through here using N-years out-of-date browsers on a regular basis... --Izno (talk) 12:37, 6 June 2016 (UTC)
@Magog the Ogre: OK, so by "the first template" you actually meant the {{PD-BG-exempt}} and by "the parent template" you actually meant the {{2-D in PD}}. To me, the first template is the {{Information}} (although the first transclusion is {{int:filedesc}}).
I note that since this thread was started, the file description page has been altered by Eesn (talk · contribs) to add {{CC-BY-SA-3.0}} which seems to be incompatible with the two claims for non-copyrightable work. --Redrose64 (talk) 12:52, 5 June 2016 (UTC)
@Redrose64: hope not to derail the technical discussion, this is a photo I've taken myself, of a government-commissioned work that is of cultural/historical significance. The work itself is uncredited, it refers to entities that no longer exist, and I don't think Bulgaria has an equivalent of PD-Polish. The three tags I've placed on it are to do with this particular situation (Two for the reason I think the work is in the public domain, and one to release my photo of it under CC.) I welcome any advise on this. — Preceding unsigned comment added by Eesn (talkcontribs) 16:05, 5 June 2016
Eesn actually added that license per my suggestion on his talk page. There is a lot to the photograph other than the 2D artwork, so I didn't think that PD-art applies. Magog the Ogre (tc) 02:04, 6 June 2016 (UTC)

Tech News: 2016-23

20:51, 6 June 2016 (UTC)

Advanced review log (reversions)

Advanced Review Log includes an introductory line stating,
This page lists the most recent pending changes reviews in a format sortable by namespace, action and type (manual or automatic).

But it lists only the articles where the changes were approved. Where is the list of pending changes reviews where the changes were reverted? —Anomalocaris (talk) 21:39, 6 June 2016 (UTC)

They are included but are relatively rare and may have no examples in the first page with 50 entries. Select "Unaccepted" in the "Review action" box to see them. PrimeHunter (talk) 21:59, 6 June 2016 (UTC)
I mean articles where the reviewer clicked the "Revert Changes" button on the review screen, and on the Special:Review revisions page, the reviewer clicked "Revert these changes". —Anomalocaris (talk) 22:31, 6 June 2016 (UTC)

Wikipedia Adventure for blocked users

I see that this indefblocked user has been enrolled into The Wikipedia Adventure: [6]. It might be an idea for the Wikipedia Adventure mechanism (which I'm surprised to see can "simulate" edits), to check for account blocking before trying to do this. -- The Anome (talk) 08:40, 6 June 2016 (UTC)

Hi The Anome, that's a good idea - did that user have their talk page access revoked? The block log suggests not, otherwise TWA wouldn't have been able to do this edit -- samtar talk or stalk 08:46, 6 June 2016 (UTC)
No, their talk page access hadn't been revoked, but I don't think we should be inviting blocked users to play a game, regardless. I think this goes for both indefblocked users, and also non-indefblocked users who might be blocked at that particular moment. I'm also a bit perplexed about "simulating" edits, as this makes tracking editor activity more difficult without any practical benefit that I can see: perhaps it might be better for the Wikipedia Adventure bot to edit under a single-purpose account name of its own? -- The Anome (talk) 08:48, 6 June 2016 (UTC)
The Wikipedia Adventure is a JavaScript script running client side and making edits with limited contribution by the user but it's still user edits and can only edit pages the user is able to edit. Blocked users can usually edit their own talk page. "simulated automatically" is just an edit summary chosen by the script. It should be possible for the script to check whether the user is blocked and refuse to run in that case, but it seems pretty harmless when some blocked users make all sorts of bad edits to their talk page. I strongly oppose users being able to make edits credited to a bot account instead of their own account. Many stages of The Wikipedia Adventure allow the user to add their own content. Even if it can only save in the user's own userspace (not sure that's the case), many important policies also apply there, and pages can also be moved later. PrimeHunter (talk) 12:18, 6 June 2016 (UTC)
Working as designed, the only edits were edits that can be made by the blocked user already - and indefinite is not infinite - maybe TWA will help them get over whatever they did to get blocked and could be used as an appeal. — xaosflux Talk 02:16, 7 June 2016 (UTC)


I'm having trouble at the moment with clicking links. I'll click a link once and it'll act as if I've double clicked it and the editing window will appear. What's causing the problem?♦ Dr. Blofeld 09:13, 6 June 2016 (UTC)

It works for me in Firefox and Vector. What is your browser and skin? Does the mouse show signs of unwanted double clicking elsewhere? Try disabling "Edit pages on double click" at Special:Preferences#mw-prefsection-editing. PrimeHunter (talk) 11:58, 6 June 2016 (UTC)
I had this problem a few months ago, and it turned out to be an issue with a corrupted mouse driver treating single-clicks as double-clicks. Does the same thing happen elsewhere (e.g., single-clicking a desktop icon acting as a double-click and opening it)? ‑ Iridescent 14:54, 6 June 2016 (UTC)
I've disabled the double click edit article feature in my preference to stop it happening. Yes, may well be a mouse problem, I've not noticed the problem elsewhere though.♦ Dr. Blofeld 12:08, 7 June 2016 (UTC)

Unable to access "Your alerts"

I have an alert notice (red with "1") and when I click on the icon it fades a bit, but does not show what the alert is about and it does not go away. I tried resetting the notifications in Preferences, but that did not fix it. Would someone please reset it? Thanks. —Ojorojo (talk) 15:07, 4 June 2016 (UTC)

Does Special:Notifications work? PrimeHunter (talk) 17:32, 4 June 2016 (UTC)
I have what seems to be the same problem: clicking on the red alert button (or Special:Notifications) puts me on the Notifications page, but all I get is a box with the rolling diagonal bars. First saw this in the small hours this morning, and still happens. Pinging you (@PrimeHunter:) to see if you get the same thing. ~ J. Johnson (JJ) (talk) 18:21, 4 June 2016 (UTC)
And instantaneous update: In the course off adding the previous comment I checked, and the situation was unchanged. Immediately after Saving I checked again, and ... it worked. Odd. Coincidental, or did the Save actually do something? ~ J. Johnson (JJ) (talk) 18:25, 4 June 2016 (UTC)
More oddness: This morning my watchlist started displaying article (main space) names in big text. (I am using the Vector skin.) Coincidentally with alerts now working my watchlist has returned to normal display (all titles highlighted, but all text normal size). ~ J. Johnson (JJ) (talk) 18:34, 4 June 2016 (UTC)
I have also noticed random updates from 'Your alerts' and have learned to ignore them. I was going to this page to report that Wikipedia:Tools/Navigation popups are working again, as of today. Navpops weren't working for me in the last day or so. --Ancheta Wis   (talk | contribs) 18:55, 4 June 2016 (UTC)
Alerts work for me. I have Firefox 46.0.1 and Vector. PrimeHunter (talk) 19:53, 4 June 2016 (UTC)
Now I get rolling diagonal bars when clicking on the red alert icon (same with Special: Notifications). —Ojorojo (talk) 13:46, 5 June 2016 (UTC)
@Ojorojo: Do you mean that it freezes and does not complete loading, after you see the diagonal bars? If so, the freeze in loading sounds like a gadget/userscript is interfering (I see you don't use any userscripts). Next time you experience this, please could you open your webconsole (per browser/OS instructions are in the last point of Wikipedia:Reporting JavaScript errors), then repeat the actions, and paste what appears in the webconsole? It should look something like this screenshot. (Specifically: Open your webconsole, refresh the wikipage you're looking at, then click the notifications badge (button), then copy&paste the text in the webconsole to here, or the WT:Notifications page.) Any additional details, such as the exact links that each error points to (e.g. this screenshot), and other thoughts on what gadget might be involved, and notes on any javascript-blocking browser-extensions you might have installed, would also be useful. Thanks. Quiddity (WMF) (talk) 19:31, 6 June 2016 (UTC)
@Quiddity: It's now back to the icon just fading a bit when I click on it (nothing else happens). The alerts are for a revert[7] and your recent ping. I haven't changed anything (gadgets, blocks, etc.) and there is nothing special in my set-up. —Ojorojo (talk) 15:06, 7 June 2016 (UTC)

What links here / templates

It would be useful if there were a filter for "What links here" results to exclude the effect of templates in articles. For example, many articles that link to Prohibition should actually link to Prohibition in the United States, typically piped as [[Prohibition in the United States|Prohibition]], i.e. Prohibition, or to the article about prohibition in some other country. Many articles that show up as linking to Prohibition don't contain the link directly, but are pulled in because of templates such as {{Prohibition}}, {{Criminal law}}, {{Social issues in Pakistan}}, or {{Social issues in Nepal}}. As another example, many articles that link to Murder, Inc. don't contain the link directly, but are pulled in because of one of the template {{Organized crime groups in New York City}}, {{Lucchese crime family}}, or {{Genovese crime family}}.

It would be useful to be able to to retrieve an instantaneously-updated list of articles that link to a page directly and not through templates that actually include the wikilink. —Anomalocaris (talk) 21:18, 6 June 2016 (UTC)

It's a frequently requested feature. Here are some of the requests:

PrimeHunter (talk) 21:50, 6 June 2016 (UTC)
PrimeHunter: Thanks. —Anomalocaris (talk) 22:31, 6 June 2016 (UTC)
For that record Search can facilitate such cleanups. Search offers linksto and hastemplate.
See Link reports at Help:Linking. — Cpiral§Cpiral 19:25, 7 June 2016 (UTC)

BLP problems with pending changes at Amber Heard

This is a weird problem, and also a serious one. It suggests that Google showed its readers the latest version of an article, regardless of whether the pending change had been approved or not. Since the whole point of pending changes is not to show readers edits that have not been automatically accepted, this may need to be looked at to prevent similar things from happening in the future. Amber Heard is now semi-protected.--♦IanMacM♦ (talk to me) 11:40, 4 June 2016 (UTC)

This rather depends on how Google is getting the data for its Knowledge Graph. It's not difficult to write a program to get the latest unapproved revision, and you don't even need to be logged in to do it. It might be that on the MediaWiki side we could make it easier for them to get the latest approved revision rather than just the latest revision, but mostly I suspect that Google just needs to fix the Knowledge Graph to make it work with pending changes. Until we know how they are getting the data, I don't think there's much we can do except for reporting it to Google. — Mr. Stradivarius ♪ talk ♪ 06:07, 6 June 2016 (UTC)
If memory serves – and it may not, so perhaps someone who knows more will speak up – Pending Changes is ignored for feeds to Google due to performance problems. Whatamidoing (WMF) (talk) 17:41, 8 June 2016 (UTC)

Special:Search question about why option to create new page does not exist

Just curious since this is probably a technical issue: I recently tried to search for an all-caps shortcut in the "Wikipedia:" namespace using Special:Search. However, usually when the search function realizes that the page doesn't exist, it provides a red linked option to create the page, but for this page, the option did not exist. However, through trial and error, I found out that this issue only happened with searches/nonexistent pages that started with "Wikipedia:" or "WP:" and the inclusion of the all-caps string of "REPORT" anywhere in the title. Is there some sort of setting causing this purposely, or is this an error of some sort? (I'm guessing that it's the former, but if it is the former, I'm curious where the page is [probably "MediaWiki:"] that is causing this to happen.) Steel1943 (talk) 15:03, 8 June 2016 (UTC)

"OR" between search terms is one of the Help:Searching#Search engine features so the software thinks it's search syntax and not a potential page to be created. OR doesn't work as search syntax in the middle of a word like REPORT but the software apparently isn't smart enough to figure that out. It's not set at Wikipedia. The same applies to other wikis with the same search software. Before this feature we got some silly page creations like page names containing prefix: or intitle:. PrimeHunter (talk) 17:32, 8 June 2016 (UTC)
See also phab:T23102: "Don't propose to create a page under a title with prefix:, intitle:, incategory: or other search operators". PrimeHunter (talk) 17:36, 8 June 2016 (UTC)
Interesting. Looks as though I will have to use one of the more complicated ways to make the referenced page(s). Makes sense though about "OR", though I was thinking there was something blacklisted instead. Anyways, thanks for the explanation. Steel1943 (talk) 18:47, 8 June 2016 (UTC)
Yes, such page names can still be created, for example via a red link WP:REPORT. Or search a similar name like WP:REPRT, click the red link in the search page and then add 'O' to the url. PrimeHunter (talk) 19:46, 8 June 2016 (UTC)


Hello all,

I've a suggestion that will allow the specific code {{Birth date|YYYY|MM|DD}} to be used to display nothing. This is useful to be able to silently include it in {{Infobox person}} (and profession-specific variants of it such as {{Infobox scientist}}) examples. For new users inserting {{Infobox person}}, I've found it common that they don't understand the significance of the birthdate being commented out (<!-- {{Birth date|YYYY|MM|DD}} -->). When they insert the dates into the {{birth date}} template within the infobox, it is still commented out and doesn't display (examples below). I've put the code at Template:Birth date/sandbox. (ping Jonesey95, Izno).

{{Infobox scientist
| name        = 
| birth_date  = <!--{{birth date |1987|02|11}}-->

This "unfilled" {{birth date/sandbox}} template can be included uncommented out (({{Birth date/sandbox|YYYY|MM|DD}})), such that it works normally as soon as dates are filled in without having to remove the commenting:

{{Infobox scientist
| name        = 
| birth_date  = {{birth date |1987|02|11}}
Example 1a
Example 1b
Example 2a
Born(YYYY-Expression error: Unrecognized word "mm".-DD)Expression error: Unrecognized word "dd"., Expression error: Unrecognized word "yyyy".Expression error: Unrecognized word "yyyy".Expression error: Unrecognized word "mm".Expression error: Unrecognized word "dd".
Example 2b
Born(1987-02-11)February 11, 1987
<!-- {{birth date |YYYY|MM|DD}} -->Example 1a (used in blank {{Infobox person}} template)
<!-- {{birth date |1987|02|11}} -->Example 1b (incorrectly filled in {{Infobox person}} template if <!-- --> not removed). Example 1b should show "February 11, 1987", but doesn't because user didn't remove <!-- -->
{{birth date/sandbox |YYYY|MM|DD}}Example 2a (for use in blank {{Infobox person}} template and doesn't need <!-- -->)
{{birth date/sandbox |1987|02|11}}Example 2b (filled in {{Infobox person}} template without need to remove <!-- -->)

Hopefully the examples make sense. Opinions and suggestions welcomed. T.Shafee(Evo﹠Evo)talk 12:01, 9 June 2016 (UTC)

Personally, I would change the code in {{infobox person}} to accept any arbitrary date format, e.g. YYYY-MM-DD, DD Month YYYY, or Month DD, YYYY. Then I would add the microformat <span class="bday">...</span> tags to that unless they were already present. That way new users can put any date format they like in the infobox field, and we don't have to treat {{birth date|YYYY|MM|DD}} as a special case. However, your proposal is better than the status quo, and I wouldn't oppose it. — Mr. Stradivarius ♪ talk ♪ 12:14, 9 June 2016 (UTC)
The {{birth date}} template is not the only one used in that infobox parameter, for living people the {{birth date and age}} template is normally used. But both of those templates are unsuitable for non-Gregorian dates; and the full birth date is not always available, for which {{birth year}} etc. are preferable. I'm sure the idea of moving the metadata classes into {{infobox person}} has come up before, and rejected as impractical. --Redrose64 (talk) 12:43, 9 June 2016 (UTC)
I see some legitimate concerns for the specific implementation, but I like the general concept of a template which, when the parameter is a placeholder simply doesn't display, but when the parameter is replaced with a valid entry then displays.--S Philbrick(Talk) 15:44, 9 June 2016 (UTC)

Wikipedia is running slowly

I have been trying to edit pages, and whenever I try to access pages on English Wikipedia, it is taking close to 15-30 seconds to load, or is giving me an error. Please {{ping}} me when you reply. --Jax 0677 (talk) 18:29, 9 June 2016 (UTC)

@Jax 0677: There is not any information in your request that we can use to troubleshoot. Is the delay/error only occurring when you try to edit, or any access at all? What is the error message? Mamyles (talk) 21:06, 9 June 2016 (UTC)
Reply - It is now back to normal. --Jax 0677 (talk) 21:15, 9 June 2016 (UTC)

[Edit] links MIA on user talk page

Hi. Corinne is missing [Edit] links on all but the last six threads on her User talk:Corinne page. Where's Waldo? Cheers! {{u|Checkingfax}} {Talk} 22:41, 9 June 2016 (UTC)

And they're back. --Redrose64 (talk) 22:59, 9 June 2016 (UTC)

Arbitration clerk call for script writer [cross-post]

The Clerks of the Arbitration Committee are looking for a script writer who will work with the clerk team to automate portions of the clerks' procedures. If you are a skilled script writer and are interested in working with us, please email the clerk team at

For the Clerks of the Arbitration Committee, Kevin (aka L235 · t · c) 05:31, 10 June 2016 (UTC)

Discuss this at: Wikipedia talk:Arbitration Committee/Noticeboard#Arbitration clerk call for script writer

Ogg video subtitles will not activate on the Wikipedia

Hi, I was referred here from the help desk. I checked this video with English subtitles here in the WP article. Before, they did come up, but now the subtitles are not activating. I thought it may be related to some language preferences of mine, but my search engine and Windows is English based, so why do not they activate, even when tapping on the "English subtitles" option. On the other hand, subtitles do activate on Wikimedia commons. Any hints welcome! Iñaki LL (talk) 07:33, 10 June 2016 (UTC)

This is related to phab:T122737. I've been reworking some parts of the subtitle support, to make it closer to HTML5 video support, but it's not completely finished yet. I've been a bit busy at work, but hope to return to this soon and get this problem fixed. —TheDJ (talkcontribs) 09:55, 10 June 2016 (UTC)
Appreciated TheDJ. Iñaki LL (talk) 10:04, 10 June 2016 (UTC)

Claude Billard

I've just created this article but I unfortunately linked it with the same title to the French article of which it is a translation whereas the link should have directed to fr:Claude Billard (écrivain). Can someone please correct my mistake and tell me how to undo such errors? Thanks in advance, LouisAlain (talk) 13:30, 10 June 2016 (UTC)

 Done: I removed the link you added to wikidata:Q2977009 and moved it to wikidata:Q2977011. —  crh 23  (Talk) 13:37, 10 June 2016 (UTC)
Thanks a lot, LouisAlain (talk) 14:31, 10 June 2016 (UTC)

Email throttle

I'm emailing people on m:Wiki Project Med#People interested, telling them the membership list is open and inviting them to fill in the form. I've just received a message, after 10 or so emails, saying "Action throttled: As an anti-abuse measure, you are limited from performing this action too many times in a short space of time, and you have exceeded this limit. Please try again in a few minutes." Is there any way I can get un-throttled? --Anthonyhcole (talk · contribs · email) 05:42, 9 June 2016 (UTC)

If you are doing that on meta: about the only way to get noratelimit would be to become a member of the accountcreator group over there (same here on enwiki) - this is not normally issued just so you can send lots of email, but you can ask at WP:PERM. If you want to notify a lot of people here on the English Wikipedia the best way is with talk page messages. Your message can be mass delivered by requesting at WT:MMS. — xaosflux Talk 00:23, 10 June 2016 (UTC)
Closed out via WP:PERM. — xaosflux Talk 15:20, 10 June 2016 (UTC)

Wikipedia to the Moon: voting has begun

Hello, after six weeks of community discussion about Wikipedia to the Moon, there are now 10 different proposals for content for the mission. Starting today, you can vote for them on Meta-Wiki, and decide what we will work on: a Wikipedia canon, different lists, the Moon in 300 languages, an astronomy editathon, featured articles, articles about technology, endangered things, or DNA-related topics. You can even vote against community involvement. Voting is open until 24 June. Sorry that this message is again in English only, but we are using village pumps to reach as many communities as possible, so that everyone knows they can vote. Best, Moon team at Wikimedia Deutschland 15:31, 10 June 2016 (UTC)

Where did the color on "Collapse top" template go?

Why is the Template:Collapse top template colorless all of a sudden, and how can I put the green color back in? For one thing, I have my Barnstars collapsed on my userpage, so they won't hog the page like a giant advertisement. But I do want people to notice that the collapsed section is there -- and for that it needs color, otherwise nobody sees a collapsed section. Why did the color get removed, and how can I make it green again? Softlavender (talk) 05:02, 12 June 2016 (UTC)

@Softlavender: Answer to question "why?" can be found here. --Edgars2007 (talk/contribs) 05:12, 12 June 2016 (UTC)
The default color was changed (with precious little participation in the decision, but whatever), and it looks like not all of the doc was updated to reflect the change. Try adding |bg=#cfc and see if that's the color you wanted. ―Mandruss  05:22, 12 June 2016 (UTC)
@Softlavender: I restored the color on your page for you. — xaosflux Talk 05:24, 12 June 2016 (UTC)
OK thanks for now. Softlavender (talk) 05:28, 12 June 2016 (UTC)
  • There should at least be instructions on the Template page on how to add color. This is an extremely drastic change to something that has been standard, and standard practice, for many many years. The color was there for a reason. Softlavender (talk) 05:28, 12 June 2016 (UTC)
The background color is specified using the |bg= parameter, which is documented. ―Mandruss  05:34, 12 June 2016 (UTC)

Template-protection of TemplateSubster force?

Should User:AnomieBOT/TemplateSubster force be template-protected, rather than fully protected? Full protection was added to the page before templateeditor came about. I believe that most template editors handle templates and substitution fairly often to understand how the page is used, how it relates to {{Subst only}}, etc. (Perhaps I should disclose that I hold the template editor right.) Note that a page like Wikipedia:AutoWikiBrowser/Template redirects affects thousands of AWB editors, and is only semi-protected. Any thoughts? — Andy W. (talk ·ctb) 07:28, 5 June 2016 (UTC)

In my understanding, substitution is actually preferred, unless a template satisfies some of these a template should be transcluded if:
  • The template contains a large amount of text that, if substed, clutters the page.
  • The template produces intricate syntax (template guts, parser function calls, etc) that makes the surrounding text hard to navigate, or, if the syntax is slightly modified on a future edit, could cause major visual damage to the page.
  • The template contains timely information that, if substituted, might be lost or not be updated.
  • The template is very visual (i.e. message boxes, infoboxes, tables, divs). Generally, incremental improvements continue to happen on these, and articles and other pages that transclude these need all the updates.
There's been a discussion on templates like {{ESp}} about whether they should be substed or not. In my understanding, the reason they're substed is that they don't contain absolutely vital information that needs to be kept up with template changes, and its substitution produces not too much code that makes a page much less navigable. Furthermore, these "minimal" templates deserve substitution to lessen the probability of major server churn upon template update. Hence, I think I'm not wrong when I say that the reason this bot service exists is due to the job queue, and performance reasons. I also think that most TEs would understand the extremely negative implications of adding an intricate template to the list. The page in question may benefit from having TEs access for management/housekeeping. Interested in hearing thoughts. — Andy W. (talk ·ctb) 08:15, 5 June 2016 (UTC)
Notifying Anomie. — JJMC89(T·C) 08:41, 5 June 2016 (UTC)
Thanks. I'm happy to let the community decide what level of protection is appropriate for the page. I've updated the bot to accept template-protection there. Anomie 11:26, 5 June 2016 (UTC)
@Anomie: Policy wise - as a bot operator you are ultimately responsible for the edits made under you bot account, so changing your bot's operating parameters isn't really a community decision; you can pick what ever level of protection you deem to be appropriate. You can't transfer your accountability to others. — xaosflux Talk 14:27, 5 June 2016 (UTC)
Corollary: You are also responsible for updating your bot to follow changes to community decisions, so this may help you let others guide that path. I've go no objections to whatever level of protection you deem appropriate. — xaosflux Talk 14:30, 5 June 2016 (UTC)
If I had an objection to reducing the protection level, I'd have objected. Template-protection should serve the purpose adequately, I think we can trust template editors to take sufficient care that they don't instruct the bot to substitute templates in a disruptive manner. Anomie 20:28, 5 June 2016 (UTC)
@Xaosflux: There doesn't seem like there's been a resolution to this yet. Any further thoughts? Thanks. — Andy W. (talk ·ctb) 17:04, 6 June 2016 (UTC)
@Andy M. Wang: What part of WP:SUBST states that substitution is actually preferred? --Redrose64 (talk) 15:06, 5 June 2016 (UTC)
@Redrose64: It doesn't, but I'm generally under the impression that the opposite (large amounts of unnecessary transclusion) can be costly in terms of performance. I'm aware of WP:PERF, but as a TE aware of the job queue, I think perf is something to keep in mind. — Andy W. (talk ·ctb) 15:44, 5 June 2016 (UTC)
WP:SUBST#Benefits, fifth bullet, states that "substituting en masse may ultimately speed up the site, but this is not a reason to prefer substitution over transclusion". Indeed, WP:SUBST#Drawbacks, fourth bullet, reinforces that. --Redrose64 (talk) 18:22, 5 June 2016 (UTC)
My two cents: since technical competence in the realm of template maintenance is assumed for template editors and they are also well trusted users, I'd see no problem whatsoever with them being able to modify that page, especially if they follow the guidelines that they normally do for editing template protected pages. Further, since the responsibility for the bot ultimately lies with Anomie and he has raised no objections, if no other objections are made I see no reason not to change the protection level. —  crh 23  (Talk) 17:13, 6 June 2016 (UTC)
I think this entire matter is completely up to @Anomie: , as an admin they can adjust the protection - or can decide they want to approve edits to their own page using the edit request process. — xaosflux Talk 17:26, 6 June 2016 (UTC)
I basically agree: I am slightly cautious as bot actions are somewhat community controlled by bot approvals, but I think this case is pretty clear: it's up to Anomie. —  crh 23  (Talk) 17:56, 6 June 2016 (UTC)
  • This would be very useful. As someone who closes and enacts consensus for TfD discussions regularly, it's often useful to convert a template to a wrapper and mass-substitute it. ~ RobTalk 17:39, 7 June 2016 (UTC)

Since no one objected here, I went ahead and reduced the protection level. Anomie 13:08, 12 June 2016 (UTC)

Bash (probably) script by another editor

Moved from Wikipedia:Reference desk/Computing: Jc86035 (talk • contribs) Use {{re|Jc86035}} to reply to me 03:36, 7 June 2016 (UTC)

I've been working near WT:RDT recently and I'm trying to run scripts which Pldx1 wrote (to standardise the {{BS}}–{{BS14}} and {{BS-2}}–{{BS14-2}} series of templates and add /safesubst subpages), probably in Bash or another command language. Unfortunately they don't seem to be working, and I don't have any idea what's wrong with them. Could someone help fix (and possibly run) the scripts? Thanks, Jc86035 (talk • contribs) Use {{re|Jc86035}} to reply to me 09:43, 6 June 2016 (UTC)

For reference, the scripts are here.

nummax=115 # (i.e.  n<15)

##----------------- old ----------------------------------------------------------------------- computing the old BSn

for ((numbo=101; numbo<$nummax; numbo++)) ; do 
  num=$(($numbo-100)) ; vers="tmp_BS${numbo:1:2}-old.txt"
  echo -n "<includeonly>" > $vers
  if test $num -lt 12 ; then prefo="O"; else prefo="O0"; fi

cat <<EOF >> $vers

  for ((i=2;i<=$num;i++)); do
    if test $num -lt 12 -o $i -gt 9 ; then prefo="O"; else prefo="O0"; fi
cat <<EOF >> $vers
{{BS-overlap|{{{PX$i|{{#if: {{{PX|}}}|{{{PX}}}|{{BSpx}}}}}}}|{{{$i|}}}|{{{$prefo$i|}}}|{{{$prefo${i}2|}}}|{{{$prefo${i}3|}}}|{{{$prefo${i}4|}}}|{{{$prefo${i}5|}}}|link={{{L${i}|file:}}}|alt={{{alt${i}|#default}}}}}
cat <<EOF >> $vers


##------------------ safesubst ----------------------------------------------------------------- computing the new BSn

for ((numbo=101; numbo<$nummax; numbo++)) ; do 

  echo "<includeonly>" > tmp_vers
  for ((i=1;i<=$num;i++)); do
    if test $num -lt 12 -o $i -gt 9 ; then prefo="O"; else prefo="O0"; fi
    cat <<EOF >> tmp_vers
{{#IF:{{{$prefo${i}|}}}|!~{{{$prefo${i}|}}} }}
{{#IF:{{{$prefo${i}2|}}}|!~{{{$prefo${i}2|}}} }}
{{#IF:{{{$prefo${i}3|}}}|!~{{{$prefo${i}3|}}} }}
{{#IF:{{{$prefo${i}4|}}}|!~{{{$prefo${i}4|}}} }}
{{#IF:{{{$prefo${i}5|}}}|!~{{{$prefo${i}5|}}} }}
cat <<EOF >> tmp_vers
      |~~ ~~{{{$(($num+2))|}}}
      {{#IF:{{{$(($num+1))|}}}{{{km|}}}|~~{{{$(($num+1))|{{{km|}}}}}}~|~~ ~}}
      |~ ~
        |~ ~~ ~
        |{{#IF:{{{bg|}}}|~ ~~ ~~ ~}}
        |~~{{{$(($num+4))|}}}{{#IF:{{{bg|}}}|~~bg={{{bg|}}} }}
        |{{#IF:{{{bg|}}}|~~ ~~bg={{{bg|}}} }}
        {{#IF:{{{$(($num+2))|}}}|~~ ~~}}
        {{{$(($num+4))|}}}{{#IF:{{{bg|}}}|~~bg={{{bg|}}} }}
           {{#IF:{{{$(($num+2))|}}}|~~ ~~ ~~}}

  sed tmp_vers -e "s¶^[ ]*¶¶ ; s¶{{#IF:¶{{{{{|safesubst:}}}#if:¶ " | tr -d "\n" | sed -e " s¶}}} }}¶}}}}}¶g; s¶}}} }}¶}}}}}¶g " > $vers
  echo "" >> $vers 



##------------------ old-2 --------------------------------------------------------------- computing the old BSn-2

function bs-double (){

  if test $num -lt 11 ; then prefo="O" ; else prefo="O0"; fi
  cat <<EOF >> $vers

  for ((i=2;i<=$num;i++)); do
    if test $num -lt 11 -o $i -gt 9 ; then prefo="O" ; else prefo="O0"; fi
  cat <<EOF >> $vers
{{BS-overlap|{{{PX$i|{{#if: {{{PX|}}}|{{{PX}}}|{{BSpx}}}}}}}|{{{$i|}}}|{{{${prefo}$i|}}}|{{{${prefo}${i}2|}}}|{{{${prefo}${i}3|}}}|{{{${prefo}${i}4|}}}|{{{${prefo}${i}5|}}}|link={{{L${i}|file:}}}|alt={{{alt${i}|#default}}}}}


cat <<EOF >> $vers

for ((numbo=101; numbo<$nummax; numbo++)) ; do 
  echo -n "<includeonly>" > $vers
  diff -s $vers ../BSn-tmp/$vers


##------------------ safesubs-2 --------------------------------------------------------------- computing the new BSn-2

for ((numbo=101; numbo<$nummax; numbo++)) ; do 

  cat <<EOF > $tmp_vers
  cat <<EOF >> $tmp_vers
      ~ ~
    ~! !
    {{{$(($num+1))|}}}! !

  for ((i=1;i<=$num;i++)); do
    if test $num -lt 11 -o $i -gt 9 ; then  prefo="O"; else prefo="O0"; fi
    cat <<EOF >> $tmp_vers
{{#IF:{{{$prefo${i}|}}}|!~{{{$prefo${i}|}}} }}
{{#IF:{{{$prefo${i}2|}}}|!~{{{$prefo${i}2|}}} }}
{{#IF:{{{$prefo${i}3|}}}|!~{{{$prefo${i}3|}}} }}
{{#IF:{{{$prefo${i}4|}}}|!~{{{$prefo${i}4|}}} }}
{{#IF:{{{$prefo${i}5|}}}|!~{{{$prefo${i}5|}}} }}

cat <<EOF >> $tmp_vers
    |~~ ~~{{{$(($num+2))|}}}
      ~~ ~
      |~ ~
        |~ ~~ ~
        |{{#IF:{{{bg|}}}|~ ~~ ~~ ~}}
        |~~{{{$(($num+5))|}}}{{#IF:{{{bg|}}}|~~bg={{{bg|}}} }}
        |{{#IF:{{{bg|}}}|~~ ~~bg={{{bg|}}}}}
        {{#IF:{{{$(($num+2))|}}}|~~ ~~}}
        {{{$(($num+5))|}}}{{#IF:{{{bg|}}}|~~bg={{{bg|}}} }}
          {{#IF:{{{$(($num+2))|}}}|~~ ~~ ~~}}

  sed $tmp_vers -e "s¶^[ ]*¶¶ ; s¶{{#IF:¶{{{{{|safesubst:}}}#if:¶g " | tr -d "\n" | sed -e " s¶}}} }}¶}}}}}¶g; s¶}}} }}¶}}}}}¶g " > $vers


(Moved comment, seeing as no one responded.) Jc86035 (talk • contribs) Use {{re|Jc86035}} to reply to me 03:36, 7 June 2016 (UTC)

What precisely happens when you try to run the scripts? "They don't seem to be working" is unhelpful.
In both scripts, try replacing all occurrences of with / in the sed line. In the second script, add nummax=115 at the top to match the first script, and replace $tmp_vers with tmp_vers. With those changes, they work for me (or seem to; I'm not sure what they're supposed to do). -- BenRG (talk) 04:21, 9 June 2016 (UTC)
@BenRG: Sorry for the lack of detail. Based on the behaviour of the first script, the scripts are supposed to create files ("tmp_BSn-old.txt" and "tmp_BSn-safesubst.txt" for script 1; "tmp_BSn-old-2.txt" and "tmp_BSn-2-safesubst.txt" for script 2) in the user's home folder, which should contain MediaWiki template markup similar to the text of Template:BS11 and Template:BS11/safesubst. Adding your changes, the problem for the first script becomes that the "-safesubst" files are empty (and a file called "tmp-vers" containing what should be in "tmp_BS14-safesubst.txt", but with all the newlines and spaces intact, is created). With the second script… nothing happens, like before. Not sure why. Jc86035 (talk • contribs) Use {{re|Jc86035}} to reply to me 14:10, 9 June 2016 (UTC)
Just... no. --MZMcBride (talk) 15:07, 9 June 2016 (UTC)
@MZMcBride: well, I could always do all 56 of them manually. :/ Jc86035 (talk • contribs) Use {{re|Jc86035}} to reply to me 10:59, 10 June 2016 (UTC)
I have no problem with quick scripts to get a job done. But the learning curve here seems pretty steep, even for a technical village pump. ParserFunctions and parts of wikitext mixed with moderately advanced shell scripting? Yowza. --MZMcBride (talk) 19:18, 12 June 2016 (UTC)

Tech News: 2016-24

18:41, 13 June 2016 (UTC)

Not getting pings

I haven't been getting ping notifications for the past couple of days though there's nothing wrong withe pings themselves. I checked my notifications settings (prefs/notifications) and nothing seems to have changed there either with web notifications selected for everything. I know of at least three pings that I did not get because those pages are on my watchlist example, not sure how many I've missed. Is this a general issue or a problem unique to me caused by something stupid I did recently? cheers. —SpacemanSpiff 01:24, 13 June 2016 (UTC)

@SpacemanSpiff: Ping! mw:Manual:Echo mentions that for pings to work "The comment must either belong in its entirety to existing sections (the diff chunk of added lines must not contain new section headers) or start a new section (diff chunk starts with a new section header)." - could be the reason it failed for the diff you provided. Besides {{U}} is a shortcut for {{User link}} and does not always work the same as {{Ping}}. Could you give more diffs? - NQ (talk) 05:27, 13 June 2016 (UTC)
@NQ: I got your ping this time, but {{U}} is what I've always used at SPI and I've never had pings not delivered before. I've also pinged both @Tokyogirl79: and @Bbb23: using that at SPI and I'm quite sure they got it, just as I've got their pings using that in the past. I also didn't get this. I can't seem to remember the other one that I didn't get the ping for but accidentally stumbled on. The last ping I received before yours was this. cheers. —SpacemanSpiff 05:59, 13 June 2016 (UTC)
@SpacemanSpiff: I was able to reproduce the faulty diff and since no pings were sent to me either, it must be something wonky with the way the diffs are being parsed by Echo. Hopefully, Redrose64 or Quiddity (WMF) can pinpoint the exact issue. Anyway nothing wrong with your notification system. - NQ (talk) 06:57, 13 June 2016 (UTC)
Thanks, makes me feel better that it's not me! cheers. —SpacemanSpiff 07:27, 13 June 2016 (UTC)
@SpacemanSpiff and NQ: In the diff above, there are 69 calls to the {{checkuser}} template, each of which creates links to a different user's userpage. Add the call of {{U|NQ}}, and that makes 70 user links. Echo, however, will not issue any notifications if more than 50 users are linked. This is to prevent someone leaving a talk page message like You have been mentioned at {{WP:ANI}}. ~~~~ by mistake, which would transclude the whole of ANI and notify anyone that happened to have their user page linked there. — Mr. Stradivarius ♪ talk ♪ 07:32, 13 June 2016 (UTC)
Thanks, that makes sense for the second example, but the first one is still a bit puzzling. I'll just ignore that as an anomaly. cheers. —SpacemanSpiff 08:43, 13 June 2016 (UTC)
The first one might have failed because it included headings (this is the "must not contain new section headers" mentioned above), I'm not sure. On the matter of templates like {{u}} vs {{ping}} and others - what matters is not what the template is called or what it looks like when rendered, but whether or not it yields a link to a user page. --Redrose64 (talk) 09:08, 13 June 2016 (UTC)
Yup, there are convoluted requirements for mentions to work. There's some ongoing work at phab:T135717 (and subtasks, primarily phab:T135719 at the moment) that is aiming to make it clearer to us when a probable attempt at a mention didn't work - the current end goal is likely to be: a notification is sent to me, when my edit contained what appears to be a mention, but no mention was actually triggered - this is more reliable and unmissable than adding the details into the popup-bubble save-message (as previously suggested). Quiddity (WMF) (talk) 20:02, 13 June 2016 (UTC)

Apparent glitch has disabled section editing

See User talk:Stunteltje. I tried a couple of possible fixes but they didn't accomplish anything.— Vchimpanzee • talk • contributions • 20:57, 13 June 2016 (UTC)

@Vchimpanzee: Whoever welcomed that user copied part of the Wikipedia:Welcoming_committee/Welcome_to_Wikipedia portal page to that talk page instead of using one of the real welcome templates. Since that is a portal page, it uses the {{Box header}} template which disables section editing by default. I cleaned up the template a bit so that it should no longer be an issue for that page. It might be a good idea to implement a similar fix to the other users that αѕєηιηє welcomed back in 2008. --Ahecht (TALK
) 21:38, 13 June 2016 (UTC)
I haven't seen the problem on any of the pages I've checked so far. Thanks. Time to return to the real world.— Vchimpanzee • talk • contributions • 21:43, 13 June 2016 (UTC)
The problem was that Wikipedia:Welcoming committee/Welcome to Wikipedia/box-header uses {{Box-header}} which suppresses section edt links by default with __NOEDITSECTION__. I have overridden the default by adding |EDIT=yes to the call [15] so the around 200 pages at Special:WhatLinksHere/Wikipedia:Welcoming committee/Welcome to Wikipedia/box-header now have section edit links. PrimeHunter (talk) 21:46, 13 June 2016 (UTC)

Cross-wiki watchlist feedback

Hi everyone,

The Wikimedia Foundation Community Tech team has started the design work for a Cross-wiki watchlist feature, which we were tasked to do by the community. Our goal is to build a beta feature that users will enable, replacing their current watchlist with a functional, easy-to-browse cross-wiki watchlist, if they want to. We'll be pinging some different Wikimedia communities throughout the process. If you're active on more than one Wikimedia wiki and would find this useful, feel free to take a look and tell us what you think.

I'm posting some early wireframes here that show the basic structure of how the watchlist items would look, along with some filters. There's two views pictured in these wireframes that you could toggle between, using the filters – a single chronological list that shows entries from all the wikis that you choose, or splitting up the list by project.

The feedback that we're looking for: Which elements would fit your workflow, and which would get in your way? Does the "chronological list" vs "split by project" filter make sense? Which one do you think you would use? What are the important missing pieces?

There's more discussion about this on Meta: m:Talk:Community Tech/Cross-wiki watchlist. Feel free to post your thoughts here, or there. What do you think? /Johan (WMF) (talk) 21:51, 13 June 2016 (UTC)

Template and template redirect transclusion counts

Ahecht and I had a good discussion here about what exactly the transclusion count yielded by this tool actually means. I have been under the impression that the count given includes itself and all the transclusions accumulated by redirect aliases. I have also never seen an alias with a transclusion count greater than the main template. (Is that just lack of experience?)

Anyway, Ahecht pointed out to me that the transclusion counts of {{Reply to}} is less than the sum of the counts of its redirects. (Again, see here.) Any details about what exactly the templatecount tool is counting? Thanks — Andy W. (talk ·ctb) 20:04, 13 June 2016 (UTC)

The thing to do is to look at a template with somewhat fewer transclusions, and add that up manually to see how the figure is reached. Consider for example Template:FC Armavir squad, which has a transclusion count of 20, whilst Special:WhatLinksHere/Template:FC_Armavir_squad has 43 entries, 40 of which are transclusions - but observe that every single one of those transclusions is duplicated, once (not indented) for the template proper, once (indented) for the redirect Template:FC Torpedo Armavir squad. This redir lists 20 transclusions at Special:WhatLinksHere/Template:FC Torpedo Armavir squad, and also shows a transclusion count of 20. So I would say that the transclusion count for the template proper includes transclusions of its redirs but does not double-count. --Redrose64 (talk) 20:26, 13 June 2016 (UTC)
Ahh, so the discrepancy is that by manually adding up the redirects, I'm double counting pages that use two different redirects (for example, a page that used both {{ping}} and {{reply to}}). Makes sense. --Ahecht (TALK
) 21:06, 13 June 2016 (UTC)
Yes, the tool counts how many pages are transcluding the template but each page is only counted once no matter how many transclusions it makes or which redirects it uses. PrimeHunter (talk) 21:20, 13 June 2016 (UTC)
I dispute the advice of my esteemed colleagues in a couple of respects. The tool does a simple count(*) query on templatelinks.tl_title [16]. tl_title can contain a template name or a redirect name. Using RR64's example, I looked at page ID 19729356, one of the footballers, and find two rows in templatelinks, one for FC_Armavir_squad and one for FC_Torpedo_Armavir_squad [17]. Based on this, and my presumptions about the table structure [18] [19] I think that:
  • Querying against the template name in the tool does not count the redirects
  • The table will contain only a single row for any combination of Page_ID and tl_title, which is to say no matter how many transclusions of the same name there are in a page, there'll only be one row, but...
  • The table can have multiple rows which resolve to the same template (i.e. the 19729356 example) because the table links a Page_id to the transclusion name as used in the page - and that might be the template name or it might be the redirect name, and if both are present then there'll be two rows.
  • Summing results from the tool will be prone to overcounting the number of pages that link to the template, since a single page can be listed once for each template/redirect name found in the page
  • But, to reiterate the first point, one would need to sum the redirects and the template results, since the template result does not include the redirects
  • And in sum: Quarry is capable of answering the question you want answering, but I'm not sure my SQL is up to it tonight. --Tagishsimon (talk) 23:07, 13 June 2016 (UTC)
Your points sound contradictory so I'm not sure what you mean. The question was about That tool counts every page where a template is transcluded, no matter whether the template was transcluded via its own name or the name of a redirect. For example, says 5 transclusions even though no page says {{Hair space}}. All 5 transcluding pages at Special:WhatLinksHere/Template:Hair space say {{hairsp}}. PrimeHunter (talk) 23:46, 13 June 2016 (UTC)
Yes. On reflection, I withdraw the above and now agree: querying the template name in the tool does include the redirects. --Tagishsimon (talk) 00:00, 14 June 2016 (UTC)
Search parameter hastemplate: given a canonical pagename it report content transclusions, but given an alias (i.e. redirect) instead of a canonical, reports named transclusions. — Cpiral§Cpiral 06:25, 14 June 2016 (UTC)

Bible Verse Finder fails with some translations

I'm not sure who maintains this tool anymore; apparently the person who migrated it a couple years ago no longer maintains it. Anyway, I noticed that it fails to find the JP and JPR translations on, because the site reorganized and the API that this script used no longer works; also, it has problems with some other translations, according to commenters on the template's talk page. In particular, I found this problem while looking at Religious restrictions on the consumption of pork and for now, I think I'll link to the JPR chapter pages as they currently exist on, bypassing this tool. Julyo (talk) 23:10, 12 June 2016 (UTC)

Although Template:bibleverse (talk links edit) is highly visible (4000 pp.), it is just a protected and simple wrapper to an unmaintained tool. Problems like this must be addressed before these tools become too highly visible, or else users will make it our problem when wmflabs tools break. — Cpiral§Cpiral 08:34, 14 June 2016 (UTC)

math ml rendering changes and scribunto

Perhaps related to MathML glitch because all of this was once working.

The cs1|2 citation templates produce COinS metadata. For those parameters that are included in the metadata it is necessary to replace strip-markers with the relevant content before the parameter value is encoded into the metadata format. For example, a |title= parameter value might be written like this:

|title=<nowik>some protected stuff</nowik>

When it comes time to generate the metadata for citation's title, Module:Citation/CS1/COinS function coins_cleanup() calls the Scributo text library function mw.text.unstripNoWiki(). In this {{cite report}} template, the content of |title= is protected with <nowiki>...</nowiki> tags:

{{cite report |title=<nowiki>some protected stuff</nowiki>}}

And it renders correctly:

some protected stuff (Report).

and the metadata are correct (title parameter metadata has the keyword &rft.btitle=

<cite class="citation report">some protected stuff (Report).</cite><span title="ctx_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Abook&rft.genre=report&rft.btitle=some+protected+stuff&" class="Z3988"></span>'"`UNIQ--templatestyles-00000070-QINU`"'

Because it is very common for technical, math, and science sources to include equations in their titles, a similar process is followed for math strip-markers. Except. Except that there are three kinds of math rendering: PNG images, LaTeX source, and MathML with SVG or PNG fallback. I speculate that the final replacement doesn't happen until the page is served to the reader whose preference may be different from the editor. Additionally, these math renderings come with varying amounts of extra stuff (MathML being the worst offender).

Because math is rendered according to user preference settings, when Module:Citation/CS1 encounters <math>...</math> tags it calls mw.text.unstripNoWiki() to get the original content and then retrieves a more-or-less plain-text version of the equation for the metadata. Here is the citation used for the subsequent examples:

{{cite report |title=<math>\Delta{H}</math>}}
(Report). – the math part rendered according to your preference settings

In the following examples, we are looking for the metadata keyword &rft.btitle= and the content of the <cite>...</cite> tags. When the preference is set to PNG images, this:

<cite class="citation report"><img class="mwe-math-fallback-image-inline tex" alt="\Delta{H}" src="//" /> (Report).</cite><span title="ctx_ver=Z39.88-2004&" class="Z3988"><span style="display:none;"> </span></span>

and when the preference is set to LaTeX source:

<cite class="citation report"><span class="mwe-math-fallback-source-inline tex" dir="ltr">$ \Delta {H} $</span> (Report).</cite><span title="ctx_ver=Z39.88-2004&" class="Z3988"><span style="display:none;"> </span></span>

But, and this used to work, when the preference is set to MathML:

<cite class="citation report">?'"`UNIQ--math-0000001C-QINU`"'? (Report).</cite><span title="ctx_ver=Z39.88-2004&" class="Z3988"><span style="display:none;"> </span></span>
(the question marks are place holders for the delete characters (0x7f) which are in this page's source)

Here is the MathML markup for this simple citation:

<span><span class="mwe-math-mathml-inline mwe-math-mathml-a11y" style="display: none;"><math xmlns="" >
    <mrow class="MJX-TeXAtom-ORD">
      <mstyle displaystyle="true" scriptlevel="0">
        <mi mathvariant="normal">&#x0394;<!-- Δ --></mi>
        <mrow class="MJX-TeXAtom-ORD">
    <annotation encoding="application/x-tex">{\displaystyle \Delta {H}}</annotation>
</math></span><img src="" class="mwe-math-fallback-image-inline" aria-hidden="true" style="vertical-align: -0.338ex; width:4.02ex; height:2.343ex;" alt="\Delta{H}" /></span>

Module:Citation/CS1/COinS attempts to fetch this stripmarker's value with a call to mw.text.unstripNoWiki() as it did for the other two preferences. For this preference selection, the module looks for the <annotation>...</annotation> tags and extracts the content using this line of code:

rendering = rendering:match ('<annotation[^>]+>(.+)</annotation>') (rendering is the variable holding the content associated with this strip marker)

Because the <annotation>...</annotation> tags are not found, the module replaces the strip marker in the metadata with an error message: MATH RENDER ERROR.

A little bit of testing shows that the call to mw.text.unstripNoWiki() returns the strip marker. We know that the Module code isn't broken because the PNG and LaTeX forms work correctly and we know that the Scributo function works properly for those two forms.

It appears that the PNG to SVG changes prevent the Scributo library's function mw.text.unstripNoWiki() from doing what it should be doing.

Trappist the monk (talk) 11:03, 14 June 2016 (UTC)

@GWicke. --Izno (talk) 11:23, 14 June 2016 (UTC)
Or maybe @Gwicke. --Izno (talk) 11:24, 14 June 2016 (UTC)
This sounds like the same problem as Wikipedia talk:Lua/Archive 4#stripmarkers and mw.text.unstripNoWiki(). That was filed as phab:T121085, but the task was closed as invalid. It should probably be reopened as a bug against the Math extension. — Mr. Stradivarius ♪ talk ♪ 12:08, 14 June 2016 (UTC)
Certainly related. Changing from mw.text.unstripNoWiki() to mw.text.unstrip() does not change how the cs1|2 Module code works so this problem is not confined solely to mw.text.unstripNoWiki().
Trappist the monk (talk) 13:01, 14 June 2016 (UTC)

Disambiguation finder script

Does anyone know of a dab finder script that currently works? I used to use User:Meteor sandwich yum/dabfinder.js, but it stopped working a few months ago and the creator of that script has been away from Wikipedia for a while. Jenks24 (talk) 12:47, 14 June 2016 (UTC)

@Jenks24: User:Anomie/linkclassifier highlights disambiguations links as well. - NQ (talk) 12:54, 14 June 2016 (UTC)
Thanks for the suggestion. Unfortunately I don't think the 'on demand' function for it works with Vector, which is what I'm after (don't want it showing up all the time, just when I want to look for it). I'll keep it in mind if no one has any other suggestions. Jenks24 (talk) 13:05, 14 June 2016 (UTC)
@Jenks24: Just tried, works for me on vector. It's under the More tab as "Link Classifier". Links from todays FA 2003 Sri Lanka cyclone light up as green using the script on demand. - NQ (talk) 13:11, 14 June 2016 (UTC)
Try the script on Wikipedia:Disambiguation - NQ (talk) 13:13, 14 June 2016 (UTC)
@NQ: Aha! Right you are, not sure what I was doing before. My quibble is that I'd prefer the whole thing to be 'on demand' so to speak, whereas currently it runs straight away on changing the colour of standard links. Is that changeable? Jenks24 (talk) 13:28, 14 June 2016 (UTC)

@Jenks24: It is supposed to be fully on-demand, but I’ve noticed that the normal links change to a different color with the script installed. Is that what you’re referring to? Or does the script load and show all the different colors by default in your case? What I’d suggest is create a separate css file in your user space such as User:Jenks24/linkclassifier.css with

A.disambiguation   { background-color:#ffff88; }
A.intentional-disambiguation   { background-color:#ffffcc; }

and call it in your vector.js

importScript('User:Anomie/linkclassifier.js'); // Linkback: [[User:Anomie/linkclassifier.js]]
mw.util.addPortletLink('p-cactions', 'javascript:LinkClassifier.onDemand()', 'Link Classifier');

You can always tweak the colors to your liking using the rules from User:Anomie/linkclassifier.css. - NQ (talk) 13:44, 14 June 2016 (UTC)

Works perfectly! Thanks so much for your help, especially walking me through it all. Jenks24 (talk) 13:49, 14 June 2016 (UTC)

Small search box

Today in Firefox 47.0 the search box at Special:Search has shrunk to only display around 15 to 25 characters at a time depending on character width. If there are more characters then some of them scroll out of view. Are others seeing the same? It's the same if I log out, also at other wikis like meta:Special:Search. The box is around three times as large in IE 9.0. It happens in all four skins. PrimeHunter (talk) 11:02, 14 June 2016 (UTC)

See phab:T137535 in the bug tracker. --AKlapper (WMF) (talk) 12:26, 14 June 2016 (UTC)
Thanks. I tried searching phab: for it at which is reached by using the initial search box. I failed to find the right search but could easily have found it at by using the "Created After" field to search recent tasks. Is there a reason (like server load?) that the former "Advanced Search" doesn't mention the latter which seems more advanced in many ways? PrimeHunter (talk) —Preceding undated comment added 12:52, 14 June 2016 (UTC)
BTW, see phab:T123907 request for wider search box. CirrusSearch added seven new, long, parameter names, and the ability to code-up regex. Typing into a tiny box is out of touch with that, esp on the page designed to help refine a search. — Cpiral§Cpiral 18:38, 14 June 2016 (UTC)


On the page Wikipedia:Autopatrolled, there seems to be some ambiguity between the "nutshell" box and the subsection "Process comparison". In the nutshell box, The autopatrolled permission automatically marks new articles created by editors with this permission as patrolled. Whereas, at "Autopatrolled process", A person with the autopatrolled right creates a page... The software (rather than a human) marks the page as patrolled.

If a user is autopatrolled, is the user autopatrolled on all namespaces, or just the article namespace? — Andy W. (talk ·ctb) 18:33, 14 June 2016 (UTC)

Scratch that, Special:Diff/699873038 leads me to believe it's on all namespaces. Yeah definitely scratch that. (I didn't need to ask) Checked patrol logs on page creations outside of the space. — Andy W. (talk ·ctb) 18:40, 14 June 2016 (UTC) 18:46, 14 June 2016 (UTC)
Yes, it's all namespaces; and you can check that by going to Special:NewPages, which by default lists pages in article space, and pick a different namespace - for example, User talk. Observe the large number of recently-created pages without a yellow background. If you ever see a yellow background here it's a bug, since pages in MediaWiki space can only be created by admins, who are by definition autopatrollers as well. --Redrose64 (talk) 18:47, 14 June 2016 (UTC)

Template:Location map Belgrade problem

Hello! Please, can someone help me and create Template:Location map Belgrade using this image (File:Map of central Belgrade.png) I was trying so much, but its complicated! Please help! --Axiomus (talk) 10:05, 15 June 2016 (UTC)

Copyvio tools

Is this what you want?

Some of you may know that I'm involved with Did You Know, and part of that review process is checking for copyvios and close paraphrasing. The DYK Toolbox within the review templates already includes Earwig's Copyvio Detector and the Duplication Detector from Labs. I'm wondering if a tool currently exists that automatically detects if a new article is partially or wholly copied from an existing Wikipedia article. It would be helpful in determining "newness" of content on nominations. Using, Battle of San Jacinto as an example, sections Retreat and Aftermath were copied a year ago by me from Texas Revolution, and I so noted in an attribution on the talk page. Yet, if I run Battle of San Jacinto through Earwig's tool, only external sites are searched; with the Labs tool, we have to know the specific article to link for search. Any suggestions? — Maile (talk) 16:35, 15 June 2016 (UTC)

Earwig's tool excludes Wikipedia sites from its top 10 listing, but if you click "9 URLs with lower confidence hidden. Show them.", it will show those hits as well, which you can then compare. Doing that, you get this Intelligentsium 18:58, 15 June 2016 (UTC)
I don't see where I can click "9 URLs with lower confidence hidden. Show them." It looks to me like you specifically told Earwig to check against Texas Revolution. What I mean is when we review a DYK nom, can we do one click and see if it was copied from any other Wikipedia article. The reviewer isn't going to know what article to search, and the nominator is unlikely to reveal that. — Maile (talk) 19:10, 15 June 2016 (UTC)
It's at the bottom of the listings. Here's a screenshot, taken from a standard Earwig search. It looks like the exclusion list is just a text list - another possibility is to ask The Earwig to add a feature whitelisting certain normally-excluded sites on an individual search basis rather than editing the global list. Intelligentsium 19:19, 15 June 2016 (UTC)
Actually if you click on "compare" to the right of "Excluded", it will show you the usual highlighted comparisons. I see it now. And the number differs. I just ran one with 40 lower confidence urls, the bottom two were Wikipedia. But it's better than nothing. And it's really good to know. Thanks for the information. — Maile (talk) 19:30, 15 June 2016 (UTC)

IP editing in spite of global lock? shows two edits made on the 10th and 12th June 2016, in spite of an apparent global lock on the IP address range covering it that has been in force from April 2016. What's going on? Does this represent a failure of the global lock mechanism? -- The Anome (talk) 09:31, 12 June 2016 (UTC)

If you look at the message more closely you'll notice that the global block is locally disabled here at English Wikipedia: "locally disabled by Mike V: IP does not appear to be a mobile provider, not a colocation service, anonymous only" --Glaisher (talk) 09:34, 12 June 2016 (UTC)
Thanks. Perhaps the clarity of the messages generated in this type of case should be improved, so that this is more easily evident to the reader? -- The Anome (talk) 10:09, 12 June 2016 (UTC)
Come up with a better wording, and it may be implemented. You're better opff finding a better forum for discussing this (perhaps WP:AN). עוד מישהו Od Mishehu 14:59, 13 June 2016 (UTC)
 Done I've updated the message to be more clear and include links, see Special:Contributions/ now. — xaosflux Talk 23:05, 15 June 2016 (UTC)

Locked, Hidden

When hovering over usernames the terms LOCKED, HIDDEN appear in capital, bold italics. What could be the meaning of that? Thanks. Dr. K. 22:13, 15 June 2016 (UTC)

Assuming you're using navigation popups? —  crh 23  (Talk) 22:16, 15 June 2016 (UTC)
Yes, thank you. Dr. K. 22:17, 15 June 2016 (UTC)
It's supposed to refer to features at mw:Extension:CentralAuth#Global users but popups is wrong. MediaWiki:Gadget-popups.js has the code:
if (globaluserinfo.locked !== null) ret.push('<b><i>' + popupString('LOCKED') + '</i></b>');
if (globaluserinfo.hidden !== null) ret.push('<b><i>' + popupString('HIDDEN') + '</i></b>');
"LOCKED" may be broken due to gerrit:45362. There is probably a similar problem with "HIDDEN". I have reported it at MediaWiki talk:Gadget-popups.js#LOCKED and HIDDEN are wrong. PrimeHunter (talk) 22:37, 15 June 2016 (UTC)
Err, that change is entirely unrelated and 3+ years old. Anyways, I fixed it by reverting the latest change to popups that had an issue. Legoktm (talk) 22:48, 15 June 2016 (UTC)
Thank you both. Now it doesn't appear anymore. Dr. K. 22:53, 15 June 2016 (UTC)
OK, thanks for the fix. For testing, User:Cmhhmfo really is locked (meta:Special:CentralAuth/Cmhhmfo) and popups reports it without making false reports for others. PrimeHunter (talk) 23:19, 15 June 2016 (UTC)

Strange error

A screen-shot of Wikipedia showing a strange error.

Here is a strange, perhaps one off, error.

All the best: Rich Farmbrough, 18:27, 10 June 2016 (UTC).

This is from Dizzy Sal in monobook. It looks like MediaWiki:Sidebar was displayed to you in a faulty raw way due to missing data. Just ignore it if it's a one off. It works for me. PrimeHunter (talk) 18:51, 10 June 2016 (UTC)
Do you need to have the image this big in a discussion thread? ---Steve Quinn (talk) 04:28, 11 June 2016 (UTC)
Made it smaller, but maybe it can be made smaller. --Edgars2007 (talk/contribs) 04:47, 11 June 2016 (UTC)
It's a thumbnail, which means people can click on it for the full-size view. I made it smaller, and replaced px with upright for scaling proportionate to the user size pref. ―Mandruss  04:51, 11 June 2016 (UTC)
Just a heads up, I just experienced the same error (OSX 9.1.1). - Yellow Dingo (talk) 10:54, 15 June 2016 (UTC)
I'm getting this error on Ming–Hồ War when logged out (so Vector) in IE11 and Firefox, on Win7. Screenshot. Eman235/talk 02:14, 16 June 2016 (UTC)

Bizarre navigation box

0, 1, print export.png

What's going on? This is intermittent (my current window displays the navigation box just like normal), appearing on various pages unpredictably, and it just started. The situation made me think of a problem I caused a few years ago (see File:Holy Cow That's Weird.png and the section where it appears in the VPT archives), but the last edit to the MediaWiki namespace was nearly 24 hours ago. Browser is whatever the latest edition of Internet Explorer is (not Microsoft Edge), and using Monobook. Nyttend (talk) 10:56, 15 June 2016 (UTC)

Related: #Strange error (above). Fred Gandt · talk · contribs 11:00, 15 June 2016 (UTC)
I filed phab:T137877 about this issue, along with a guess as to what's causing this. Anomie 11:36, 15 June 2016 (UTC)
Someone fed a mogwai past midnight. Fred Gandt · talk · contribs 12:14, 15 June 2016 (UTC)
Or got it wet! KoshVorlon 15:50, 15 June 2016 (UTC)
Monobook? Wow... brings back memories. Intelligentsium 19:01, 15 June 2016 (UTC)

Time comparison

I'm trying to come up with a way of comparing times on Wikipedia. Specifically, I want to create a template that would display an error message if a certain page hasn't been updated in 36 hours or so, the logic of it being something like

if ({{CURRENTTIMESTAMP}} > {{REVISIONTIMESTAMP:Wikipedia:Article alerts/Last update}} + 36 hours)

or some such. I can't figure it out right now, so help would be appreciated. Headbomb {talk / contribs / physics / books} 15:03, 16 June 2016 (UTC)

@Headbomb: The expression
{{#expr:{{#time:YmdHis|-36 hours}} > {{REVISIONTIMESTAMP:Wikipedia:Article alerts/Last update}} }}
will return either 0 or 1. --Redrose64 (talk) 15:33, 16 June 2016 (UTC)
(edit conflict)
{{#ifexpr:{{#time:U}} > {{#time:U|{{REVISIONTIMESTAMP:Wikipedia:Article alerts/Last update}} + 36 hours}}|ERROR MESSAGE}}
Pages are cached so you may want a {{Purge}} for the user to trigger a current calculation of the passed time. PrimeHunter (talk) 15:34, 16 June 2016 (UTC)
Much appreciated! Headbomb {talk / contribs / physics / books} 15:56, 16 June 2016 (UTC)

AuthManager - is this the correct behaviour?

I've noticed a rather worrying behaviour with the new login system - I've not rigorously tested it, but I'd like to check at this point if it's expected.

  1. I log in on Machine 1, selecting the "Keep me logged in" checkbox.
  2. I log in on Machine 2, without selecting the checkbox.
  3. I close the browser on Machine 2.
  4. I open the browser on Machine 2 again. I am still logged in. (This is the worrying bit).
  5. I select "Log out" explicitly on Machine 2.
  6. I am now logged off on Machine 1 as well.

If this is the expected behaviour, I think that we should take some fairly urgent steps to notify people of the security implications of item 4 above. If it isn't, and it's not a problem at my end, we should take even more urgent steps to fix it. But I hope it is just a problem at my end. Tevildo (talk) 22:26, 13 June 2016 (UTC)

This is how it's worked for 7+ years. --Redrose64 (talk) 22:34, 13 June 2016 (UTC)
That's (I suppose) reassuring, but I've not noticed it before. I've always been logged out at Step 4 previously. Perhaps this time I just opened the new browser too quickly. Tevildo (talk) 22:42, 13 June 2016 (UTC)
The act of logging in sets a cookie on your machine; other login cookies that were set on other machines remain valid. Whenever you visit a Wikimedia page, your login cookie is sent back - if that is missing, or arrives corrupted, or is not recorded as a valid entry in the table held on the servers, you are treated as "not logged in"; you are only treated as "logged in" if the cookie arrives uncorrupted and also matches a valid entry in the table. The act of logging out invalidates all login cookies bearing your user name, whatever machine they were set on. Opening a browser normally leaves cookies alone; closing a browser may delete local cookies, it depends upon your browser's settings. --Redrose64 (talk) 23:12, 13 June 2016 (UTC)
Thanks for the clarification. I'll make sure that I explicitly log out when I'm not at my home machine from now on. Tevildo (talk) 23:20, 13 June 2016 (UTC)
No, it hasn't! If you do not check "Keep me logged in", closing the browser on machine 2 should end the session. This changed since the last deploy (see tech news), the session is no longer terminated. This is a rather large security risk! -- [[User:Edokter]] {{talk}} 10:46, 14 June 2016 (UTC)
I have reported this trough the appropriate channels. Normally I'd blank this section, but I think the cat is way out of the bag already, so leaving it for now. —TheDJ (talkcontribs) 11:33, 14 June 2016 (UTC)
This should be fixed for logins since about 15:35 UTC. If you logged in before that, you'll want to make sure to log out manually. As noted above, logging out anywhere will log you out everywhere (this is a feature of CentralAuth), so if you're concerned you can simply log out and back in from any machine. BJorsch (WMF) (talk) 17:04, 14 June 2016 (UTC)
Why would closing the browser on Machine 2 terminate your session, unless you delete cookies? Ideally, it shouldn't even be possible for a remote site to know you closed your browser. Though I suppose session cookies are generally interpreted that way, come to think of it. Wnt (talk) 12:11, 16 June 2016 (UTC)
@Wnt: Yes, because it's a session cookie. The remote site doesn't know that you closed your browser, but your browser knows to delete the cookie on exit. So, the next time the browser is opened and the site visited, there is no cookie to send with the page request, so it's as if you are logged out, even though you had not sent an explicit logout request. --Redrose64 (talk) 13:41, 16 June 2016 (UTC)
Both Chrome and Firefox apparently have (or had) behaviors where they don't delete session cookies, though I'm not sure about the specifics. [20] In any case the option to continue a session described there seems like a desirable user feature, since usually we want our browsers to tell other people only what we want them to say. Wnt (talk) 16:36, 16 June 2016 (UTC)

Accessdate in one ref transferred to the other


I may not get a chance to look into this soon, so I should post the bug: a version with two copies of a ref, the second with an accessdate, previewed and saved with both copies showing an accessdate. Taking out only the second copy got rid of both accessdates. As I recall that kind of nonlocal behavior worries developers... at the moment I haven't made even the most basic experiment to figure out what happened though. The versions are before and after this diff. Wnt (talk) 12:04, 16 June 2016 (UTC)

I see this after your edit:
Mauney, Matt (June 15, 2016). "2012 documentary shows Omar Mateen working security during BP oil spill". Orlando Sentinel. Retrieved June 15, 2016.
Are you saying you don't see "Retrieved June 15, 2016" at the end? Were you maybe thinking of another situation where the references have the same name but different content? PrimeHunter (talk) 12:24, 16 June 2016 (UTC)
OK, that's weird. I was positive the first Orlando Sentinel accessdate disappeared at the same time, but now I see it and there's nothing out of the ordinary. I looked at both previews and looked at both history versions before I posted this. Now I have no idea. Could there be a caching issue deep in the cite template? I doubt it. Maybe I was just having an unusually persistent hallucination... in any case, I'll withdraw this because there's nothing to look at now. Wnt (talk) 16:45, 16 June 2016 (UTC)

New formulas layout

Question Remark
Who and why have changed layout of formulas? Now I can't see formulas in my browser Opera 9.27 [21] . Yes, it's old but my computer has not enough resources for newer browser. Please explain what gain gives new layout and what was purpose.

Also W3 validator shows errors. I think , the simplier interface is the better for community. And blocking some browsers is equivalent to denying access to wikipedia for some people. It's much worse than any gains from optimisation, server load, etc.

Please, restore old layout.

By default SVG images with non-visible MathML are generated. The older PNG images can be set via user preferences.[b]

But I don't want to register and login (I haven't e-mail , time and desire to do it). (talk) 11:07, 11 June 2016 (UTC)

Just so you know, an email address is not required to make a Wikipedia account - you need nothing more than a username and password. As well as being able to edit your preferences, there are various other benefits. —  crh 23  (Talk) 13:21, 11 June 2016 (UTC)
If you can chooose a username and password quickly then it takes less than a minute to create an account. I don't know which Opera versions have problems with formulas but Opera 9 and older have less than 0.1% of our traffic. Opera is free and version 10 is from 2009. There comes a point where old browsers are not worth supporting 100% if it limits what a website can do or forces it to spend considerable time on browser support. PrimeHunter (talk) 14:28, 11 June 2016 (UTC)
There was a temporary bug in the PNG fall-back generation, which has since been fixed (see T136905). It is likely that Opera 9 is correctly loading PNG fall-back images now. If you could check this, it would be much appreciated. -- GWicke (talk) 19:47, 13 June 2016 (UTC)
No, still broken as far as I can tell. It looks like it is trying to load the SVG, but failing. If I open the image by itself in a tab, it displays fine. Pretty sure it is a problem with outdated handling of SVGs rather than a wikimedia problem. —  crh 23  (Talk) 19:59, 13 June 2016 (UTC)
I see, that's too bad. I had hoped that document.implementation.hasFeature('', '1.1') would return false in Opera 9, which would cause it to switch to the PNG fall-back. @Crh23:, do you have JavaScript enabled? If yes, could you test what this JS snippet returns? While I agree that Opera 9 is old, it might still be easy enough to support it well, if we can figure out why the PNG fall-back is not loaded. -- GWicke (talk) 01:20, 14 June 2016 (UTC)
I couldn't find a proper JS console, but putting javascript:{document.implementation.hasFeature('http//', '1.1')} in the address bar indeed returns true. I had quite an adventure trying to get a JS console to work, including executing code from —  crh 23  (Talk) 10:41, 14 June 2016 (UTC)
The SVGs load correctly if <img .../> is replaced with <object .../>, as suggested. —  crh 23  (Talk) 11:45, 14 June 2016 (UTC)
Pinging GWicke —  crh 23  (Talk) 12:25, 14 June 2016 (UTC)
A right, so we have the wrong feature detection, because SVG support is not the same as supporting SVG inline images... —TheDJ (talkcontribs) 15:18, 15 June 2016 (UTC)
I went ahead & created T137765 for this browser-specific issue, so that we don't forget to investigate this. -- GWicke (talk) 01:37, 14 June 2016 (UTC)
Question Quote
When I set option > Math > PNG images , I see images with javascripts enabled and with javascripts disabled (I prefer disabled) .

But also I see , web-developers have changed layout when radiobutton "MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools)" checked (or when user is logged out). Earlier it was meta -tag, but now img-tag. In both cases web-developer is wrong . If you use SVG you must use

<object data=""></object>

Just read standards and don't use undocumented features of trash-browsers.

No, still broken as far as I can tell.

Yes, you give me some patch through preferences, but you still have not explained the reason and purpose of the changes of layout. (talk) 05:03, 14 June 2016 (UTC)

I was gonna answer this seriously, but there is little use in talking to people who just want to be angry with other people and are not really interested in the actual challenges involved, and whose knowledge of website development is clearly stuck in the year 2008. You have been heard, I'm sure the rest of us will look at finding a solution. Thank you for your report. —TheDJ (talkcontribs) 15:18, 15 June 2016 (UTC)

Nothing was angry in my post and I didn't try to offend anybody. But if you have ability to make pages working in all browsers, why not apply it? Moreover site was working, but consciously converted to partially worked. For whom do you write html-code? If for yourself then OK, go on. But if for humankind, then stop. W3C recommends object tag [Secondly, it is worth pointing out that only <object> works consistently with the W3C standards, meaning at this time one must choose between standards consistency, or browser consistency.]. And please don't allege HTML5, as Opera 9 has not support of it. Also Opera 9.64 has many bugs, and is unstable (that's why there are many fans of 9.27 what may seem illogically. Actually no more Presto Opera was in such degree of stability, versions >9 were raw as Opera Software concentrated force on promosion for money instead of development, optimization, testing, etc. , Tetzchner left company because of this reason). Note that Opera 9.27 is Final for Merlin project . (talk) 20:10, 16 June 2016 (UTC)

Discussion at Template talk:This is a redirect#Accessibility issue

You are invited to join the discussion at Template talk:This is a redirect#Accessibility issue. nyuszika7h (talk) 20:49, 16 June 2016 (UTC)

One-Click Archiver sending threads to the wrong page's archive

I just noticed this from looking at my past edits. I have been One-Click Archiving some resolved or very stale threads on Wikipedia:WikiProject Resource Exchange/Resource Request, but they went to Wikipedia:WikiProject Resource Exchange/Archive 1 (the archive of Wikipedia:WikiProject Resource Exchange) instead of the Resource Requests page's archives, which would have currently been Wikipedia:WikiProject Resource Exchange/Resource Request/Archive 28. ClueBot III gets it right, so why is One-Click Archiver botching up? I've never had this problem before with One-Click Archiver. I don't know all that much about archiving coding so I can't figure it out and can't figure out how to fix it. Anybody? Softlavender (talk) 12:39, 16 June 2016 (UTC)

For the record, User:Softlavender/common.js imports User:Technical 13/Scripts/OneClickArchiver.js, documented at User:Technical 13/Scripts/OneClickArchiver which says: "This tool works, but due to Technical 13's indefinite ban, it is currently unmaintained. Other users are encouraged to take on the task of developing and maintaining it." The script has not been edited since Technical 13 was blocked a year ago. PrimeHunter (talk) 15:05, 16 June 2016 (UTC)
@Softlavender: If a page is archived automatically - for instance, by ClueBot III (talk · contribs) - there should not normally be a need to manually archive threads. OneClickArchiver counts as manual archiving: just leave the threads alone for the bot to pick up. About a month ago, Sunekit (talk · contribs) caused some serious mess by OneClickArchiving a number of threads at various Village Pumps, see Wikipedia:Village pump (technical)/Archive 146#OneClickArchiver misarchiving. --Redrose64 (talk) 15:25, 16 June 2016 (UTC)
One-Click Archiver should work no matter what, and there are many instances where it is more useful than the auto-archiving (WP:ANI is a prime example). I'm asking for help in how to remedy this situation. As Oiyarbepsy mentioned in the thread you linked, one issue is the archive template parameters on the page in question. Softlavender (talk) 23:46, 16 June 2016 (UTC)
  • One-click archiver does as it's told, and what tells it is User:MiszaBot/config. Cluebot uses a different config template. So, check the Miszabot template and make sure it matches the Cluebot template. Add a Miszabot template if it doesn't exist, otherwise one-click-archiver will just guess. Oiyarbepsy (talk) 01:15, 17 June 2016 (UTC)

Redr template giving incorrect advice

I noticed that at Miley cyrus, it says "Please do not replace these redirected links" just because it is semi-protected. I don't understand what's the reasoning behind that advice. Most people, including vandals, would reach the target rather than the redirect page, which would also be protected if necessary. Anyway, in this case, the links should be replaced with the correct form, because it is simply wrong. (I'm not sure why it exists anyway. Are we going to create redirects for every single person with their last name lowercased?) nyuszika7h (talk) 20:46, 16 June 2016 (UTC)

The text is from {{R semi-protected}} by Paine Ellsworth. I don't know why the quoted text is there. The redirect was made in 2006 where the go feature of the search box was case sensitive so without a redirect, a search on Miley cyrus or miley cyrus would have given a page of search results instead of going directly to Miley Cyrus. Many such redirects seem pointless today but redirects are cheap. PrimeHunter (talk) 21:25, 16 June 2016 (UTC)
To Nyuszika7h: Thank you for the ping, PrimeHunter – I think all the protection rcats are worded that way, so I'll look into changing them because there are many cases like the Myley cyrus redirect. The miscatorization rcat tells people the right thing to do; however, these cases can be misleading and confusing. Good catch, nyuszika7h!  What's in your palette? Paine  06:58, 17 June 2016 (UTC)

Posting of DYK on talk page and user page

This DYK article Template:Did you know nominations/Shumen Plateau Nature Park posted by me is presently in the lead on the main page. But it is not reported either on the talk page of the article or on my user page. There seems to be some technical hitch. Could you please look into it and see how it can be rectified and I get the article's endorsement on my talk page? Similarly, GA review notifications and a GA approval also do not appear on my talk page. Can these issue be resolved, please?--Nvvchar. 05:47, 17 June 2016 (UTC)

DYKUpdateBot missed two runs so {{Did you know}} was updated manually without the associated talk edits. The operator said "DYKUpdateBot is back in action!" four hours ago at User talk:Shubinator#DYKUpdateBot stopped. The next run would usually be around 12:00 (UTC). I don't know whether it will make the missed talk edits but wait and see for now. PrimeHunter (talk) 10:40, 17 June 2016 (UTC)l
@Shubinator: Can DYKUpdateBot do it retroactively or should it be manual? PrimeHunter (talk) 12:57, 17 June 2016 (UTC)
  • @Shubinator: Thanks. It is posted now on article talk page and my talk page.Nvvchar. 14:17, 17 June 2016 (UTC)
Thanks Maile66 for doing it manually: Special:Contributions/Maile66. PrimeHunter (talk) 14:27, 17 June 2016 (UTC)
@Nvvchar, PrimeHunter, and Legoktm: the Good Article nominations are monitored by a different bot. As mentioned on WT:GAN, for reasons not yet known, that bot has been running but skipping GA. Legoktm has not edited in two days, and he's the one to take care of this issue. — Maile (talk) 14:53, 17 June 2016 (UTC)

Magic word question

Is the a magic word that will return a value that equals the page which the magic word is placed rather than the name of the page where the magic word is transcluded? To better explain what I am trying to do as an example, let's say that I have pages "Foo A" and "Foo B": This magic word would be placed on "Foo A". "Foo A" would then be transcluded on "Foo B". The magic word should allow the value of "Foo A" appear on "Foo B" when it appears on "Foo B". (So, magic words like {{PAGENAME}}, {{FULLPAGENAME}} cannot do this since when viewing the transclusion of "Foo A" on "Foo B", they would return the related values of "Foo B" instead of "Foo A" in these scenarios.) Steel1943 (talk) 18:26, 17 June 2016 (UTC)

Why do you want to do that? Whatamidoing (WMF) (talk) 18:33, 17 June 2016 (UTC)
(edit conflict) I'm trying to make some edits to Template:Rfd2 that require this specific type of functionality. Last year, I made some edits to it to help with the creation of autopopulated edit notices in the event that a discussion is relisted, but in the process, had to hardcode instances of {{PAGENAME}} in the template as well as disable the related links when viewed directly from Wikipedia:Redirects for discussion since clicking on them from that page would not autopopulate the daily subpage's name into the edit notice due to the {{PAGENAME}} returning the value of the page which it is being viewed instead of the page which it is placed. (Previously, when the template was placed, the instances of {{PAGENAME}} were substituted when the Template:Rfd2 was substituted, but if that discussion ended up being relisted, then those links that were reliant on the substitutions of {{PAGENAME}} world no longer be correct, creating a broken link in the autopopulated edit notices.) And thus, the reason I'm asking if such a magic word exists. Steel1943 (talk) 18:43, 17 June 2016 (UTC)
Could hardcode an instance using {{#ifeq:{{PAGENAME}}|NAMEOFPAGE|actionX|actionY}}? --Izno (talk) 18:38, 17 June 2016 (UTC)
(edit conflict) @Izno: Unfortunately, that wouldn't work because that function would still be reliant on the page being viewed. See my explanation above. Steel1943 (talk) 18:46, 17 June 2016 (UTC)
How about the magic-word param? See first "Note:'" under the first table at Help:Magic words#Variables. I used the magic-word param in the {{Talk page of redirect}} template, and it works sort of like, well, magic.  What's in your palette? Paine  19:05, 17 June 2016 (UTC)
Paine, after reviewing that note, that idea unfortunately wouldn't seem to work either since that would still require either a substitution or a hardcoded instance of {{PAGENAME}} or one of its related words. When a discussion is relisted on WP:XFD forums that use daily subpages (such as WP:RFD), the subpage which the discussion appears changes when the discussion is relisted. (Which is why the template currently uses hardcoded instances of {{PAGENAME}}: to allow the value to automatically change when the discussion is relisted.) Steel1943 (talk) 19:15, 17 June 2016 (UTC)
Makes me wonder if there might be something about Fred's recently closed proposal that would help with this?  What's in your palette? Paine  21:52, 17 June 2016 (UTC)
Gnarly little challenge. Earlier I was thinking about an adaptation of Izno's idea above:
  • {{#ifeq:{{PAGENAME}}|{{safesubst:PAGENAME}}|actionX|actionY}}
Nah.  What's in your palette? Paine  22:01, 17 June 2016 (UTC)
There is no such magic word. It would for example be helpful in navboxes with V T E (view, talk, edit) links. They currently need a name parameter to tell them their own name so the links go to the right place. The parameter is frequently wrong so we have Wikipedia:Database reports/Invalid Navbar links. A template move requires updating of the parameter like [22]. PrimeHunter (talk) 18:45, 17 June 2016 (UTC)
That's what I thought. Is the addition of such a magic word even possible through the wiki software used here (like as a gerrit request or however that is done?) I'd have to imagine that such a question has been brought up in the past. Steel1943 (talk) 19:07, 17 June 2016 (UTC)
Anything can be requested at phab:.
Subst:PAGENAME isn't going to work when you want the page name to change if the RFD gets relisted (and therefore moved to a new date). I think what you really want is the advanced features for WP:Flow, which was originally envisioned as being a workflow tool that could handle everything at XFD from nominations and alerts through discussion and even reporting that it's time to close or re-list at the end, and make all of that work as efficient as possible for the admin. (Think Super-Twinkle for XFD. ;-) Unfortunately, that's not going to be available this year.
In the meantime, perhaps a bot or AWB-style script could handle this? Whatamidoing (WMF) (talk) 21:43, 17 June 2016 (UTC)

Signature timestamp does not agree with page history and watchlist

This edit by me shews in the page history and on my watchlist as having been made at 00:07, but the signature says 00:06. What is occurring? DuncanHill (talk) 00:20, 18 June 2016 (UTC)

There is often a 1 minute difference and sometimes more. I have seen 4 minutes.[23] It seems harmless so just ignore it. The time shown in logs (page histories, user contributions, watchlists, recent changes and so on) should always be the same. The time in a signature is registered at a different time, possibly by a different server. It is stored as wikitext and becomes an editable part of the page just like if the user had written it manually. It doesn't matter whether it's a little off. See also Wikipedia:Village pump (technical)/Archive 139#2 minutes difference between time in signature and history. PrimeHunter (talk) 00:56, 18 June 2016 (UTC)
Thanks - looks like I was quite active on VPT at the time but I have no memory of that thread! DuncanHill (talk) 01:02, 18 June 2016 (UTC)

"Edits by user" down

Attempt to use "Edits by user" at Shooting of Michael Brown produces "Found 0 edits by on Shooting of Michael Brown (0.0% of the total edits made to the page)". Search of VP archives tells me little but the fact that this is another thing that goes down a lot. Is there anything that can be done to bring it back up? Is there anything that can be done to prevent it from going down a lot? Why do working infrastructure tools break? Thank you. ―Mandruss  04:35, 18 June 2016 (UTC)

On second thought, I was expecting this to list the major contributors to the article, in descending sequence. Am I looking in the wrong place? Is "Edits by user" only for a single user? If so, that link could be clarified as "Edits by a user", perhaps. ―Mandruss  04:52, 18 June 2016 (UTC)

Found it in "Revision history statistics" Never mind. This could be removed as useless brain fart, or retained as education for other uninformed people like me. Over and out. ―Mandruss  04:56, 18 June 2016 (UTC)

Greater color contrast ratio

In regards to the use of {{Greater color contrast ratio}}, why does it not work when the first parameter is another template that returns a colour? It always picks the third parameter. For example (it IS meant to return white):

Alex|The|Whovian? 14:13, 18 June 2016 (UTC)

@AlexTheWhovian: {{NarniaColor|England}} is outputting <nowiki>#B30000</nowiki>, which is input to Lua as a strip marker. If you use mw.text.unstripNoWiki then Lua can convert the strip marker back to the original text of #B30000. — Mr. Stradivarius ♪ talk ♪ 14:47, 18 June 2016 (UTC)
And this edit seems to have fixed things. — Mr. Stradivarius ♪ talk ♪ 14:56, 18 June 2016 (UTC)
@Mr. Stradivarius: My thanks. Alex|The|Whovian? 15:11, 18 June 2016 (UTC)

Cascading semiprotection

WP:CASCADE says nothing about cascading semiprotection. Does anyone know where I could find a simple writeup that discusses why we don't have it? I remember that there was some sort of bug, whereby anyone could semiprotect a template by transcluding it onto the page with CC-SP enabled, but I'd like to see a page giving a better description than I could give. Nyttend (talk) 13:34, 18 June 2016 (UTC)

  • Security reasons, basically: By cascade semi protecting a page, every editor who can edit semi protected pages could semi protect other pages by simply transcluding them in such a cascade protected page. Your memory is fairly accurate, it's been referred to in Phabricator phab:T10658.Jo-Jo Eumerus (talk, contributions) 13:40, 18 June 2016 (UTC)
    • How does one search Phabricator? It runs in my mind that there was a much older bug request, in which someone had discovered the security flaw (when CC-SP was still an option) and filed an emergency request that got marked as "Unbreak Now!" and was fixed rather quickly because of the security issues. I've searched for such a thing, but I couldn't find it. Nyttend (talk) 15:38, 18 June 2016 (UTC)
      • Via the search box, but you have to check for "Closed" issues. I am guessing you refer to phab:T10796.Jo-Jo Eumerus (talk, contributions) 16:29, 18 June 2016 (UTC)
        • That's it; I used the search box, but I wasn't aware that I had to specify closed bugs. Final request — how do I find "bug 8658" that's mentioned in T10796? Apparently the bugs got renumbered when they switched from Bugzilla to Phabricator, because phab:T8658 is a problem with running update.php that has nothing to do with this. Nyttend (talk) 18:52, 18 June 2016 (UTC)
          • I believe you add 2000 to the number - is it phab:T10658 you're looking for? BethNaught (talk) 18:55, 18 June 2016 (UTC)
            • Yes, that's it, but I found it independently and was planning to report it here so you didn't have to look. I just added something to WP:CASCADE explaining why CC-SP isn't an option and linking to 10796, and in the process noticed that the policy still has some Bugzilla links, e.g. Bugzilla:8658. This auto-redirects to the relevant Phabricator ticket, which indeed is 10658. I'm surprised (the toolserver pages became 404 errors when they took it down, so I assumed that Bugzilla also wouldn't work), but it's nice. Nyttend (talk) 18:59, 18 June 2016 (UTC)

Move an article back

Nikon AF-S DX Zoom-Nikkor 55-300mm f/4.5-5.6G VR redirects to a very short version of the name. I think it should be moved back to the full name, but I think that a move cannot be done to an article that already exists, right? How can this be done? (And do you agree that it should be done?) Bubba73 You talkin' to me? 16:51, 18 June 2016 (UTC)

Administrators can move over a redirect with a page history by deleting it. Others can request it. See Wikipedia:Requested moves. PrimeHunter (talk) 18:22, 18 June 2016 (UTC)
Thanks, done. Bubba73 You talkin' to me? 20:35, 18 June 2016 (UTC)

Line-breaks in Special:Contributions search-options box

The "Search for contributions" box where one selects username, namespaces, various checkboxes, etc causes side-scrolling on small monitors. The underlying cause seems to be that each row of controls cannot line-break. It's good that there is no break between a checkbox and its label (&nbsp; is used), and reasonable not to break within a label (class mw-input-with-label sets white-space:nowrap). But there is also no possibility to break between a label and the next checkbox, which means that whole

□ Deleted only □ Only show edits that are latest revisions □ Only show edits that are page creations □ Hide minor edits

line can be forced to run wider than my window. Compare that to the analogous "Watchlist options" box in Special:Watchlist, where breaks can occur among the different "Hide:" options. There, we have a whitespace between each subsequent <span class="mw-input-with-label">...</span>. Where does one fix the Contributions layout? DMacks (talk) 21:04, 18 June 2016 (UTC)

The spans with class mw-input-with-label have nowrap set by way of CSS from the mediaiki.legacy.shared stylesheet. That file is part of the software. We could override it with with local CSS in Common.css, and file a bug to have it fixed. -- [[User:Edokter]] {{talk}} 21:16, 18 June 2016 (UTC)
Maybe I tangled the explanation here. That CSS class itself is fine. But the page that uses it (whatever generates Special:Contributions itself) should insert a space between the consecutive spans that use it. That is:
<span class="mw-input-with-label">...</span><span class="mw-input-with-label">...</span>
should insert a space ( ):
<span class="mw-input-with-label">...</span> <span class="mw-input-with-label">...</span>
DMacks (talk) 21:50, 18 June 2016 (UTC)
The page is generated in core (SpecialContributions.php). -- [[User:Edokter]] {{talk}} 22:19, 18 June 2016 (UTC)
Thanks! Beyond my understanding of the internals, so Phab'ed. DMacks (talk) 03:24, 19 June 2016 (UTC)

Question mark renders in read mode without being added in edit mode

In the read mode of my edit, the sentence "Cyprus is located just 75? km to the south of Turkey" has a question mark after the number "75". But if you look at what I actually wrote in the diff, I did not add a question mark after "75". In a subsequent edit, I removed the question mark but the actual diff shows no question mark being removed. How can this happen? Dr. K. 21:55, 18 June 2016 (UTC)

The character in question was a U+007F <control-007F>... No idea what it does, but may show as a question mark by some systems. -- [[User:Edokter]] {{talk}} 22:29, 18 June 2016 (UTC)
Thank you Edokter. I cannot see any character at all. How did you identify it and how did that character go into the edit? Dr. K. 22:37, 18 June 2016 (UTC)
7F is the good old "delete" character (bottom right corner of this chart). In the days of paper tape input, if a character was punched incorrectly, you'd punch a row of holes across the tape at that point and the character would be ignored. --Redrose64 (talk) 23:01, 18 June 2016 (UTC)
I see. Thank you very much Redrose64. That explains my questions. Dr. K. 23:15, 18 June 2016 (UTC)
I use the tool (unrelated to WMF or wikis) to get information about special characters. Copy-paste a string to the "Characters" field and click "View in Uniview". In this case MediaWiki changes the character so you would have to copy the string with 75 km from the diff and not the rendered page. Some browsers may also change certain special characters when copy-pasting but this case works for me in Firefox. PrimeHunter (talk) 00:04, 19 June 2016 (UTC)
I myself use the Unicode Analyzer extension for Chrome. -- [[User:Edokter]] {{talk}} 08:32, 19 June 2016 (UTC)

Route diagram templates in mobile view

Since… well, when the mobile view was implemented, the display of route diagram templates used in railway line articles and canal articles has been completely broken, probably because of table CSS. How could this be fixed (either through modifying the templates themselves, if possible, or MediaWiki:Common.css)? Jc86035 (talk • contribs) Use {{re|Jc86035}} to reply to me 16:11, 19 June 2016 (UTC)

In mobile view, tables are given a 1em top- and bottom margins. Setting a zero margin (inline) on the inner tables should clear any broken display. (And I strongly advice getting rid of the <center> tags.) -- [[User:Edokter]] {{talk}} 18:42, 19 June 2016 (UTC)
@Edokter: Thanks; I've fixed that in {{Routemap}} and several of the auxiliary templates which are used in it (but not yet {{BS-map}} or the legacy {{BS-table}} group). (The padding on the sides isn't perfect, but it's usable. I'm honestly quite surprised that no one's bothered to fix it since 2012.) However, the collapsible rows option (example; see diagrams 4.1, 4.2 and 6.2) still isn't working; is collapsibility entirely disabled in mobile view? Jc86035 (talk • contribs) Use {{re|Jc86035}} to reply to me 14:33, 20 June 2016 (UTC)
There also seem to be some remaining issues with the display of {{BSto}} inside infoboxes. Jc86035 (talk • contribs) Use {{re|Jc86035}} to reply to me 14:44, 20 June 2016 (UTC)
All I know is that tables in mobile view have different styling then desktop, and that it often clashes with templates using wikitables. Mobile view does not load Common.css/.js, so anything template related there, like collapsing, is usually not loaded in mobile (unless copied to Mobile.css). The collapsing functionality provided by MediaWiki itself (mw-collapsible) might work, but I have not tested. I am not that deep into mobile. -- [[User:Edokter]] {{talk}} 15:30, 20 June 2016 (UTC)

Tech News: 2016-25

19:14, 20 June 2016 (UTC)

Markup renders as

The technique at User:Iceblock/Markup-render may be useful for Template:Markup and probably other places too.

Try the user template by, for example, typing {{subst:user:iceblock/Markup-render|{{CURRENTYEAR}}}} inside table markup (brace and pipe, and pipe and brace). This will result in |- | {{CURRENTYEAR}} | 2019

Iceblock (talk) 21:30, 20 June 2016 (UTC)

Compact language list

There probably will be questions, so... If you don't like those compact language links, you can disable them at Preferences section "Languages" at the bottom of page. --Edgars2007 (talk/contribs) 15:37, 20 June 2016 (UTC)

The change was already reverted. --Stryn (talk) 17:36, 20 June 2016 (UTC)
Thanks for the note. --Edgars2007 (talk/contribs) 19:14, 20 June 2016 (UTC)
When Compact Language Links is deployed, the setting to disable/enable it will be under the heading "Languages" at the bottom of the Appearance tab (or equivalent for your language) in your preferences.
The deployment schedule is at phabricator:T136677. --Pipetricker (talk) 08:39, 21 June 2016 (UTC)

Can't set up vandal fighter

I can't set up last version of vandal fighter with mac os, no way to automatically detect my browser (chrome), with manual setting up of the browser can't find the right file. How to resolve this issue? Thanks, bye.--Giacomo Seics (talk) 14:43, 21 June 2016 (UTC)

@Giacomo Seics: Symbol move vote.svg Have you tried the Computing section of Wikipedia's Reference Desk? They specialize in answering knowledge questions there; this help desk is only for technical issues with Wikipedia. For your convenience, here is the link to post a question there: click here. I hope this helps. --Redrose64 (talk) 22:22, 21 June 2016 (UTC)

Preview-only template warnings using REVISIONID magic word

I've noticed several heavily used templates using the hack of checking if REVISIONID is empty to detect preview mode in order to show excess/misused template parameters, namely:

  • Module:Check for unknown parameters
  • Module:Convert
  • Module:Citation/CS1
  • Module:TemplatePar

A few templates also use Module:If_preview for the same reason.

These have the side effect of causing an extra parse on page edit and disabling the edit stash mechanism (pre-parsing the page as the user types the edit summary or otherwise looks like they might press save soon). A few suggestions were made for how to mitigate this. This affects a large percentage of pages and negatively impacts page save time performance.

The use of REVISIONID (and REVISIONTIMESTAMP likewise) should never really occur given the performance impact. It should probably be disabled server side at some point, since it's basically just a pitfall in the waiting.

I think CSS based hiding and/or HTML comments would work fine, leaving the HTML the same regardless of preview. I see that Module:Check_for_unknown_parameters already adds a tracking category to pages, which seems like it would work well when combined with comments/warnings that pinpoint what part of pages have parameter problems. Are there any experienced template editors interested in fixing these templates and modules? Aaron Schulz 19:35, 16 June 2016 (UTC)

See Wikipedia:Village pump (technical)/Archive 145#"Under Title" issue. --Edgars2007 (talk/contribs) 19:50, 16 June 2016 (UTC)
HTML comments should avoid the Google and internal search index pollution. Aaron Schulz 20:28, 16 June 2016 (UTC)
Aaron, could you please look at #Time comparison, if you have general concerns about using REVISIONTIMESTAMP in templates? Whatamidoing (WMF) (talk) 00:29, 17 June 2016 (UTC)
There is magic word and parser function for REVISIONTIMESTAMP (the later takes a title argument). The parser function is fine unless the page itself is used as the title. It's only when the revision id/time of a page itself is in that page's content that parsing becomes duplicated and impossible to cache. Aaron Schulz 01:08, 17 June 2016 (UTC)
For those wanting to know the background, convert recently (1 June 2016) switched to using REVISIONID so someone adding converts has a better chance of seeing an error. The following shows what "the statue is {{convert|12|m|junk}} high" looks like in preview and in a saved article.
the statue is 12 metres (Error in convert: Unit name "junk" is not known (help)) high
the statue is 12 metres ([convert: unknown unit]) high
It might be ok to put a message at the top of the page saying there is a problem providing there is an anchor at the place of the problem so the message could link to it. The anchor could be left in the page if it were saved with an error, and the anchor could be the same for each error so the link only goes to the first occurrence. Johnuniq (talk) 00:36, 17 June 2016 (UTC)
I would just output a big red error message every time. Now that Category:Convert errors is empty the error message should only appear in a few articles, and if anyone saves an edit with a big red error message it should get fixed pretty quickly. — Mr. Stradivarius ♪ talk ♪ 01:08, 17 June 2016 (UTC)

Pinging Trappist the monk because of the Citation/CS1 connection. — Mr. Stradivarius ♪ talk ♪ 01:15, 17 June 2016 (UTC)

Module:Citation/CS1 uses the preview detection as an editor-aid for the case where a link to Internet Archive is malformed. When one of those links is malformed, the module rewrites the link to point to a directory page at the archive so that the editor may choose an appropriate snapshot. Right now, every invoke of the module inpect the revision timestamp. I suspect that this could be changed so that revision timestamp is inspected only when there is an applicable error.
Clearly there is a desire for preview-only messaging. What happens if a module that needs to know about preview mode got the value of REVISIONID passed to it in the {{#invoke:Module|function|{{REVISIONID}}? Is this still 'bad'? MediaWiki knows that it is rendering a preview, can it not provide an 'environment' sort of table that modules could query for this and perhaps other useful information?
Trappist the monk (talk) 11:06, 17 June 2016 (UTC)
There is a patch proposing a magic word to expose preview mode as one experimental option at T137900, but momentum there seems to be toward exposing a method to register errors in Lua (I merged a patch for that, but that method does not yet give location context, which Jackmcbarn is looking into). Deferring the REVISIONID call until errors are discovered is probably the best short-term workaround. Aaron Schulz 20:46, 17 June 2016 (UTC)
Good, and Module:Citation/CS1 also uses hidden-text CSS to always have "invisible" warning messages displayed with custom CSS by wp:wikignomes or other users wanting to view the hidden CS1-cite warnings on live pages. -Wikid77 (talk) 23:10, 21 June 2016 (UTC)
I have implemented module:TemplatePar in several templates, especially for this helpful preview-warning. Adding a tracking category is not an alternative (it is a ´´parallel´´ error handling). I understand I can expect a software change or meta-module editing (or is this usage deprecated and should I edit my editor-level templates)? I`ll follow this. -DePiep (talk) 09:01, 17 June 2016 (UTC)

I am the author of Module:TemplatePar.

  • I don’t think there is a performance “problem”.
  • REVISIONID would be touched, if ever, only in case of errors.
    • Just don’t make errors.
    • If any error on template usage arrives, the target is to eliminate the mistake asap.
    • If no error, no cache issues are in question.
  • There are various methods to deal with detected faults.
    • See de:Wikipedia:Lua/Modul/TemplatePar/en #error handling.
    • On error detection
      1. a maintenance category may be thrown and/or
      2. an error message may be displayed in red
        • always visible
        • always but hidden
        • never
        • on preview only.
    • The last case is the only one that utilizes current REVISIONID, and only if there is a problem to be reported.

Greetings --PerfektesChaos (talk) 11:33, 17 June 2016 (UTC)

I can confirm the lightning-fast SAVE, after edit-preview with clean infoboxes having Module:TemplatePar to check for invalid parameter names. In fact, I was ultra-shocked at how fast (2 seconds) the edit-save operation of a large page occurred (after edit-preview), not aware then of the pre-processed templates and style cache which allowed the edit-save to instantly display the saved page. -Wikid77 (talk)

Pinging Frietjes for the unknown parameter module. – Jonesey95 (talk) 15:51, 17 June 2016 (UTC)

if there is an alternative way to show an error only in preview, then we can certainly put that in Module:Check for unknown parameters. we can also invert the if logic to put try to make sure the REVISIONID check happens only when there is output from the module. Frietjes (talk) 16:07, 17 June 2016 (UTC)
pinging Evad37 who is using Module:If preview in 650 or so articles. Frietjes (talk) 17:34, 17 June 2016 (UTC)
My use case was to provide some help/information links for editors at the bottom of an infobox, without them being visible for readers – I never realised using REVISIONID was a big deal. Is ~650 pages alright for the short term (until there's another magic word or lua method)? - Evad37 [talk] 02:10, 18 June 2016 (UTC)

Since it's probably not obvious to everyone, I want to point out Aaron's not just a random volunteer. Aaron and his four teammates are the people whose primary job duties involve deciding whether something constitutes a WP:PERFORMANCE problem. We don't need to worry about performance, but Aaron does. Whatamidoing (WMF) (talk) 18:31, 17 June 2016 (UTC)

It might have helped if he had posted with his WMF tag. :D --Izno (talk) 18:39, 17 June 2016 (UTC)
Special:CentralAuth/Aaron Schulz (WMF) seems a bit thin compared to Special:CentralAuth/Aaron Schulz. :D PrimeHunter (talk) 18:59, 17 June 2016 (UTC)

One thing someone noted above is calling the magic word only when an error is actually known to have happened. As long as there is also a tracking category in all such cases when the error is saved (so people can keep the number of affected pages very low), this could be a good short-term work around. Given how often the server logs are reported vary_revision disabling edit stashing (and also leading to double parsing on page views incidentally), there are either errors on a huge percent of articles or the magic word is used unconditionally for many modules. Aaron Schulz 19:44, 17 June 2016 (UTC)

It looks like Module:Citation/CS1 is unconditionally checking REVISIONID. The checks could be done via a method call and deferred until errors are already known like some of the other modules. Aaron Schulz 20:31, 17 June 2016 (UTC)

Module:Convert only reads REVISIONID if about to output an error message. It adds a tracking category if the page is an article (namespace 0), and the category is usually empty, or may have half a dozen pages for a few days. I guess that means there is no problem with convert continuing the way it is? Johnuniq (talk) 00:16, 18 June 2016 (UTC)

Are we sure that this is a problem? At Help talk:Citation Style 1#archive url checks and preview mode, Editor Aaron Schulz provided a link to a page of graphs, in particular saveTiming, that purport to illustrate performance (I guess) where the x-axis is readily identifiable as date/time, but the y-axis is just a meaningless (to me) number. The claim is that use of {{REVISIONID}} is detrimental to system performance. Module:Citation/CS1 started using that magic word at 17:42, 4 June 2016 (UTC) so any detrimental effect should be noticeable in the saveTiming graph if scaled to the most recent month, which see.

It is not clear to me that there was a substantial spike or level change in the graph as a result of adding {{REVISIONID}} detection to Module:Citation/CS1.

Trappist the monk (talk) 12:38, 18 June 2016 (UTC)

There are too many things going on in that time period. I'd defer the REVISIONID until errors are known and watch the graph after making that change. Aaron Schulz 18:14, 18 June 2016 (UTC)
To confirm, I have noticed, during the prior 2 weeks, how edit-save of many pages with clean wp:CS1 cites (after edit-preview) has always seemed slower, compared to pages with only infoboxes where the edit-preview evidently generated a cache to enable rapid edit-save (displayed within 2 seconds of click "Save page"), even for a large page with large infobox(es), an instant save. -Wikid77 (talk) 23:10, 21 June 2016 (UTC)

WP articles >1366pix width (31% of users have this size) don't show well, is there an easy way to find >1366pix WP pages?

This morning I reduced the headers on a couple of columns (to reduce the column-width) so the chart in an article would fit on a common laptop display. Only about 1/4 of all computer screens are <1366pix but 31% of users have 1366p screens. Is there an easy way to find other pages on WP that are >1366pix so they may be easily fixed to fit?

Are there any WP policies on page-width? Raquel Baranow (talk) 22:14, 20 June 2016 (UTC)

FI: The article mentioned above is Epidemiology of motor vehicle collisions.
I don't think this is something we need to be overly concerned with (within reason), as browsers are designed to allow easy navigation of varied sizes and shapes of page. Wikipedia's HTML markup is carefully parsed to ensure it's compatible with even ancient technology. An issue we cannot do anything about is how the user sets their browser's user-agent i.e. their text could always be much larger or smaller than normal, and we have no way to detect it or force them to use our size (for essential accessibility reasons). This means that no matter what efforts we make to have content be of a specific size, it simply won't be for everyone.
IMO: It's far more important that the content is semantically correct, than looking pretty. Although every effort should be made (within reason) to maintain a clean and aesthetically pleasing layout, if a table needs to be e.g. 4000px wide to make sense, then it should be. Fred Gandt · talk · contribs 14:36, 21 June 2016 (UTC)
One of the basic principles about web design is that you should assume nothing about the other person's system setup. They might not have the same hardware as you; they might not have the same screen dimensions; they might not have the same operating system or browser; they might not have the same fonts installed. If even one of these differs from your own, they will see tables differently. --Redrose64 (talk) 22:28, 21 June 2016 (UTC)
I wouldn't worry too much about this. Wide tables are notoriously problematic, but they do 'work'. I'd focus on putting less data in tables and more in prose, and hopefully at some point, we can improve the table viewing for smaller form factor devices, but those will always be problematic. (Try viewing a large spreadsheet on a phone). —TheDJ (talkcontribs) 09:00, 22 June 2016 (UTC)

Edit summary and <Enter>

I wrote an edit summary after editing an article and without thinking pressed the Enter key. To my horror, the edit was saved before I could vet it with Show preview. I can't recall hearing of any prior complaint, but it seems to me that it's bad policy to allow a stray finger to hit Enter, and one should be forced to make a deliberate and conscious save by moving the cursor and clicking Save page. Akld guy (talk) 02:37, 20 June 2016 (UTC)

Hitting Enter while the cursor is in the "edit summary" text box has always saved the edit outright, and it's been helpful to me (and many others I'm sure). Perhaps there could be a preference that could change the default behaviour from "save edit" to "show preview".  · Salvidrim! ·  03:32, 20 June 2016 (UTC)
This has been declined previously, it should be able to be made with a user javascript - not sure if someone has made one? — xaosflux Talk 03:48, 20 June 2016 (UTC)
I'm using User:Anomie/nosubmitsummary.js to disable the enter key. -- John of Reading (talk) 06:22, 20 June 2016 (UTC)
Thanks. That's a very very old Phabricator ticket. I guess it's pointless starting another, so I'll just go with the flow and take no special precautions to avoid hitting Enter; carelessness is sure to increase my edit count and who cares if there's extra server traffic and wear & tear on Wikipedia's storage media as a result :) Akld guy (talk) 23:15, 20 June 2016 (UTC)
Probably no one to be honest: don't worry about performance. —  crh 23  (Talk) 12:22, 21 June 2016 (UTC)
Does that .js listed above solve your needs? If enough people want it, it could be turned in to a gadget. I know some other projects default to require preview before save - personally I hate that feature when I have to go to a language I don't know and for some reason to contact someone. — xaosflux Talk 13:19, 21 June 2016 (UTC)
I've been using it for lots of months, and it's stopped me fubar-ing several times. I almost built it into other scripts I use, but am lazy ;-) Fred Gandt · talk · contribs 14:04, 21 June 2016 (UTC)
@Xaosflux: I'm not suggesting that the default behavior should be changed, merely that the Enter key function be removed. That would avoid inadvertent saves by careless fingers, and would force user to deliberately move the cursor to Show preview or Save page and click. His/her choice, as current. Akld guy (talk) 20:40, 21 June 2016 (UTC)
For enwiki, we have a few options: (1)Users can do this themselves - see the code snippet above, you can add this to your own .js file to try it out. (Let me know if you need help). (2) We could make this a gadget to turn on and off via preferences/gadgets; (3) We could change this site-wide (don't think we will get mass support for that). — xaosflux Talk 21:59, 21 June 2016 (UTC)
It is a common feature of browsers that the submit button of a form is triggered by the Enter key. This is also brings accessibility features (or rather, removing that would mess with the expectations of people who rely on keyboard navigation for accessibility reasons). —TheDJ (talkcontribs) 09:05, 22 June 2016 (UTC)

Talk pages in Google

Should wikipedia talk pages be showing up in google searches? For example a search for Ceropegia nampyana (without quotes) just now turned up a link to Wikipedia talk:WikiProject Plants. Sminthopsis84 (talk) 14:56, 21 June 2016 (UTC)

AFAIK this is expected behaviour: search engine indexing is controlled by the __NOINDEX__ magic word, and WP:Controlling search engine indexing says basically that everything except things in userspace or draftspace is indexed by default, with a few exceptions. —  crh 23  (Talk) 16:33, 21 June 2016 (UTC)
Yeah, they rank a lot lower than article pages, so that's why you might not have noticed one before, but they are indexed. I regularly use google to look for specific things in talk pages. —TheDJ (talkcontribs) 09:14, 22 June 2016 (UTC)

history revision

Softlavender (or whomever has a moment) the edit counter is having some problems apparently on wiki en, thank you--Ozzie10aaaa (talk) 11:07, 22 June 2016 (UTC)

New Tool to Help Transfer Files to Commons: MTC!

Hi all, I have written a new tool, MTC!, to help transfer free files to Commons. I'm looking for some victims beta-testers to help test and/or provide feedback about the tool. Please try it out, have fun, and let me know how it goes! :) -FASTILY 11:32, 22 June 2016 (UTC)

Notices on banned users' talk pages

I followed a rabbit trail and ended up at User talk:Technical 13, wanting to find a link to an arbitration page (presumably left by the user who blocked him most recently), but the history is clogged up with lots of deletion notices that obviously won't get a response from this indef-blocked user. Would it be possible for us to reconfigure the auto-notify scripts so that they don't leave notes on the talk pages of indef-blocked users? It seems rather pointless to give full protection to a user talk page just to ward off notices of "your page has been nominated for XFD", especially as it wouldn't impede such notifications when the nominator's an admin, and with most indef-blocked users (this one being an exception) there's no good reason to prevent them from making {{unblock}} requests at their talk pages. Nyttend (talk) 19:16, 18 June 2016 (UTC)

Sounds like a good idea. I would like to point out, though, that at least Twinkle, when used by an admin, will give a warning about editing a fully protected page and allow the user to skip that step. עוד מישהו Od Mishehu 03:12, 19 June 2016 (UTC)
Adding AFD notices to the {{bots}} template might help. AWB respects that; I don't know if Twinkle also does. WhatamIdoing (talk) 18:05, 22 June 2016 (UTC)

Search box no longer accepts pages titles?

Every time I enter a page, I end up on the search page(!). What gives? -- [[User:Edokter]] {{talk}} 10:32, 18 June 2016 (UTC)

Never mind, javascript was disabled. Still begs the question: why do I need javascript to navigate to a page directly? -- [[User:Edokter]] {{talk}} 10:36, 18 June 2016 (UTC)
Can you be a little more specific? With javascript disabled when loading a page are you finding the focus on the search input box? — xaosflux Talk 14:44, 18 June 2016 (UTC)
OK I duplicated this, however it is not consistent - with javascript disabled, but using monobook it seems to work (e.g. searching for "Heat" goes to Heat not the results) - but with Vector it goes to the results page. — xaosflux Talk 14:52, 18 June 2016 (UTC)
If memory serves, this was one of the old complaints about Vector: Monobook's default is "Go", and Vector's default is "Search". (Monobook's search box has two buttons under it.) Whatamidoing (WMF) (talk) 18:00, 22 June 2016 (UTC)
Annoyingly, at commons:, the default is "Search" even in MonoBook. You need to make a conscious effort to click the "Go" button, which isn't always easy if the drop-down suggestions are hiding the buttons. --Redrose64 (talk) 22:33, 22 June 2016 (UTC)

Twinkle and Urdu wp

User:Hindustanilanguage and I have been looking at an issue related to user notification on Urdu WP. It seems that Twinkle does not warn the page creator or create the AfD page on the Urdu WP. I am not verzed enough in either Twinkle or Urdu to give much help.

Any ideas?

All the best: Rich Farmbrough, 09:36, 23 June 2016 (UTC).

MW Error message when undoing "content model" change

I accidentally changed some dude's userpage "content model" from wikitext to JavaScript:

I tried undoing the edit (clicking undo, which tried to load, but received a MediaWiki error message:

  • [V2YW0ApAMFMAAJQqpJAAAABL] 2016-06-19 03:51:44: Fatal exception of type MWException

Using the built-in "rollback" function (not Twinkle's version, which I haven't tried) successfully reverted the change, though.  · Salvidrim! ·  03:55, 19 June 2016 (UTC)

For future reference: mw:Help:ChangeContentModel; you can always revert from the log page. -- [[User:Edokter]] {{talk}} 08:40, 19 June 2016 (UTC)
Here are 2 other ways to revert a content model change: you can either use Special:ChangeContentModel again to change the content model from JavaScript back to wikitext, or you can delete the page and then undelete all revisions except the content model change. (talk) 14:10, 19 June 2016 (UTC)
  • I know of the workarounds, but it remains that "rollback works, undo throws a fatal MW error" is not a desirable situation.  · Salvidrim! ·  17:58, 19 June 2016 (UTC)
  • This smells a lot like phab:T75490 - which apparently was closed as a wont fix? — xaosflux Talk 22:30, 19 June 2016 (UTC)
    • As Salvidrim said, "throws a fatal MW error" is not a desirable situation. T75490 seems to have been closed on the basis that it was about being unable to revert such edits at all and that ability was made available, not noticing that the normal undo link still throws exceptions (I also see that some exception-throwing was fixed as T73163). Anomie 15:22, 20 June 2016 (UTC)
  • I'll look into the exception, but how do you "accidentally" change the content model of a page? Legoktm (talk) 22:29, 22 June 2016 (UTC)
  • By being a klutz like me. I intended to test the "change content model" on my sandbox page to test out stuff but accidentally did it on some guy's userpage.  · Salvidrim! ·  17:02, 23 June 2016 (UTC)

Article count down

I was surprised when looking at my article count to see that it said I'd only created 383, when I've kept an accurate count and should have over 400. Doing a spot check, I found that articles I'd created from betwen 12 October 2008 and December 2008 no longer appear. For example this and this etc, no longer appear in my count. I assume that this doesn't only affect me. Any reason? Valenciano (talk) 15:00, 20 June 2016 (UTC)

@Valenciano: Looks like a problem with the tool, or the database copy it is working off of, not with enwiki. Your article creations contributions list does show the pages you referenced above. @Cyberpower678: or @MusikAnimal: do you have any info that may help? — xaosflux Talk 13:16, 21 June 2016 (UTC)
I heard a rumor that the DBs on labs are suffering from corruption, and that the problem is getting worse. I can't confirm that though at current.—cyberpowerChat:Online 13:57, 21 June 2016 (UTC)
Yeah it seems to be an issue with X!tools and the "missing" articles seem to date from late 2008. Valenciano (talk) 13:59, 21 June 2016 (UTC)
See T134203 for more details, indeed there is some data drift in the database - which xtools works off of. ~ Matthewrbowker Drop me a note 20:43, 23 June 2016 (UTC)

Google actively blocking an innocuous Wikipedia article

Google Search results are actively blocking the Wikipedia article Tommy's Honour, even though it comes up very easily (top four of the results) on Bing [29] and Yahoo [30] searches. The "Tommy's Honour" Wikipedia article does not even show up on a site-specific search (even though every Wikipedia article or category the title is mentioned comes up): [31].

Even if I search for a specific phrase which has been in the live article from the beginning, it doesn't come up on Google: [32], although again it does come up on Bing [33] and Yahoo [34].

Why is this? The article is more than a week old, and I have checked every day and it has been blocked on Google from the beginning, even though every other article on Wikipedia comes up on a Google search within minutes of its creation.

Can anyone figure this out? As far as I can tell there are no codes or magic words in the article that would cause this, but maybe someone with tech savvy can check further. As far as I can tell, I have been the only one editing the article or its talk page except for a category added.

Thanks for any help. Softlavender (talk) 08:08, 22 June 2016 (UTC)

Sometimes it takes a while before Google picks up on a new article. A week is not common, but not impossible either. —TheDJ (talkcontribs) 09:12, 22 June 2016 (UTC)
In my experience that might have been true 7 to 10 years ago, but nowadays Google picks up Wikipedia changes and articles instantly. In my opinion there's no reason why Bing and Yahoo would index the article but Google would actively skip over it except something is blocking it or there is some error involved. Softlavender (talk) 09:45, 22 June 2016 (UTC)
It looks to me like Google is currently unable to read pages with an apostrophe in the title. See phab:T106793: "Pages with single quote in title are inaccessible by some clients (redirect loop)". I tried 10 articles with an apostrophe created two days ago:
None of them were found by Google. I also tried 10 pages with no apostrophe and all except ‎Ruggles Beach, Ohio were found. In addition, if a Google search like'Reilly+site: finds older articles with an apostrophe then none of the cases I tested have a "cached" link on the green triangle after the url. Other articles have that. PrimeHunter (talk) 10:42, 22 June 2016 (UTC)
Yeah, what PH said; punctuation in titles confuses Google's crawlers. I've just created Tommys Honour as a redirect, which Google ought to pick up within half an hour or so, although its pagerank is likely to be fairly low. ‑ Iridescent 10:49, 22 June 2016 (UTC)
Yup, tommy's honour now working as a Google search term. The same "create a redirect with no punctuation" trick will work on any page Google's ignoring. ‑ Iridescent 10:51, 22 June 2016 (UTC)
Iridescent, Google never blocked mention of the article within Wikipedia; it's blocking the article itself. Please click all the links in my OP. Softlavender (talk) 10:54, 22 June 2016 (UTC)
Yes, which per my comment above I've now fixed for you. Google's crawlers don't index pages with apostrophes in the title, so if you want a page to show on Google you need to create an unpunctuated version of the title as a redirect, which I've now done. Tommy's Honour is now showing the correct article as the first hit. (Searching just "Tommy's Honour" without the site.wikipedia is currently showing the Wikipedia page on p4 of the results, but that's because the redirect has only just been created; it will rise through the pageranks quite quickly as people visit it.) ‑ Iridescent 10:58, 22 June 2016 (UTC)
OK looking a tiny bit better now. Thanks very much for your help. Softlavender (talk) 11:20, 22 June 2016 (UTC)
  • I have created a thread on this issue on Google Help Forums: [35]. Evidently Google has known about the issue for at least 8 months. -- Softlavender (talk) 13:44, 22 June 2016 (UTC)

Hi :Softlavender, The article also does not display Wikipedia's Page view statistics (all days are 0):'s_Honour

I have posted about the wikipedia problem at: Wikipedia:Help_desk#wmflabs_Pageviews_Analysis_broken.3F. Ottawahitech (talk) 20:51, 22 June 2016 (UTC)please ping me

Ottawahitech, I've been checking page-views daily on that article and it has always worked. However today for some reason the page-views don't show up unless you actively choose a specific date-range, at which point the page-views do show up. I'm not sure if other articles are having this issue. Softlavender (talk) 05:38, 23 June 2016 (UTC)
  • Iridescent, the article is back to being fully blocked on Google, even with a site-specific search or phrase-within-article search. Softlavender (talk) 05:44, 23 June 2016 (UTC)

Before we start a massive redirect creation effort, I'd like to do a bit more research.

For example, I just did a Google search for Connecticut Huskies women:

and the first page of results includes:

How does this square with the claim that titles with apostrophes are not indexed?--S Philbrick(Talk) 18:48, 23 June 2016 (UTC)

That said, I do not see:
or others in the first ten pages of results.--S Philbrick(Talk) 18:53, 23 June 2016 (UTC)
Sphilbrick: (1) The redirect solution doesn't work, as I informed Iridescent above. It worked for a few hours but Google is again completely blocking the article he made a redirect for. (2) The Google issue has only existed recently. The articles that are blocked are articles that were created after the onset of the problem. The articles that you are seeing that do appear on Google were created before the problem started. All of the articles you are looking for are indexed by Google and will come up if you use quotation marks in the search. Try looking for more recently created articles with apostrophes, and use quotation marks when you search (that's the only way to determine whether Google is indexing it; without quotation marks you are going to get a random assortment of results). Softlavender (talk) 23:27, 23 June 2016 (UTC); edited 23:52, 23 June 2016 (UTC)

Disenrolling from course

I have enrolled in Education Program:Hanyang University/Sociology of Everyday Life (Spring 2016) by mistake but when I try to disenroll, a blank page is shown. Another user will have to remove me from that course. GeoffreyT2000 (talk) 03:06, 24 June 2016 (UTC)

 Done You have been removed. — xaosflux Talk 03:39, 24 June 2016 (UTC)

Special:Notifications page failure

The Special:Notifications page no longer displays notifications for me. I'm guessing there was a software update about two weeks ago and my system (Mac OS 10.4.11 Safari 4.0) does not support the upgrade. Is there a way to access the previous version of the software so that I won't have to rely on email notifications? Reidgreg (talk) 14:51, 22 June 2016 (UTC)

@Reidgreg: No, that is not possible. Could you try to be more exact as to what is not working ? Is the entire page just blank ? Anyways, you are already receiving a fallback interface (without any javascript enhancements), and apparently even the fallback interface is no longer working for you. Such a browser is HIGHLY HIGHLY deprecated and I would not expect it to work without major problems on most websites. MediaWiki only guarantees the most basic support for reading articles for such browsers, and they can (and are allowed to) break at any time. If you have a friend who has knowledge about HTML and CSS, he might be able to figure out why it would fail, and then we might be able to workaround it. —TheDJ (talkcontribs) 16:13, 22 June 2016 (UTC)
I filed a ticket, but it will probably be difficult to find the cause and not of the highest priority. —TheDJ (talkcontribs) 16:23, 22 June 2016 (UTC)
The Special:Notifications page is displayed as a regular Wikipedia page with the top and left bars, the title is Notifications under which is a link to Preferences and under that is a short, wide rectangle with scrolling diagonal bars as if something is supposed to be there. I receive no error messages. Thanks for filing a ticket (I'd attempted to investigate this on Phabricator but couldn't get those pages to load, either). I know I'm in a small minority, but I always appreciated Wikipedia's accessibility and can't say I favour upgrades which disenfranchise even a small portion of the userbase. Reidgreg (talk) 17:07, 23 June 2016 (UTC)
@Reidgreg: To confirm, you're seeing something like this? That is the loading indicator, and if you're seeing it, that would mean that your browser is actually getting JavaScript, and fails in the middle of executing it. Can you try disabling JavaScript support in your browser and seeing if you can load the page then? We apparently still serve JavaScript to Safari 4 (and only blacklist Safari 3), maybe we should blacklist it too… Matma Rex talk 07:39, 24 June 2016 (UTC)
I don't believe I'm normally given to superlatives, but OMG it worked! Thank-you so much, I bow to your awesome technomagery! I'll just disable javascript to load that as a normal page when I get a notice of an alert waiting. (I wonder what changed that made it stop working -- I'd guess it had been trying the javascript but failed after a certain time and defaulted to the regular page, but for some reason it stopped defaulting?) Again, much thanks all around! - Reidgreg (talk) 10:12, 24 June 2016 (UTC)
At this point we don't really know. I guess either we accidentally started serving JavaScript code to Safari 4 when previously we did not, or some recent change to Notifications started requiring some feature of JavaScript that Safari 4 doesn't support. Hopefully we'll find out soon and this workaround will no longer be necessary :) Please watch the Phabricator task linked above for updates. Matma Rex talk 11:16, 24 June 2016 (UTC)

Gallery tags: Make mode=packed the default

I think there are very few galleries still using the old grid-style, e.g. this one (content taken from Wikipedia:Featured pictures/Fungi):

We have a better solution, but it's not the default: adding mode=packed gives us:

When this was launched, it was suggested the default could be changed to mode=packed in future. I think it's time. Any disagreement? Adam Cuerden (talk) 11:42, 23 June 2016 (UTC)

I happen to agree, but this needs a {{RFC}