Wikipedia:Village pump (technical)

 Policy Technical Proposals Idea lab WMF Miscellaneous 

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

The technical section of the village pump is used to discuss technical issues about Wikipedia. Bug reports and feature requests should be made in Phabricator (see how to report a bug). Bugs with security implications should be reported differently (see how to report security bugs).

Newcomers to the technical village pump are encouraged to read these guidelines prior to posting here. If you want to report a JavaScript error, please follow this guideline. Questions about MediaWiki in general should be posted at the MediaWiki support desk.

« Archives, 193, 194, 195, 196, 197, 198, 199, 200, 201, 202, 203, 204, 205, 206, 207, 208, 209, 210, 211, 212, 213

IP range blocks

Tracked in Phabricator
Task T243974

User:Vituzzu has blocked multiple ranges, and has not been responding to multiple requests at meta:User_talk:Vituzzu for unblocking, despite reminders at itwiki where he is most active.

These blocks are affecting many editors who use VPNs and cloud services. Some of the IP range blocks include English Wikipedia, while other ranges are only blocked on e.g. Commons, Wikidata & other language Wikipedias.

How can we get someone else to review these blocks, please? – Fayenatic London 14:29, 24 January 2020 (UTC)

Only stewards from metawiki can review them. Asking here is useless. Ruslik_Zero 19:10, 24 January 2020 (UTC)

Note that one can override global blocks locally, at Special:GlobalBlockWhitelist. Jo-Jo Eumerus (talk) 19:26, 24 January 2020 (UTC)

Thank you both. I found meta:Steward_requests/Global which is the place to appeal.

Can the blocking message be changed to link to that page, instead of saying "You can contact Vituzzu to discuss the block" since that has proved fruitless and frustrating for many would-be editors? – Fayenatic London 13:03, 27 January 2020 (UTC)

IP's can't place requests there, Globally blocked or locked users should appeal to stewards(at)wikimedia.org. — xaosflux ^Talk 14:49, 27 January 2020 (UTC)

@Xaosflux: Thanks; I confirm that that email address is displayed in a different message to blocked IP users. The problem remains for logged-in users who are blocked; we see the message that I pasted at meta:User_talk:Vituzzu#Unblock_Puffin_browser, which only refers us to one steward (who may, as in this case, choose not to respond to talk page requests). The message displayed needs to be changed. – Fayenatic London 21:01, 28 January 2020 (UTC)

@Fayenatic london: can you point to the specific MediaWiki: messages that are being shown? — xaosflux ^Talk 21:21, 28 January 2020 (UTC)

LIkely one of these. — xaosflux ^Talk 21:23, 28 January 2020 (UTC)

It was not one of those, but it is now! I checked many on that page, then tried Search, which did not find it; so I tried a Google search, which found a mirror of it, via which I traced this copy: MediaWiki:Globalblocking-ipblocked-range/en. I edited that, but I assume that the original version is the locked one at meta:MediaWiki:Globalblocking-ipblocked-range/en. – Fayenatic London 22:23, 28 January 2020 (UTC)

@Fayenatic london: ok - so your concern is that you don't like the message on meta-wiki, not here. Here is the challenge, meta-wiki is multi-lingual so we don't like to translate those messages directly. I think a good idea may be to request that that message has an additional line, something like {IF X exists: "For additional information see X"} where X is a new message that can point to a page. To request a change like this you can ask at phab by filing a feature request. — xaosflux ^Talk 23:19, 28 January 2020 (UTC)

This complaint has now brought to light a wider problem. Thank you MarcoAurelio for diagnosing this and for the fix in process, and Xaosflux for your help.

For the record, the link above for Steward requests is not the appropriate place to appeal against these blocks, but meta:SRIPBE. – Fayenatic London 12:26, 1 February 2020 (UTC)

Note, SRIPBE is not the place to "appeal" a global IP block, it is the place to request an exemption around all global IP blocks. — xaosflux ^Talk 12:57, 1 February 2020 (UTC)

For the record, the blocking message is being changed, see meta:MediaWiki talk:Globalblocking-ipblocked-range/en. It will no longer refer people to individual stewards (who may be unresponsive). As for the frustrated editors unanswered at Vituzzu's page, I pinged those from the last 6 months myself (as WP:TPS), advising them to make a request on a generic page; some have already done so. – Fayenatic London 13:26, 7 February 2020 (UTC)

Post 6 million article reflections: What does Wikipedia do "wrong" technically?

Having just passed the 6 million article mark, it's clear that Wikipedia is going to be around for a long time. Maybe it's time for some introspection. What do you think are major technical decisions in the design of Mediawiki, that were mistakes in hindsight? I will mention the two that I think were particular bad:

• Wikitext itself is not defined by a grammar. As for the markup itself, the use of the same character to markup bold and italics, in particular, seems like a mistake.
• Talk pages and the way we use them are a disaster.

With the wikitext, I don't know what the current status is but there have been efforts to define Wikitext. Those should be high-priority efforts by the WMF. As our article count grows, so too does the need for automated parsing. This is an issue where the technical debt will get worse with time and make the encyclopedia harder to maintain. (If my understanding about the current status is out-of-date, please point me to updates.)

As for threads, some sort of threaded implementation is needed. mw:Structured Discussions is our attempt but so far the first implementation, Flow, has been unacceptable. Something more compact and minimal in style like reddit's comment system would have been leaps and bounds better. One of the major complaints I have is that the entire concept of "archiving" or closing discussions, where we tell users to not to add to an older discussion, is a big mistake. One should ALWAYS feel free to add to a discussion even if you are years late to the action. Why? Because people still read those discussions years late too. It doesn't make sense to leave questions unanswered just because nobody arrived in time to answer them or not to post a winning idea because you as a late arrival were the first to have it. Even when the user is still motivated enough to start a new discussion (and not all will be), it often doesn't make sense to fragment the discussion this way because it often leads to pragmatic issues because the new thread doesn't link to the original discussion. Plus too, the rate of discussion is dramatically different depending on the article so it's hard to even define what "late" to a discussion even means. It is normal to wait years for a question to be answered on a very obscure or highly technical article.

I've always felt like our subpage system could be modified to handle talk discussions. (I've been under the impression for a long time that this was one of our Wikipedia:Perennial proposals but maybe not.) Picture this: a new threaded discussion could be started as a subpage of the talk page, replies as sub-subpages to that subpage, replies to replies as sub-sub-pages, and so on. The talk page itself would be generated by Mediawiki to show threads (subpages) with most recent activity. I think this idea could work brilliantly. 1) Threaded discussion (as a tree) is enforced by the storage of the data itself. 2) It would discourage the accidental editing of other's comments while still allowing for it when needed (with isolation of the history of modification to just that particular comment). Subpages NOT intended to be discussions could be marked so they are excluded from auto-generation of the talk page. I think this has HUGE potential for revolutionizing the way we handle discussions. It would also eliminate the need for our clumsy archiving process.

Anyway, those are the top two problems that I'd like to see be fixed so future generations of editors aren't saddled by poor decisions forever. Fixing each of these would require some work but I think the WMF and the community could tackle these. What do you think? What about your own pet peeves? Jason Quinn (talk) 04:36, 28 January 2020 (UTC)

Without comment on the general concept of modifying talk pages, you have pretty much described Flow, which worked on the tree discussion model and each thread as a subpage. Risker (talk) 06:08, 28 January 2020 (UTC)

We did just hold an immensely detailed cross-project talk page consultation on this issue, which seems to have given preference for an improved version of the current setup, designed to be more user-friendly, rather than any whole-hearted shift. Nosebagbear (talk) 09:50, 28 January 2020 (UTC)

The reason I heard why Flow was never adopted (besides being tightly integrated with VE, which was received with a marked lack of enthusiasm) was that no one could figure out how to port all of the old discussions into Flow. IMHO, & speaking from my experience with computers, is that the old discussions should have been left alone & everyone used Flow (or whatever) from a given point on. As long as it worked, after the first few weeks of bitching from one & all, it would have been accepted. (Or maybe the response to VE dissuaded those who made the decisions from implementing Flow, because without the first the second wouldn't work.) -- llywrch (talk) 00:34, 29 January 2020 (UTC)

Nope, it just had so many many issues and many people had "showstopper" type issues (and there were many of these) - one of mine is infinite scrolling - YUCK. — xaosflux ^Talk 00:46, 29 January 2020 (UTC)

Second the disdain for infinite scrolling. Also second that Flow had more issues than just how to import old discussions. As I said above, the style was far too "fluffy" and not compact enough. As for old discussions, I think my idea of using subpages easily handles this: old discussions are simply moved to a special subpage called "old". Done and done. Jason Quinn (talk) 01:45, 29 January 2020 (UTC)

@Jason Quinn: Also, the parsing of Wikitext has been made much more rigorous with the creation of Parsoid, which is now the definitive parser for Wikitext, and therefore also acts as the de-facto specification for that notation. It has a much cleaner internal architecture than the previous software, and in particular, the low-level tokenizer is now defined rigorously in terms of a Parsing Expression Grammar, making things like the apostrophe ambiguity at least well-defined, if not sensible. -- The Anome (talk) 12:07, 28 January 2020 (UTC)

@Jason Quinn, @The Anome 'pings' notify when the addressee, the text, and the signer are edited in one, unitary message. --Ancheta Wis   (talk | contribs) 12:25, 28 January 2020 (UTC)

Alright, add that to the list: forgetting a ping is so user-unfriendly that people mess it up in a discussion about wikitext parsing. (Edit summary pings have helped in this area though.) --AntiCompositeNumber (talk) 12:41, 28 January 2020 (UTC)

Absolutely. If "belated ping" happens this often, it's no longer a user mistake, it's a use case. Just check if the editing user account name matches the account name in the signature (even if it's already there on the page), and if so, trigger the ping event. -- The Anome (talk) 18:12, 28 January 2020 (UTC)

The Anome, A {{#ping:The Anome}} parser function might be a better idea, to explicitly say "Yes, I want to ping this person". If it gets evaluated in the expanded wikitext, send a notification. --AntiCompositeNumber (talk) 18:36, 28 January 2020 (UTC)

The main reason to require a signed edit is to avoid unwanted pings when you refactor or archive a talk page or copy existing posts, and the diff engine detects a userpage link in your edit. A parser function wouldn't change that problem unless it saved something else (or nothing) to indicate the ping is old and should not be repeated if the saved code is copied later. PrimeHunter (talk) 22:37, 28 January 2020 (UTC)

• Our Main page is in need of some technical upgrades, especially so we can take back control of how it appears to mobile users. This has been challenging to deal with in the past, as efforts to improve technical components often creep in to "content" management - even when they don't have to. — xaosflux ^Talk 12:17, 28 January 2020 (UTC)

  @Xaosflux: I tried to nudge the current TemplateStyles proposal implementer but it didn't pan out. Want to make a go at it with me and see how many millions of people we can scare with our horrid web design? :) --Izno (talk) 01:48, 29 January 2020 (UTC)

• I don't care too much about the talkpages and wikitext - yes, it could be improved but both do properly serve their purpose and do not really pose a problem. I am much more worried by the total neglect of bugs and maintenance on some very old extensions. My pet peeve is there the spam-blacklist extension, but also our AbuseFilter and CheckUser and similar are in a very bad shape (the latter now does get some attention). There are bug reports of more than 10 year old that nonetheless are still utterly ignored. --Dirk Beetstra ^{T C} 12:27, 28 January 2020 (UTC)
  • Yes, considering the scale of the WMF's budget, it would be good to see some more of it spent on software maintenance, rather than just developing new features. -- The Anome (talk) 12:34, 28 January 2020 (UTC)

  +1 to that. It seems that once a WMF team declares a project "good enough for production", all development effort stops. I have noticed that WMF development teams have been starting to be more explicit about their development priorities (or lack thereof). This is a good start, but doesn't actually help anything get fixed. --AntiCompositeNumber (talk) 12:47, 28 January 2020 (UTC)

  +2. I would much rather see a massive bug-fixing effort than yet another new approach that is then essentially abandoned at the late beta / 1.0 stage with hundreds of unpleasant bugs outstanding. – Jonesey95 (talk) 15:16, 28 January 2020 (UTC)

  +3. MER-C 15:32, 30 January 2020 (UTC)

  If we're in the business of spit-balling crazy ideas, I personally don't care for visual editor, and I don't understand edit filters, but it would be super helpful to have a visual editor for edit filters. GMG^talk 15:19, 28 January 2020 (UTC)

  • Yes, and this applies to other tools too. For example, I have a five year old request for a simple feature addition to navigation popups, to make them show Wikidata IDs. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:53, 28 January 2020 (UTC)
• The biggest thing in my mind is how we edit articles. Visual Editor is a great step forward, but it's not done yet. Improving VE, and especially citation handling, should be one of the highest priorities. We need to get it to the point where it's the default editor. -- RoySmith (talk) 17:16, 28 January 2020 (UTC)
  • RoySmith, before we could do that, we'd have to have a long discussion about what it means for any piece of software to be "the default editor". There are many Wikipedias at which a majority of editors use the visual editor (i.e., the visual mode, showing rich text, not mw:Extension:VisualEditor's wikitext mode, which we usually call the 2017 wikitext editor) for at least some edits. There are many editors who use it exclusively. Some editors care about what's popular, rather than what's good for a particular purpose. One of the problems we've had in deciding which is more popular is that we don't tag all the editing environments that are used on wiki. Looking at the 500 most recent non-bot changes to articles here, I can tell you that 7 happened in Twinkle, 8 used STiki, 5 used HotCat, 1 used AutoEd, and that at least 40 used AWB ("at least", because only some AWB users tag their edits with that tool). 22 used the visual editor on desktop, 12 used the visual editor on mobile, and 10 the 2017WTE, which is part of VisualEditor. 57 used the wikitext mobile web editor. But how many of the other ~350 edits used the 2010 wikitext editor? Or WikEd? Or the 2003 editor? Or another editing environment? Or didn't really use any editing environment? Nobody really knows. I've heard estimates that around half of edits on this wiki are bots or use scripts (like Twinkle), but we don't know whether, say, WikEd is used for more edits than the 2003 wikitext editor. Whatamidoing (WMF) (talk) 19:58, 28 January 2020 (UTC)
    • I'm talking about what a brand-new desktop user sees for the first time. Not bots. Not power-users with AWB, Twinkle, etc. Open an incognito window in your browser. Go to Special:Random. Click the edit tab. Right now, you get taken to a "Welcome to Wikipedia" screen, with two buttons: "Switch to the visual editor", and "Start editing". The very first thing a new user is confronted with is the requirement to make a choice that they can't possibly understand the implications of. That's a problem. They should just be immediately dropped into the Visual Editor, but along with that, VE also has to improve. And, yeah, it has to work on talk pages too. -- RoySmith (talk) 20:33, 28 January 2020 (UTC)
      • We could probably get the devs to make that change (newbies at enwiki start in the visual mode, just like they do at many other wikis), but it'd probably take a local RFC, because of the history of individuals making contradictory pronouncements about what The Community™ wants. As for using it on talk pages, see #Eventualities. Whatamidoing (WMF) (talk) 20:53, 28 January 2020 (UTC)
  • @RoySmith: What improvements to citation handling do you see needed in VE? Just curious :) Sam Walton (talk) 23:40, 29 January 2020 (UTC)
    • I don't know how much of what I'm going to say here is VE per-se, how much is Citoid, and how much is just the design of the {{cite}} templates. But, it's all part of the overall UX, which is really all that matters.
    1. It too often can't find the metadata it needs in automatic mode. That's not an easy one to solve, since it's usually a case of the website having sucky markup. But, maybe we could log every URL it fails on, keep track of what domains it fails on the most often, and use that as feedback to build smarter (perhaps domain-specific) parsers.
    2. Related to the above, it too often mis-categorizes things as web, when they really should be book, or magazine, or whatever. And when it does that, it's not easy to fix.
    3. It can be hard to find the field you want. Some of the templates have way crazy number of fields (last2, last3, last4, etc) and you have to scroll through all of those to find what you want.
    4. It should insert the reference in a human-friendly form, i.e. with one field per line. That way, if you need to go in and fix something in source editing mode, it's possible to find what you need to fix.
    5. References get renumbered weirdly in preview mode under certain circumstances (T207182 and T52474)
    6. It should be smarter about integrating edit filters. Try citing this book, for example. You don't get the warning until you actually publish the page. And then you get a dialog box with inscrutable "Dismiss" and "Continue" buttons (I have no idea which button does what). I tried clicking "Dismiss", then I realized I had lost the error message, so I figured I'd just click "Publish" again to see it. To my surprise, the second publish actually published it. Anyway, the warning should come up immediately as soon as you try to insert the reference. Yeah, I know, that's a messy fix because the filter architecture doesn't support that, but it's all part of a bad UX.
  • Don't get me wrong, what we've got now is pretty cool. It just can be a whole lot better/easier, and it needs to get there to make this really usable for newbies. -- RoySmith (talk) 01:03, 30 January 2020 (UTC)
• Speaking of the 2019 Talk Page Consultation, would you all please go to https://en.wikipedia.beta.wmflabs.org/wiki/Talk:Dog and log in (unless you want your IP address exposed) and try out the "Reply" tool. mw:Talk pages project/replying/prototype testing suggests a testing script to follow. Feedback can go on the talk page, or ping me, and I'll pass it along. Whatamidoing (WMF) (talk) 20:06, 28 January 2020 (UTC)

@Whatamidoing (WMF) I can post my just-now experience on the beta.wmflabs with a blow-by-blow set of hints for just getting to respond on Talk:Dog. Would you like this? I can put these hints on your talk page for now. --Ancheta Wis   (talk | contribs) 21:07, 28 January 2020 (UTC)

@Whatamidoing (WMF):, am I supposed to see something different on the beta Dog talk page? I don't see any options other than the normal "edit" links next to each section. --Ahecht (TALK
PAGE) 22:07, 28 January 2020 (UTC)

@Whatamidoing (WMF) and Ahecht: I guess we found a bug here, the "Reply" tool doesn't work if the discussion page has any comments before first section heading, and that page now does thanks to @Ancheta Wis:'s edit. I filed phab:T243869 and submitted a patch to fix this. Matma Rex talk 00:19, 29 January 2020 (UTC)

In the meantime, I've added a section heading, so the 'Reply' buttons are visible again, and any interested person can try out https://en.wikipedia.beta.wmflabs.org/wiki/Talk:Dog . Whatamidoing (WMF) (talk) 23:26, 29 January 2020 (UTC)

• We breach the principles of DRY ("Don't repeat yourself"), on a massive scale. For example, by storing citation metadata dozens - sometimes thousands - of times in individual articles rather than one, centrally. And we're currently importing six million subject descriptions from Wikidata, rather than working with the Wikidata community - our colleagues and friends, and in some cases ourselves - to develop common standards, and then transclude them. For a volunteer project with a surfeit of outstanding tasks, such behaviour is self-harming. A professional organisation would prohibit such nonsense. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:48, 28 January 2020 (UTC)

  Pigsonthewing, Sigh. WD has, on a massive scale, imported data without any control. Their anti-vandalism (in the broadest sense of the word) is practically non-existing. We need eons to check all the data and get it up to par, including having proper ways of protecting it. At the moment it is a blessing that some of our featured articles are not using the WikiData data as it has been vandalised without anyone noticing. Tell me, what is the use of a massive amount of data if you have no clue what is blatantly and scarily wrong? Dirk Beetstra ^{T C} 11:46, 29 January 2020 (UTC)

  More hyperbolic criticism of Wikidata not rooted in facts: "without any control" and "practically non-existing" are the scary words folks keep repeating to spread FUD. Wikidata's quality is ever-increasing, with more and more Wikipedia editions using Wikidata-driven infoboxes, bringing more eyeballs and quality checks. Wikidata does need more technical features to help out, like signed statements, but the future is inevitably going to include massive use of Wikidata. The sooner English Wikipedia wakes up to this, the better. Otherwise it's the same old saw: editors are quick to brush off Wikipedia's significant failures while deeming any problems with Wikidata as fatal. As to what Pigsonthewing said above: if we would only consolidate our citations into a structured database via Wikidata or some other system, that would bring orders of magnitude more efficiency to our references, and provide powerful new capabilities around quality, reliability and fact-checking. Rather than English Wikipedia trying to wing it with manual infoboxes, manual citations and manual short descriptions, it should attempt to join the 21st century and start unifying and collaborating. -- Fuzheado | Talk 02:19, 30 January 2020 (UTC)

  Fuzheado, OK, one of the WikiData items connected to one of our Featured Articles here is heavily vandalised, noted by ORES, no-one is doing anything about it (and the infobox on the only project (not a small one) that is using it is hence .. utterly incorrect .. an no-one eyeballs it). There are spamlinks all around (some more than a 1 year old), no-one does anything about it. Some items get spam-links added over and over, and instead of protecting all wikis from that rubbish only after 5 or so attempts of spamming someone protects it (if it even gets detected). And there are continuous attempts to vandalise up to hundreds of pages on all Wikimedia projects that use the data (of which above Featured article is a successful attempt), and that attempt is related to requests like this (if we would grant more sites like that for sites discussed in that thread the vandalism of the Featured article would have succeeded on an earlier attempt by the same user).

  I totally, completely, utterly agree with user:Pigsonthewing that that would be a major improvement, but you really have to get your data correct, protected. There is absolutely no reason that the 'immutable' data is freely editable (read: vandalisable, because that is the only reason to change it), and that is likely even true for a good portion of the other data. You may very well be right that the reliability of WikiData is improving continuously, but you have no clue if you are currently at 10% mistakes, 1% mistakes or at 1-in-a-million mistakes. That you can confidently say that you are improving however suggests that we are more in the percentages than in the parts per million. (and no, WIkiData is NOT a reliable source, it is just as volatile as any wiki). Dirk Beetstra ^{T C} 05:39, 30 January 2020 (UTC)

  I should have added here: now, if WMF had it's antivandalism tools up to date and controlled, then that vandalism would possibly have been detected and reverted (well, ORES did detect it ..). If WikiData had a way to protect correct and immutable data then you at least know that that is good data. At the moment, here on en.wikipedia, and also on WikiData and anywhere else, it is an utter mess. Protect your correct and immutable WikiData data, and you can fully automatically detect that data on en.wikipedia is wrong, and people will trust that data and just re-use the WikiData data. Now, WikiData is just a tool to vandalise up to a couple of hundred of pages across the same number of wikis in one go. --Dirk Beetstra ^{T C} 11:55, 29 January 2020 (UTC)

• It was written in PHP [1][2]. This is one of the underlying reasons why WMF software is so crap. There are other, more competently designed programming languages that are better able to handle the level of complexity of MediaWiki currently, and even more. Ducks and runs. MER-C 15:45, 30 January 2020 (UTC)
  • I feel you. I've done a little PHP work (inherited a project). In the few months I was working with it, I developed a deep loathing for the language. But, I've always looked at Wikipedia and said, "On the other hand, if it was possible to write one of the world's largest websites in PHP, that's obviously evidence that it can be a productive tool". Isn't FaceBook also a PHP shop? Same observation there. For that matter, English is a disaster of a language. The grammar, vocabulary, and spelling are all bizarre. I didn't even know that conjugating verbs was a concept until I learned Spanish in junior high school. Yet, we've managed to make effective use of English in transportation, literature, and information retrieval. There's certainly lots of problems with WikiMedia, but I'd be hard pressed to trace them back to the language it's implemented in. -- RoySmith (talk) 16:20, 30 January 2020 (UTC)
    • True, but it takes more effort, stricter discipline and better software engineers in order to do so. Facebook can afford to pay billions to make that happen. The WMF? Not so much. MER-C 17:12, 30 January 2020 (UTC)

      MER-C, mwaagh ... they do have a lot of money to spend and developers to make stuff. See, we have Flow, MediaViewer, Visual Editor ... SuperProtect. We even have sarcasm.php (strangely, onlyrequired once ...). Dirk Beetstra ^{T C} 06:51, 31 January 2020 (UTC)

      FYI, mw:Extension:Sarcasm was not developed by and is not maintained by the WMF.^[whoosh] Anomie⚔ 14:49, 31 January 2020 (UTC)

      Anomie, I knew that. <sarcasm>quite a bit of the material that however was developed by the WMF is not maintained by them either.</sarcasm> Dirk Beetstra ^{T C} 10:34, 2 February 2020 (UTC)

  MER-C, the crappiness of php as a language is mostly based in it's history. As a language in itself (today, post 7.0) it is not that bad any longer. No, the crappiness now mostly comes from that fact that we have 16 year of accumulated crappiness, no specifications and a gazillion unwritten use cases that need to be supported. —TheDJ (talk • contribs) 11:26, 31 January 2020 (UTC)

• +1 on fixing old bugs, like the >6-year-old phab:T223002 in NavPops. —[AlanM1(talk)]— 08:32, 31 January 2020 (UTC)

  AlanM1, that's a community supported tool, it does not fall under the foundation. —TheDJ (talk • contribs) 11:22, 31 January 2020 (UTC)

• The talkpages work reasonably well, at least in editing terms. OK we need an “autosign on talkpages” option and it would be nice to be able to watchlist sections as well as whole pages. The bigger problem is the number of unanswered queries on article talk pages. Such things could be done if we didn’t have a philosophy among the developers that if you neglect software for long enough you can persuade people to ditch it for something shiny and new. It would be really helpful if there were a report listing talk page queries relevant to each wikiproject, with the option to mark such sections as resolved and/or not relevant to individual wikiprojects. On a bigger scale, we need to take a leaf out of the Chinese Wikipedia and give our readers the option of displaying Wikipedia in American English, Indian English or English. Most of that could be done programatically, and where it needs hidden templates or links to show whether you are using bonnet as in headgear, part of a car or type of chilli it would make for uncontentious reader involvement opportunities. Oh and probably our most urgent software need, we need to make this site easily editable on smartphones. Editing levels are still clearly above the 2014 lull, but there are lots of community skews that come from our being almost a broadcast medium to the smartphone generation. ϢereSpielChequers 07:20, 8 February 2020 (UTC)

Deleting and reordering parameters using AWB

Is it possible to delete and reorder template parameters using AWB? I can't seem to find that option anywhere. If not, does anyone have any regexes for deleting the content of entire parameters within a template or for reordering existing ones? Thanks, – Srdjan m (talk) 19:47, 29 January 2020 (UTC)

You could do that with regular expressions, though making an expression that would always match every possible parameter value could be tricky. I would not like to give examples, unless I try it out first! Graeme Bartlett (talk) 13:03, 31 January 2020 (UTC)

@Graeme Bartlett: Well, if you can figure out how to do it reliably, do let me know. You can use any example you want for deleting parameters and reordering them. Infobox examples might be best. – Srdjan m (talk) 11:11, 4 February 2020 (UTC)

Srdjan m, I am not sure if it helps. Years and years (and years) ago we ran AWB to convert tables of data into a sectioned infobox. That required the bot to move lines around. I think the trick I used was to put a dummy text in front of each line, with a number for in which section a piece of info needs to go. Then you have a regex that swaps 'line2 (data)' and 'line1 (other data)' and all combinations, and you repeat that swap x-times. At some point, there are no cases where there is a 'line2' before a 'line1' and you are all sorted. The code can be found here: User:Beetstra/Chemicals. It results in edits like diff (note that in the old table there is no 'order' in the information in the old diff, and that in the output the data is put into the correct section. Note that the script is 'buggy' (in that it may miss misspelled stuff, and that it eats other things on the page). Dirk Beetstra ^{T C} 13:35, 4 February 2020 (UTC)

How to using unstripNowiki in subst: ?

Why {{unstripNoWiki}} notwork in {{subst:}}?

• See Special:Diff/938286178
• In lua console, I try to print the return value returned from mw.text.unstripNowiki, then I got a string like ?UNIQ...-nowiki-00000000-QINU?, so I think mw.text.unstripNowiki notwork in {{subst:}}.

BUT WHY? And How to fix it?--Yu-Fan 宇帆 (talk) 07:42, 30 January 2020 (UTC)

Tested it and it does seem to work, the substitution simply places the template parameter in the Unstrip module invocation, so it will look the same but it is actually displayed from the function invocation. --qedk (t 桜 c) 15:29, 30 January 2020 (UTC)

@QEDK:

{{#invoke:Unstrip|unstripNoWiki|<nowiki>[[12]]</nowiki>}} → 12 (work)

but {{subst:#invoke:Unstrip|unstripNoWiki|<nowiki>[[12]]</nowiki>}} → [[12]] (not work)

{{safesubst:#invoke:Unstrip|unstripNoWiki|<nowiki>[[12]]</nowiki>}} → [[12]] (not work)

I think it doesn't work. (see sandbox test Special:diff/938411191)--Yu-Fan 宇帆 (talk) 01:00, 31 January 2020 (UTC)

• Does anyone know how to DELETE <nowiki><nowiki/> tag and keep the inner text in {{subst:}}? --Yu-Fan 宇帆 (talk) 06:57, 5 February 2020 (UTC)

  I think you're misunderstanding the purpose of the template. It displays the text as it is supposed to be displayed without the tags (by invoking the module), it does not substitute and remove the nowiki tags itself. --qedk (t 桜 c) 07:27, 5 February 2020 (UTC)

"Mark all notifications as read" is broken on mobile

If I click the "mark all notifications as read" button on mobile, the notifications clear on mobile, but on the desktop version I still see unread notifications. I've had this bug for a few weeks now, is anyone else having this problem? Narutolovehinata5 ^tccsdnew 00:14, 3 February 2020 (UTC)

It eventually remembers. I have had to click mark all over several sessions before it stuck. At this point, I just click desktop mode on the bottom of wikipedia and clear notifications from there. Slywriter (talk) 14:31, 5 February 2020 (UTC)

Tech News: 2020-06

Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.

Changes later this week

• The new version of MediaWiki will be on test wikis and MediaWiki.org from 4 February. It will be on non-Wikipedia wikis and some Wikipedias from 5 February. It will be on all wikis from 6 February (calendar).

Tech news prepared by Tech News writers and posted by bot • Contribute • Translate • Get help • Give feedback • Subscribe or unsubscribe.

20:04, 3 February 2020 (UTC)

Is that all? Looks like a boilerplate with no extras. --Redrose64 🌹 (talk) 20:19, 3 February 2020 (UTC)

Redrose64, I was surprised too. Is that really it? moonythedwarf (Braden N.) 20:20, 3 February 2020 (UTC)

It appears so from [3]. PrimeHunter (talk) 22:42, 3 February 2020 (UTC)

Pinging BBearnes (WMF), who populated the changelog file for the previous version. Is there a changelog for this version? Sorry to bother you if you are the wrong staffer to ask. – Jonesey95 (talk) 00:11, 4 February 2020 (UTC)

It's complete. There's just not much to say this week. – Ammarpad (talk) 04:44, 4 February 2020 (UTC)

Apologies - we've still got a bug with automatic generation of deploy notes. See T243330: train-deploy-notes Jenkins job fails in conjunction with branch.py for that. We'll get something posted here. - Brennen (talk) 18:46, 5 February 2020 (UTC)

Oldest "living people"?

Is there a way to generate a list of the oldest (or, put another way, earliest born) people in Wikipedia listed in Category:Living people? I am thinking that we might have some older articles on people who were listed as living at the time, and have since died. I think the best place to start looking is with people who would be remarkably old if they were in fact still alive. BD2412 T 03:22, 4 February 2020 (UTC)

I'm just spit-balling here, but creating a WikiData entry for each person with their birth year would make creating such a list much easier. We already have categories for people born in certain years. Maybe go through categories like Category:1900 births, Category:1901 births, etc. up to, say, Category:1950 births. For each name, check to see if the person is in a "deaths" category, Category:Living people, or neither, then make your list from there. Slogging through hte resulting list will be a chore though - it will only include people who weren't well-known enough at the time of their death for someone to see an obituary and say "hmm, better check to see if Wikipedia has an article about them." Many newspaper obituaries go behind paywalls after a few days or weeks, making verification more difficult. davidwr/(talk)/(contribs) 04:01, 4 February 2020 (UTC)

Here's a dumb idea: an insource search for "birth date and age\|18" to (try to) show people born in the 19th century that are allegedly still alive. This will more likely show that the wrong template is being used, or some other error, or commented-out templates. You can do the same thing for birth years starting with 190. There are lots of commented-out templates in the results, but you might see something interesting, like Virginia McLaurin, who is (according to Wikipedia) 110 years old.

Here's how to do a cross-category search with PetScan. Change the year in the Categories box to modify the search. – Jonesey95 (talk) 04:34, 4 February 2020 (UTC)

Our search box can also do this simple case: incategory:"Living people" incategory:"1920 births". PrimeHunter (talk) 04:39, 4 February 2020 (UTC)

d:Q23074256 offers 1909 or 1917. both allegedly imported from enwiki. Very vaguely I recall a rule (on commons?) that nobody can be 130 for the purposes of copyright. –84.46.52.96 (talk) 04:47, 4 February 2020 (UTC)

We seem to have a wealth of techniques. Excellent, thanks! BD2412 T 05:17, 4 February 2020 (UTC)

Throttling of new users

throttling message shown to a Wikipedia editor

For the first time today, I've had trainees throttled while making constructive edits (screenshot above). The only new aspect of is that they created accounts this norming, whereas usually I get them to do in advance (but often some do not). Is that the cause, or has something else changed? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:37, 4 February 2020 (UTC)

@Pigsonthewing: was the only edit they were doing "edit" at the time, or something else? (Perhaps adding external links?). If you have live attendees, you can use your eventcoordinator access to add the +confirmed flag to them (for up to 10 days) to help bypass certain newbie filters. — xaosflux ^Talk 12:43, 4 February 2020 (UTC)

It sounds like the old problem of a maximum number of edits per minute allowed per IP address for IPs and Newbies. The guaranteed way of hitting it is to include the phrase "all hit save now" in your training presentation. Using a Wifi hotspot for some editors, or as Xaosflux mentions setting some editors as confirmed usually solves the problem. Of course the more editors on one IP address the more likely this is to happen. ϢereSpielChequers 20:35, 7 February 2020 (UTC)

More template troubles, different problem

I'd like some assistance with a template coding problem that's been brought to the attention of WikiProject Canada. Our List of villages in Canada is split up into separate sublists for each province, most (but not all) of which have been made collapsible due to the number of villages in each list. However, in all of the collapsed lists, the template doesn't actually display any column headers to explain what the data in each column represents. But if we attempt to edit the templates, column headers are present in them — which means they're just not displaying to the end user in page view mode.

So I wanted to ask if somebody with more experience in template coding can figure out if there's something in the template coding that can be changed to make the column headers display properly. Thanks. Bearcat (talk) 14:49, 4 February 2020 (UTC)

Before attempting to solve that issue, it should be better to observe MOS:DONTHIDE and rework the layout (or split if needed). --Gonnym (talk) 14:53, 4 February 2020 (UTC)

Just show everything by default. The article is not that long, and you'd be complying with MOS. – Jonesey95 (talk) 15:56, 4 February 2020 (UTC)

Connection issues, Feb 4, 2020

Well, here we go again. A few minutes ago there were severe connection problems with all server clusters (At least for me!), and, as of writing this, ESAMS and EQIAD are at 0% ATS backend availability for upload per grafana. Anyone know what's going on? --moonythedwarf (Braden N.) 17:01, 4 February 2020 (UTC)

The Main Page technical layout has gone live

Should there be any issues, please report them to Talk:Main_Page#Tech_update_has_gone_live. Thank you, — xaosflux ^Talk 00:08, 5 February 2020 (UTC)

Xaosflux, thank you —TheDJ (talk • contribs) 08:38, 5 February 2020 (UTC)

Some cleanup templates are lacking substituted dates

{{More citations needed}}, {{COI}}, {{No footnotes}}, {{Disputed}}, – examples of cleanup templates that have working, functional substituted dates

{{Tone}}, {{POV}}, {{Cleanup reorganize}}, {{Like resume}} – examples of ones that don't

Same thing with inline templates:

{{Citation needed}}, {{Incomprehensible inline}}, {{Clarify}}, {{Who}}, – examples of ones that work

{{POV statement}}, {{Date missing}}, {{Dubious}}, {{Excessive citations inline}}, – examples of ones that don't

When you just insert the template, this is supposed to show up:

Month and year
{{subst:CURRENTMONTHNAME}}
{{subst:CURRENTYEAR}}

I don't know why this does not occur with some of them. I've scoured the source code for all that I could find. I looked long and hard for similarities between the ones that worked and the ones that didn't. Couldn't find any. They all seem to have the code in place to substitute dates, and no other abnormalities.

It has been this way for some years now, I believe. Please can someone investigate this issue so that it can be fixed. Thank you. · • SUM1 • · (talk) 04:33, 5 February 2020 (UTC)

{{Tone|{{subst:DATE}}}} works fine for me. It produces {{Tone|date=February 2020}}. I did the same thing with {{POV}}, {{Cleanup reorganize}}, and {{Like resume}}, and they worked fine. – Jonesey95 (talk) 05:27, 5 February 2020 (UTC)

@Jonesey95: Oh, that's not what I meant. The thing I said is supposed to show up when you insert the template does not show up for {{Tone}} and the others. · • SUM1 • · (talk) 05:40, 5 February 2020 (UTC)

@Jonesey95: Don't worry, I've figured it out. Some of them have

== Template data ==

<templatedata>

{

"params": {

"date": {

"label": "Month and year",

"description": "Month and year of tagging; e.g., 'January 2013', but not 'jan13'",

"type": "string",

"autovalue": "{{subst:CURRENTMONTHNAME}} {{subst:CURRENTYEAR}}",

"suggested": true

}

},

"description": "a description"

}

</templatedata>

in their documentation, and some of them don't. Perhaps someone could help me with adding this to all of them. · • SUM1 • · (talk) 05:54, 5 February 2020 (UTC)

Good detective work. That is the Template Data programming code used by the Visual Editor. If you insert one of those templates using the Visual Editor, and the template does not have the Template Data programming code in its documentation (please don't ask why template programming code lives in the unprotected documentation page), you won't get the date automatically. Most, or maybe all, of the above templates will have a date automatically added by a bot, so it should not be a problem. – Jonesey95 (talk) 06:00, 5 February 2020 (UTC)

@Jonesey95: Indeed. Though, I was considering mentioning, just like with the {{Unsigned}} and {{Unsigned IP}} templates, I've seen multiple times the bots fail to add the dates at all, years later. I also find out TemplateData is also used for the 2017 wikitext editor, which explains why it works the same way when I edit in source mode while not logged in, as of today. · • SUM1 • · (talk) 06:33, 5 February 2020 (UTC)

There's something wrong with the Harvard cites at Golden plates

I fell down the rabbit hole of trying to fix the Harv error on ref #5 - spent way TOO much time but when I tried to fix it, my so-called "fix" then would cause other refs to fail etc.. I cannot figure out what is wrong. Please someone correct this and post here what caused the Harv error. Thanks, I give up. Good night. Shearonink (talk) 08:36, 5 February 2020 (UTC)

I've fixed one error, only for another error to pop up instead. Please see my comment at Talk:Golden plates. rchard2scout (talk) 15:14, 5 February 2020 (UTC)

P.s.: note for anyone else, in order to see Harv errors, you need this userscript. rchard2scout (talk) 15:14, 5 February 2020 (UTC)

The cascading Harv errors happened to me too - there is something technically wrong with the Harv-cite coding on the article. Shearonink (talk) 19:12, 5 February 2020 (UTC)

After a quick inspection, I don't see that there is anything technically wrong at Golden plates. Rather, it would appear that editors have been careless in writing either the citation templates or the {{harv}}, {{harvnb}}, and {{harvtxt}} templates. (Why does the article use three variants of the harv templates?) This issue is best handled at the article's talk page. If there is anything technically wrong, let me know.

—Trappist the monk (talk) 15:32, 5 February 2020 (UTC)

I see Harv errors quite a bit on articles that use Harv cites (thanks to Ucucha's wonderful script). Usually I can fix the issues myself but whatever is going on with this article is something beyond my poor expertise. I don't know what is technically wrong, I was unable to fix the problem and so was Rchard2scout. I don't know why the article has so many Harv cite variants and, frankly, it is huge and I don't have the time to bring them all into agreement with each other. I just want to know what's wrong so I can learn and fix it the next time I see it. I think the extra eyes and the more robust traffic here at Village pump:technical (3,305 watchers!) will help solve whatever is going on, the article only has about 160 watchers. Even if this post doesn't quite belong here please leave this thread up so we can all figure it out. Shearonink (talk) 19:12, 5 February 2020 (UTC)

What was wrong here was just many, many Harv errors. And Ucucha's script wasn't showing them all at the same time, so I went through them one by one. No actual bugs here, except maybe the fact that the script doesn't show all errors at once. rchard2scout (talk) 16:14, 6 February 2020 (UTC)

THANK YOU. Heh, I've never seen an article that had so many Harv errors that the script (sort of) gave up and only mentioned the first occurrence - that's a new on on me. Shearonink (talk) 20:11, 6 February 2020 (UTC)

Toolforge questions

This is the second time I'm attempting to get my bots set up at Toolforge and failed. Is there anyone who is willing to give me some guidance? Steps I've completed:

• Wikitech account (developer account)
• Created a tool on Toolforge
• Gotten shell access
• Added public key to Wikitech account
• Private key stored in /.ssh w.r.t root
• tools-login.wmflabs.org added to known_hosts

Steps where I'm stuck:

• user@tools-login.wmflabs.org: Permission denied (publickey,hostbased).
• What to do next? --qedk (t 桜 c) 13:35, 5 February 2020 (UTC)

  QEDK, are you trying to log in as user and not qedk? --AntiCompositeNumber (talk) 13:44, 5 February 2020 (UTC)

  I feel amazingly stupid, thanks. I added a custom host-user configuration to resolve this. --qedk (t 桜 c) 13:55, 5 February 2020 (UTC)

  @QEDK: Live help is available at #wikimedia-cloud ^connect, so you know :) — MusikAnimal ^talk 19:16, 5 February 2020 (UTC)

  You'll also want to subscribe to the cloud services mailing list -- RoySmith (talk) 20:57, 5 February 2020 (UTC)

• @AntiCompositeNumber, MusikAnimal, and RoySmith: Since I'm unable to get help on the IRC channel, I thought it better to ask here. Why doesn't $ source ~/.bash_profile persist between sessions on Toolforge, I've tried it with and without a venv, in my toolforge account, as the tool, saving it to ./profile, just doesn't work and it's very annoying working on macOS with Toolforge. Secondly, Pywikibot seems unable to make newlines from "\n", so how do I deal with that. It was generating newlines properly for a while and then it didn't, some secret I'm missing out? --qedk (t 桜 c) 08:25, 9 February 2020 (UTC)

@QEDK: Unfortunately, "on Toolforge" can mean any of a number of different things. Are you talking about directly on the bastion hosts, or grid engine, or a docker instance, or maybe something else? Could you give the exact sequence of commands you're using? -- RoySmith (talk) 14:58, 9 February 2020 (UTC)

$ ssh tools-login
$ become qedkbot
$ source venv/bin/activate
$ cd pywikibot-core
$ python3 pwb.py login.py

@RoySmith: This will then throw a Unknown locale: UTF-8 error, which I have to fix by adding the variables to bash_proifle every session. tools-login is aliased to tools-login.wmflabs.org. --qedk (t 桜 c) 15:12, 9 February 2020 (UTC)

Hmmm. I took the liberty of poking around in /data/project/qedkbot. All I see in your .bash_profile there is setting PYTHONPATH. Does that not get set properly? In the sequence of commands above, if you do printenv right after the become qedkbot, does your PYTHONPATH not include the value you set in .bash_profile? -- RoySmith (talk) 15:24, 9 February 2020 (UTC)

@RoySmith: That is exactly the issue yep, I set the variables two hours ago on my latest bot run (had to!), and now you don't see them. I know setting the variables at the start of each session works because then I don't get the Unknown locale error, but if you want I can try printenv. --qedk (t 桜 c) 15:43, 9 February 2020 (UTC)

Yes, please run the printenv and see what you've got after become qedkbox That should at least narrow it down to whether the environment variable isn't getting set, or if it's getting set but maybe overridden by the venv or something. Another thing to try is after you activate the venv, fire up an interactive python and do import sys; print(sys.path) to see if it made it that far. You might also try sticking echo "I'm here" in your .bash_profile and see if that gets printed; that should confirm the .bash_profile is getting run. -- RoySmith (talk) 15:55, 9 February 2020 (UTC)

• @QEDK: I saw your BRFA. I assume that means you've solved this. What turned out to be the problem? -- RoySmith (talk) 01:28, 10 February 2020 (UTC)

  @RoySmith: I wish it was fixed, a bit loaded up on coursework so didn't get to debug it yet. Still setting the variables every session, I'm guessing submissions to the grid engine will not throw an Unknown locale error. --qedk (t 桜 c) 06:48, 10 February 2020 (UTC)

  @RoySmith: This is very, very weird - the environment variables show up just fine in printenv (inside and outside venv). Here's the output for sys.path: ['/data/project/shared/pywikibot/core', '/shared/pywikibot/core/scripts', '/mnt/nfs/labstore-secondary-tools-project/qedkbot/venv/lib/python35.zip', '/mnt/nfs/labstore-secondary-tools-project/qedkbot/venv/lib/python3.5', '/mnt/nfs/labstore-secondary-tools-project/qedkbot/venv/lib/python3.5/plat-x86_64-linux-gnu', '/mnt/nfs/labstore-secondary-tools-project/qedkbot/venv/lib/python3.5/lib-dynload', '/usr/lib/python3.5', '/usr/lib/python3.5/plat-x86_64-linux-gnu', '', '/mnt/nfs/labstore-secondary-tools-project/qedkbot/venv/lib/python3.5/site-packages', '/mnt/nfs/labstore-secondary-tools-project/qedkbot/pywikibot-core', '/mnt/nfs/labstore-secondary-tools-project/qedkbot/venv/lib/python3.5/site-packages/pycparser-2.19-py3.5.egg', '/mnt/nfs/labstore-secondary-tools-project/qedkbot/venv/lib/python3.5/site-packages/IPython/extensions', '/mnt/nfs/labstore-secondary-tools-project/qedkbot/.ipython']. Tells you something? --qedk (t 桜 c) 09:44, 11 February 2020 (UTC)

  I was indeed correct, for some reason source ~/.bash_profile was not setting the variables in .bash_profile (which makes me wonder where they are being stored). Configured around Jupyter Notebook to let me edit hidden files (not a Vim fan) and then changed .bash_profile myself, seems to stick between sessions once added - which leads me to think it is the command itself that is not working (which is weird!). --qedk (t 桜 c) 09:56, 11 February 2020 (UTC)

I'm pretty confused at this point. The sys.path output you include above does indeed include the PYTHONPATH entries from your .bash_profile. One odd thing I noticed is that your .bash_profile doesn't end with a newline. I don't know if that bothers bash or not. I've never used pywikibot. I remember looking at it briefly a long time ago and coming to the conclusion that it was crazy complicated. I took a look at the pwb.py source. Yeah, it's nuts. It's doing all kinds of crazy stuff with mutating sys.path and os.environ, and then executing other scripts. I honestly can't tell WTF it's trying to do. This looks like a pywikibot-specific problem, so I suggest asking the pywikibot developers. Sorry I can't be more help. -- RoySmith (talk) 16:56, 11 February 2020 (UTC)

@RoySmith: I don't think it's pywikibot, it is definitely something to do with the variables not getting saved in .bash_profile. I still appreciate all the help. Thanks a ton! -qedk (t 桜 c) 11:30, 12 February 2020 (UTC)

Better "n changes since last visit" in Watchlist

Before I ask for this in Phabricator I want to see if anyone besides me supports it.

Special:Watchlist shows the number of changes for each day as a clickable link. This is great for the current day but there is no single link that will show the diff from your last visit until now if there were unseen changes before today.

For example, this is a line in my watchlist from yesterday and today:

5 February 2020

filled dot, expansion triangle, timestamp Wikipedia:Village pump (proposals)‎‎(3 changes | history)

4 February 2020

filled dot, expansion triangle, timestamp Wikipedia:Village pump (proposals)‎‎ (30 changes | history)

I recommend changing it so it shows the "changes since last visit" so in addition to showing it if either there are a mix of unread and read messages on a given day, it also shows with the total number of unread messages if there are unread messages across more than one day.

This would now look like:

5 February 2020

filled dot, expansion triangle, timestamp Wikipedia:Village pump (proposals)‎‎(3 changes | 33 changes since last visit | history)

4 February 2020

filled dot, expansion triangle, timestamp Wikipedia:Village pump (proposals)‎‎(30 changes | 33 changes since last visit | history)

Note the "33 changes since last visit" is the same for both days. I could click on either link to "fully catch up" for this page. Right now, I click on History, scroll down to the newest link I have read, and click on the "cur" link to get the same "diff" page.

Would others find this useful enough to suggest at Phabricator? davidwr/(talk)/(contribs) 17:29, 5 February 2020 (UTC)

This would be phab:T4877. --Izno (talk) 18:07, 5 February 2020 (UTC)

(See also phab:T10681. --Izno (talk) 18:12, 5 February 2020 (UTC))

Unexpectedly caught by pending changes

I'm autoconfirmed and all my edits should be automatically accepted on pages with pending changes. Bat as food has had pending changes since the end of last month, and I've edited the article many times since then, with all my edits automatically accepted. My most recent edit today was not automatically accepted, however, and had to be accepted with someone with pending change review rights. As far as I can tell, I have not had any rights changed. Any reason why that would have happened? Enwebb (talk) 21:56, 5 February 2020 (UTC)

It's probably because this IP edit had to be accepted first. Graham87 03:05, 6 February 2020 (UTC)

I had an edit since then that was autoconfirmed. If you look at the history, I made an automatically confirmed edit, then another editor did the same. My next edit was then pending, despite being autoconfirmed and no other pending revisions. Enwebb (talk) 16:22, 6 February 2020 (UTC)

Since the PC has expired and there is no way to see the PC history (afaik), I say we can assume it was a one-time glitch. Either way, not a deal-breaker and what Graham87 said has the highest probability (maybe you missed a pending revision somewhere?). --qedk (t 桜 c) 16:47, 6 February 2020 (UTC)

This may be another instance of phab:T234743, which is discussed at #Permissions pop-ups? further up this page. -- John of Reading (talk) 16:53, 6 February 2020 (UTC)

John of Reading, seems probable, thanks. I just wanted to make sure I surfaced this if it was a new bug. Enwebb (talk) 18:05, 6 February 2020 (UTC)

@QEDK: Go to the page history; click View logs for this page, and look for "configured pending changes settings". If there are lots of entries, you can restrict it by clicking the "All public logs" dropdown and selecting "Pending changes log". --Redrose64 🌹 (talk) 20:02, 6 February 2020 (UTC)

@Redrose64: I was referring to the tags added in the page history itself, "automatically accepted", "accepted by xx" and so on, which are permanently lost (again, afaik and as per WP:PC#Pending changes adds highlighting that is lost when disabled) when PC is removed or it expires. --qedk (t 桜 c) 20:09, 6 February 2020 (UTC)

Oh, that's the Review log. --Redrose64 🌹 (talk) 20:39, 6 February 2020 (UTC)

Table sorting fustercluck

I've come across a significant problem with List of Canadian Inuit, which I'd like to solicit some help in cleaning up.

The default sort order in the list is by surname, using the {{Sortname}} template without issue. The problem comes up in the other columns which are using the generic {{sort}} template instead — in order to impose extra control on the specific order in which the entries would be sorted if a reader attempts to re-sort the list on that column, each entry is simply sortkeyed with sequential numbers: the entry that should become first is sortkeyed "001", the entry that should become second is sortkeyed "002", the entry that should become third is sortkeyed "003", and on and so forth. And this is, further, being used to control more than just each column in isolation: for example, in order to ensure that the four people in the list who were born in Manitoba remain in alphabetical order even after the list is re-sorted on the "region" column, those four people are sequentially sortkeyed as 003, 004, 005 and 006 in that column.

What this means, however, is that I can't simply add a new entry for an Inuit person who isn't already in the list, such as the Inuit musician whose article I was attempting to add when I discovered this — each time I want add even one new person, I would also have to undertake an extended project, anywhere between one to three hours long, of manually readjusting every single numbered sortkey in the entire rest of the article to reflect each individual column's new desired sort order. If there's a new Inuit person from Manitoba to add, for example, I would have to manually +1 every sortkey in that entire column for every single entry through the entire rest of the list in the process of adding that one person. Which in turn ends up meaning that the list can't be properly added to or completed at all, because no editor can or should ever reasonably be expected to commit to actually doing that much work. Only one new person has actually been added to this list since 2015, even though there are literally dozens of missing entries for Inuit people who do have articles to add — and I suspect that the opaqueness and dauntingness of this sequential sortkeying system is a big part of the reason for that.

Obviously, we shouldn't be using opaque coding systems like this, or creating this much excess work for editors who try to actually add new entries to the list — the template coding needs to be as transparent and simple as possible, and the only time a column should ever be sorted on numbers is when the visible data in the column is numbers. So is there any way to quick-fix this in an automated manner, or is somebody just going to have to buckle down and manually remove all the numbered sortkeys? Bearcat (talk) 16:48, 6 February 2020 (UTC)

@Bearcat: A search-and-replace from {{sort| to {{subst:P2| will remove the calls to the {{sort}} template, since Template:P2 is a template that outputs its second parameter. The space at the start of the replace string makes sure that the table cells are not completely empty; without that space, the table formatting is damaged. I've tried this at User:John of Reading/X1. -- John of Reading (talk) 18:20, 6 February 2020 (UTC)

Okay, thank you, that worked. I got confused by it for a second and thought it had broken other things, but it actually didn't. Bearcat (talk) 18:33, 6 February 2020 (UTC)

I advice giving Help:Sorting a proper read as well. —TheDJ (talk • contribs) 18:42, 6 February 2020 (UTC)

RegExp rule for changing every Hegira years to Gregorian years

Hello, Can someone write a rule for me? I want to change every Hegira years to Gregorian years. For example: Hegira years + 622 = Gregorian years. I have an incomplete rule. It's /\b\d{4}\b/g. Thanks, ⇒ AramTalk 19:19, 6 February 2020 (UTC)

Sadly, I don't think you can do that with a simple regex. Regexes alone cannot do math. The language you're using may be able to accomplish this in a different way, however. What language are you using?--Jorm (talk) 19:41, 6 February 2020 (UTC)

@Jorm: I speak in Central Kurdish (Sorani) and ckb:دەستپێک is my home Wikipedia. ⇒ AramTalk 20:03, 6 February 2020 (UTC)

@Aram: I meant the programming language you are using, not your local language.--Jorm (talk) 20:26, 6 February 2020 (UTC)

You might want to look at some of the templates in Category:Date-computing templates. If you can isolate the Hegira year, you can use {{#expr:622+HEGIRAYEARGOESHERE}} Disclaimer: I haven't actually tried this. See m:Help:Calculation for details on #expr. davidwr/(talk)/(contribs) 20:13, 6 February 2020 (UTC)

It would need to be substituted to work (but yes, it does work): {{subst:#expr:622+HEGIRAYEARGOESHERE}}. So, if you can isolate exactly the year with regex and somehow store it as a variable, it is possible, preferably using pywikibot/AWB. --qedk (t 桜 c) 20:23, 6 February 2020 (UTC)

It's not a constant 622 either, since the Islamic calendar is a lunar calendar of 354 or 355 days, and so the Islamic New Year occurs ten or eleven days earlier each Gregorian year. This means that whilst in the early days of the Islamic calendar there was a 622 year difference in the numbering, this has now shrunk to 579 (2020 CE spans 1441-1442 AH, and 1441 AH spans 2019-2020 CE). It can be shown that in several thousand years time, the two calendars will be in step (as far as numbering goes) for a few years, before diverging again. That aside, there are formulae at Hijri year#Formula. --Redrose64 🌹 (talk) 20:34, 6 February 2020 (UTC)

@Jorm: Unfortunately, I can't understand JS. @Davidwr: Actually, I didn't understand you. Sorry! @QEDK: Yes, I want to use the rule in AWB software. And by using "Find and Replace" do it. @Redrose64: Approximately, 622 is correct. Thank you for helping! ⇒ AramTalk 20:53, 6 February 2020 (UTC)

@Aram: Either of the following will give you the date you want, just replace 1398 with the correct Hegira year

{{User:Davidwr/Add622|1398}}

or, preferably,

{{subst:User:Davidwr/Add622|1398}}

The first one will stop working once I delete the page User:Davidwr/Add622. The page consist of only one line:

{{#expr: (622+{{{1}}}) }}

Here is what it renders as:

2020

davidwr/(talk)/(contribs) 21:04, 6 February 2020 (UTC)

Self-reply: I didn't realize that you were looking at the Arabic calendar. The above code won't work except when the year difference is 622 years. Sorry. davidwr/(talk)/(contribs) 21:06, 6 February 2020 (UTC)

@Davidwr: Thank you for your good explanation! But this way is so hard and we should do it one by one. I want to change all by one click. In past, my teacher wrote a code for website and i saw it worked properly, but after putting it to Wikipedia JS file, it didn't work. the following is my teacher codes:

function hejiraYearFix(){

    var txt = document.getElementById("wpTextbox1").value;

    
    var res = txt.replace(/\b\d{4}\b/g,function(m){
        return parseInt(m)+622});
    
    document.getElementById("wpTextbox1").innerHTML = res;
}
window.hejiraYearFix = hejiraYearFix;

Hope It is a useful code for more explaining. ⇒ AramTalk 21:26, 6 February 2020 (UTC)

@Aram: You say Approximately, 622 is correct. - but for present-day dates, it's way off. The current Hijri year is 1441 AH, but 1441+622=2063 (alternatively, 2020-622=1398), so calculating on the basis that the difference is 622 means that it's 43 years out. --Redrose64 🌹 (talk) 22:43, 6 February 2020 (UTC)

Wikipedia does provide date conversion functions. This {{#time:xmj xmF xmY|6 February 2020}} renders as 11 Jumada al-thani 1441. See Help:Time function for details. Granted, it's not the search-and-replace that you wanted, but it may be useful. 21:32, 6 February 2020 (UTC)

Template:The Interviews name usage adds Category:Academy of Television Arts & Sciences Foundation Interviewees

I have a list of interviewees from the website and I want to add [[Category:Academy of Television Arts & Sciences Foundation Interviewees]] to each page. There are about 900 persons with pages that use Template:The Interviews name. Is there a software solution?

T3g5JZ50GLq (talk) 02:00, 7 February 2020 (UTC)

I could do it using WP:AWB, but I wouldn't want to do it without consensus. — Arthur Rubin (talk) 02:32, 7 February 2020 (UTC)

@T3g5JZ50GLq: I see you created Category:Academy of Television Arts & Sciences Foundation Interviewees yesterday. It does not seem suitable for a category and could easily be nominated for deletion as non-defining. See WP:COPDEF, WP:NONDEFINING, WP:CATDEFINING. PrimeHunter (talk) 12:20, 7 February 2020 (UTC)

@PrimeHunter: "a defining characteristic is one that reliable, secondary sources commonly and consistently define"

The fact that "TV Academy" spent money to get 2 hours of video of TV industry personnel to record for posterity, with Google paying for online distribution, seems pretty significant. It ranks these 900 people with a very high notability out of the more than (160,000+17,500+9,081+7,000+350) personnel in the U.S. TV industry. T3g5JZ50GLq (talk) 12:53, 7 February 2020 (UTC)

Page preview needs Delay before the preview appears

Can you reuse the interface of Reference Tooltip with Delay before the tooltip appears (in milliseconds) for Page preview interface?

"Page previews (get quick previews of a topic while reading a page)"

T3g5JZ50GLq (talk) 02:27, 7 February 2020 (UTC)

Some templates do not expand in a Citation bot's talkpage archive

In User talk:Citation bot/Archive 19 there are multiple instances of templates {{fixed}}, also {{wontfix}}, {{notabug}} and possibly others, which display their names in double curly braces instead of their content. At the same time they're correctly displayed at the main page User talk:Citation bot (at least until the tagged issue report is moved to archives). Any ideas what went wrong? --CiaPan (talk) 08:04, 7 February 2020 (UTC)

@CiaPan: Nothing went wrong, as this is a documented feature of the archiving bot being used there. See User:ClueBot III#Optional parameters, The bot will also convert {{templates}} in this list to {{tl|escaped templates}} upon archival -- John of Reading (talk) 09:04, 7 February 2020 (UTC)

Wow, it hasn't even passed my mind that might be deliberate. Thank you John of Reading for the prompt and clear explanation. --CiaPan (talk) 09:19, 7 February 2020 (UTC)

It was a bit of a surprise to me when I first ran into it. -- John of Reading (talk) 09:22, 7 February 2020 (UTC)

Error deleting file: An unknown error occurred in storage backend "local-multiwrite".

Tracked in Phabricator
Task T244567

Attempting to delete the expired PRODs File:CallipygianVenus.jpg and File:LocationAbkhaziaFinal.png throws this error; other files delete as normal. Jo-Jo Eumerus (talk) 10:07, 7 February 2020 (UTC)

Current state, both files are again visible in Category:Expired proposed deletions, one attempt to move the newer file without leaving a redirect + move it back without leaving a redirect apparently didn't help. –84.46.52.123 (talk) 09:52, 10 February 2020 (UTC)

Is there still a minor edits tickbox? - in Preferences, that is

I couldn't find one but might have missed it. I ask because I sometimes see new editors marking content edits as minor, and before warning them I want to make sure I'm right. Thanks. Doug Weller talk 11:17, 7 February 2020 (UTC)

It's still below the edit summary box to the left of "Watch this page". Do you see it here? PrimeHunter (talk) 12:01, 7 February 2020 (UTC)

@PrimeHunter: I'm an idiot, I meant in Preferences. I've edited the section heading. Doug Weller talk 18:00, 7 February 2020 (UTC)

@Doug Weller: No, it was hidden for English Wikipedia in 2011. --Izno (talk) 18:10, 7 February 2020 (UTC)

(In fact, you participated in support of removing it at the related RFC. --Izno (talk) 18:11, 7 February 2020 (UTC))

@Izno: I have a vague memory of that. User:Bishonen and I were discussing it recently and I'm sure Bish will be glad to know that also. So when I see new editors marking major content edits as minor from the getgo, I'll be sure to kindly tell them it's not a good idea. Strange how often those edits are problematic. Doug Weller talk 19:01, 7 February 2020 (UTC)

We do have {{subst:uw-minor}}. --Redrose64 🌹 (talk) 22:41, 7 February 2020 (UTC)

Extended content
jQuery(document).ready(function($) { if(mw.config.get('wgAction')==='edit') { document.getElementById('wpMinoredit').checked=true; } }); If anyone needs/wants to, this should work? Use at your discretion. --qedk (t 桜 c) 18:34, 7 February 2020 (UTC)

So new users who mark every edit as minor must be employing a custom Javascript? And as a corollary, IP editors aren't able to mark edits as minor since they can't customize any scripts? EdJohnston (talk) 18:49, 7 February 2020 (UTC)

No, users who want to have edits marked minor by default must use javascript; the second part doesn't apply because the "minor edit" checkbox is still there on the editor. They just have to mark them minor intentionally. No one should be marking edits as minor by default, and I actually would remove the "minor edit" thing entirely if it was up to me.--Jorm (talk) 19:03, 7 February 2020 (UTC)

Thanks for the explanation. A new editor, or an IP, who is marking everything 'minor' is presumably checking the 'minor' box before every edit. Unless they are on Commons, on the French Wikipedia or some other Wikimedia site that still supports the setting of 'Minor' by default as an option in their editing Preferences screen. There is a slight chance that a new editor might be using a script to automatically mark all their edits as minor. If that causes a problem, their scripts are visible and can be scrutinized. EdJohnston (talk) 01:55, 8 February 2020 (UTC)

I did wonder whether a global preference might cause this. So, I went to fr:Spécial:GlobalPreferences#mw-prefsection-editing where I found the entry "Considérer mes modifications comme mineures par défaut"; in that, I enabled the two checkboxes at the start of the line and saved. I then went to a page on French Wikipedia and clicked an edit link, where I found that the "Modification mineure" checkbox was set. Then I went to a page on English Wikipedia and clicked an edit link, where I found that the "This is a minor edit" checkbox was not set. So, if there is no local preference available, a global preference can't force an override. --Redrose64 🌹 (talk) 10:54, 8 February 2020 (UTC)

. Logged in users on enwiki have two checkboxes "wpMinoredit" and "wpWatchthis" when editing, while IPs have none. They are completely absent from the html of anon edit-forms, not merely "hidden" by css. Also, "minoredit" is enumerated here as a separate user-right, which implies a wiki's LocalSettings.php file could be configured to grant "minoredit" to IPs, or limit it to some subset of logged in users, or disable it for everyone. So different rules may apply on other wiki sites. ―cobaltcigs 11:19, 8 February 2020 (UTC)

Special:RandomInCategory with action=edit

Tracked in Phabricator
Task T244641
Resolved

I have, in my web browser, a frequently used bookmark with the url:

• https://en.wikipedia.org/w/index.php?title=Special:Random&action=edit

I don't know if this functionality is widely known, but it does successfully redirect an article title determined by Special:Random and opens it in edit mode (saving one click). Further testing reveals that other actions such as history, watch, etc. also succeed.

When I became aware of Special:RandomInCategory (see example), I tried to create an action=edit bookmark for that as well:

• https://en.wikipedia.org/w/index.php?title=Special:RandomInCategory/Articles_with_short_description_added_by_PearBOT_5&action=edit

Unfortunately the latter Special: page ignores the specified action. Any idea why? ―cobaltcigs 03:44, 8 February 2020 (UTC)

It does work for me with [4], perhaps this is some kind of bug? Jo-Jo Eumerus (talk) 09:15, 8 February 2020 (UTC)

It chooses a random article from that category and opens it in (the default) action=view mode. For example, I just clicked it and got redirected to the url https://en.wikipedia.org/wiki/Corbin_Allred with no editing box.

This differs from the behavior of the regular Special:Random urls, which do "remember" the specified action in all cases.

At least for me, using the exact same browser, etc. So when you say it "does work" for you… ―cobaltcigs 10:38, 8 February 2020 (UTC)

It drops any query string. Compare for example https://en.wikipedia.org/w/index.php?title=Special:Random&useskin=modern&uselang=de to https://en.wikipedia.org/w/index.php?title=Special:RandomInCategory/Articles_with_short_description_added_by_PearBOT_5&useskin=modern&uselang=de. https://phabricator.wikimedia.org/rSVN65054 from 2010 is "Special:Random now carries over query string parameters". You want the same for Special:RandomInCategory. You could file a Phabricator request. I didn't find one. PrimeHunter (talk) 12:53, 8 February 2020 (UTC)

As the creator of PearBOT 5 I'm a bit curious what you want to do. If you have any feedback feel free to tell me and I'll do my best to improve it. The bot still has many tens of thousand edits left so any little improvement will likely have a significant impact. ‑‑Trialpears (talk) 14:02, 8 February 2020 (UTC)

Manually review a handful of your short descriptions each day without exhausting any particular section of the alphabet, or bumping into other users. As for feedback: if I were running that kind of bot I'd just form descriptions by dropping "s" or "es" from end of the "most important" category name (according to some proprietary ranking), rather than trying to parse any prose from the lead paragraph. Because those tend to be inconsistent and polluted with irrelevant details. ―cobaltcigs 14:31, 8 February 2020 (UTC)

Anyway, ticket opened: phab:T244641. ―cobaltcigs 14:31, 8 February 2020 (UTC)

What's with images becoming unclickable?

Arising from an edit war over an article image that was reported at AN/I, I noticed that the image had an "i" in the bottom corner and tried to go to the image page to discover whose watermark this was, only to find the image is unclickable and the "i" is apparently the new way of actually going to its page. I assume this isn't some intrusive new feature of my browser but rather yet another intrusive new feature snuck in by the WMF? I don't like the defacement and I don't like the reduction in intuitiveness. Was there some place we were informed that this would be happening? ... and more importantly is there a way to turn it off in my preferences? Yngvadottir (talk) 17:57, 8 February 2020 (UTC)

Could you provide a link to the image and/or page in question? -- RoySmith (talk) 18:03, 8 February 2020 (UTC)

Was about to ask the same question. I was thinking it was Wikipedia:Administrators'_noticeboard/Incidents#User:Cambial Yellowing but that one works correctly for me. Jo-Jo Eumerus (talk) 18:04, 8 February 2020 (UTC)

The article is Yeouido. Is the infobox generated from Wikidata and that's the problem?? I haven't happened to do any edits recently involving images, so I honestly don't know how new this is. Yngvadottir (talk) 18:07, 8 February 2020 (UTC)

Ah. Thanks for the link. Looking at the source, I see the image includes <imagemap> File:Yeouido Skylines - 201912211335.jpg|340px|</imagemap>. Wow, imagemap? I haven't seen a imagemap used since the paleozoic era. No clue why the author did that, but I'm sure it's what's causing the issue. You own the WMF a healthy dose of WP:AGF. -- RoySmith (talk) 18:14, 8 February 2020 (UTC)

The imagemap was added in Special:Diff/816169558 -- RoySmith (talk) 18:18, 8 February 2020 (UTC)

Looks like we've got over 900 imagemaps right now. Mind blown. -- RoySmith (talk) 18:21, 8 February 2020 (UTC)

(edit conflict) Yes, I've just tried editing the article and discovered the HTML command. Thanks, RoySmith, for tracking it down faster and for verifying my hunch that we don't tend to use that. I'm afraid my assumption that if it wasn't my browser, it was the WMF, is all too natural; to name just one supposedly well-intentioned change, I still get tension headaches when I have to look at the Media Viewer, though they're less violent than when it was originally served up. Can the rest of the imagemaps be removed, please, unless the image actually has defined links within different parts of it? Yngvadottir (talk) 18:27, 8 February 2020 (UTC)

I think it unlikely they will be removed. I didn't know about this before today, but apparently it's an officially supported feature of MediaWiki. See mw:Extension:ImageMap. -- RoySmith (talk) 19:03, 8 February 2020 (UTC)

In addition most of the uses are appropriate with only 24 looking like they are without any links. ‑‑Trialpears (talk) 19:28, 8 February 2020 (UTC)

So, only 24 don't seem to go to different parts of a mosaic image, or such - can those be removed as obfuscatory clutter? Yngvadottir (talk) 19:34, 8 February 2020 (UTC)

Sure, I don't have time right now but feel free to replace them with normal syntax if you want. Don't trust my search too much though and make sure it actually doesn't do anything first. ‑‑Trialpears (talk) 19:44, 8 February 2020 (UTC)

Imagemaps are not a new feature, by a long way. They have been mentioned on this page several times in the past; and indeed, in one thread above. In fact, they are mentioned in this thread which dates from eight days after I registered. So in my personal experience, they've always been around. --Redrose64 🌹 (talk) 21:45, 8 February 2020 (UTC)

We also have a category system for imagemaps: Category:Wikipedia image maps. I made the poorly populated Category:Articles containing image maps in 2014. PrimeHunter (talk) 22:09, 8 February 2020 (UTC)

Need a list of all articles with multiple stub templates

Is there a way, any way, to get a list of all articles that have more than one stub template on them? Example: 1 Decembrie 1918 metro station. SD0001 (talk) 21:03, 8 February 2020 (UTC)

SD0001, if you're willing to torture the search function this kind of works. You can restrict the search further to avoid the timeout, for example by first letter in the pagename. ‑‑Trialpears (talk) 21:09, 8 February 2020 (UTC)

@Trialpears: Thanks! That search query works with a prefix qualifier. But it's worth noting there seems to something funny going on in the search engine: A search for:

incategory:"All stub articles" prefix:"1" insource:/stub\}.*stub\}/

gives no results, but a search for:

incategory:"All stub articles" insource:/stub\}.*stub\}/ prefix:"1"

works (just the order of parameters changed). Any idea why? SD0001 (talk) 21:41, 8 February 2020 (UTC)

@SD0001: Help:Searching#prefix: says: "prefix: must only ever be given at the last part of a search box query". PrimeHunter (talk) 21:55, 8 February 2020 (UTC)

Ahh, gotcha. MediaWiki is fun! SD0001 (talk) 22:06, 8 February 2020 (UTC)

It's not a crime for an article to have more than one stub template, see WP:TAGSTUB, paragraph beginning "If an article overlaps several stub categories". --Redrose64 🌹 (talk) 21:50, 8 February 2020 (UTC)

For sure. What makes you think I want to remove the multiple stub templates? SD0001 (talk) 22:06, 8 February 2020 (UTC)

Indeed, they should be merged into a single template that accepts multiple parameters. ―cobaltcigs 00:31, 9 February 2020 (UTC)

This has been discussed before and rejected, see e.g. Wikipedia talk:Manual of Style/Layout/Archive 10#Proposal: Only one stub template per article. I think also at WP:VPR. --Redrose64 🌹 (talk) 10:52, 9 February 2020 (UTC)

Talk page edits not appearing

I've been attempting to edit Talk:Shirley Ann Jackson, but none of my changes are visible, though the edits are present in the edit history. I've checked in a few different browsers. Any ideas? Zagalejo^^^ 04:25, 9 February 2020 (UTC)

In fact, there are multiple other sections of that talk page that aren't being displayed. I don't immediately see where the problem is. Zagalejo^^^ 04:27, 9 February 2020 (UTC)

There was an improperly closed "hidden comment" tag at the end of the last section that was visible, meaning that every single thing in the entire rest of the page was being swallowed by it and treated as more hidden comment. I've removed it, and the page is now behaving properly. That's almost always the answer when something like this is happening; it's almost always because there's an improperly closed coding tag (hidden comment, gallery, reference, nowiki, etc.) somewhere in the article that's breaking the page. Bearcat (talk) 04:34, 9 February 2020 (UTC)

Thanks for your help! Zagalejo^^^ 14:42, 9 February 2020 (UTC)

template/lua to determine if X is in category Y

Just wondering if there is a way to get a boolean answer to the question "is article X in category Y?", using template markup or Lua. I did search for this before asking! Outriggr (talk) 08:58, 9 February 2020 (UTC)

I think the category toolbox extension would do what you want in Lua, but I don't know if it is installed on the English Wikipedia. Another way would be to use the title object and search the contents for the category (it wouldn't get parent categories though).   Jts1882 | talk  09:37, 9 February 2020 (UTC)

I think it's not installed as mw.ext.cattools.hasPage('Category:Florida', 'Florida') does not work. --Gonnym (talk) 09:47, 9 February 2020 (UTC)

Right. Special:Version shows installed extensions. mw:Extension:CategoryToolbox is not installed. PrimeHunter (talk) 13:08, 9 February 2020 (UTC)

Thanks. From the above, I gather that what I ought to do is get the page contents using Module:Page (which I just found by guessing names) and use GetContent(), e.g. {{#invoke:Page|getContent|page=Foo}} and go from there. I've now got #invoke:String|count to work with the results of getContent(), by previewing in my sandbox, so that's cool. I would think this task is general-purpose enough to be generalized into a template but am not aware of one.... Outriggr (talk) 10:42, 9 February 2020 (UTC)

Please don't. First, there are other ways for a category to be added to an article other than plain wikitext (from templates). Second, the overhead of reading and parsing wikitext to find a category string is not worth it. Some JavaScript could probably use the API to get accurate information, if there were a demonstrated need. Johnuniq (talk) 22:18, 9 February 2020 (UTC)

Template:Party name with color

Can someone proficient in template coding take a look at my unanswered edit request at Template talk:Party name with color#Piped links, namely add a {{{2|}}} somewhere so that this supports piped links when used with full. For instance, {{Party name with color|Foo Party|German Foo Party|full=yes}} would become [[Foo Party|German Foo Party]]. – Finnusertop (talk ⋅ contribs) 12:27, 9 February 2020 (UTC)

I figured out how DAB works and added it to Template:Party name with color/doc for Constitution Party (United States). For what you really want I'd need a party with sub-pages /meta/color (ideally) + /meta/shortname (required) for a decent full example on the /doc page, the functionality might already exist. –84.46.52.123 (talk) 09:35, 10 February 2020 (UTC)

Using your example: {{Party name with color|Constitution Party (United States)|U.S. Taxpayers' Party|full=yes}} doesn't show up as [[Constitution Party (United States)|U.S. Taxpayers' Party]] but [[Constitution Party (United States)|Constitution Party]]. – Finnusertop (talk ⋅ contribs) 17:19, 11 February 2020 (UTC)

Yes, there is no 2nd parameter as you suggested it. But {{Party name with color|Constitution Party (United States)|full=yes}} should have an effect for {{Constitution Party (United States)/meta/shortname}} = Template:Constitution Party (United States)/meta/shortname.
Without "full" I'd expect "Constitution", with "full=yes" I'd expect "Constitution Party". I've extended the examples to show that something is wrong. If it would work (it doesn't) you could get a /meta/shortname Foo Party for a full name German Foo Party. –84.46.52.187 (talk) 13:42, 12 February 2020 (UTC)

But that's the thing. I don't want to alter the shortnames every time someone uses an ad hoc name in a table. I don't want to bypass the piped linked either, so as to preserve whatever semantic difference there is. That's why I'm suggesting the second unnamed parameter. – Finnusertop (talk ⋅ contribs) 16:00, 12 February 2020 (UTC)

If you copy the template as is to a sub-page /sandbox {{Party name with color/sandbox}} and the /doc as is to a sub-page /testcases {{Party name with color/testcases}} not logged-in users could help to figure it out, i.e., the /testcases should be limited to the 5 examples in /doc (nothing else) and use the /sandbox instead of the real thing. But doesn't enwiki have a "template workshop" or project, is VP/T supposed to be that place? –84.46.52.187 (talk) 17:03, 12 February 2020 (UTC)

Template:Expand Arabic and Category:Articles needing translation from Arabic Wikipedia

Forgive me if this is wrong the place to ask about this. Category:Articles needing translation from Arabic Wikipedia contained over 200 articles in January, now it is down to 32. It's not that the translations are complete, because articles such as Al-Faw and Abyan Governorate still need to be translated but have dropped out of the category. I suspect something may be wrong with Template:Expand Arabic, which is supposed to place articles in the translation category. Any ideas? --Cerebellum (talk) 13:29, 9 February 2020 (UTC)

Cerebellum, I'm probably responsible and did some major changes to {{expand language}} recently. I thought everything was alright based on my checks on {{expand Spanish}}, but seems like that's not the case. I'm currently away from my computer but will take a look at it tonight. ‑‑Trialpears (talk) 13:50, 9 February 2020 (UTC)

It's now dealt with. Sorry for the inconvenience. I'm on holiday and had a quite spotty internet connection these last few days so it took some more time than expected. ‑‑Trialpears (talk) 18:48, 11 February 2020 (UTC)

Many auto edits and account risks

Sorry for asking here!

Hello, I'm from ckbwiki, I want to say... our project is so small and we have no bot to help us in some aspects. So, i want to use many auto software programs like AutoWikiBrowser, Huggle, WPCleaner (if you use another good one, please suggest to me) and many user scripts. For example: May i do hundreds or thousands of auto edits by using AutoWikiBrowser at the same time, then is there any risk to my account? Thanks! ⇒ AramTalk 13:49, 9 February 2020 (UTC)

@Aram: tools like AutoWikiBrowser generally respect maxlag, so your "account" should be OK from a global perspective. Most projects have guidelines or rules about making lots of edits quickly, as it can make patrolling changes hard for others - but that is something the community at ckbwiki would need to decide. When communities are concerned with this, it is normally recommended to make fast repetitive edits with a dedicated account that can be flagged as a "bot", w:ckb:ویکیپیدیا:بۆت has a little bit of information on this. — xaosflux ^Talk 15:46, 9 February 2020 (UTC)

@Xaosflux: Thanks for your replying! But we haven't any skill about programming languages to creating a bot. So, I want to use AWB, and/or the others. So we have two options. First, i must use auto programs by logging in with my account (Aram). Second, i must create another account (not a bot) especially for these edits. Which one is your choice? Thanks! ⇒ AramTalk 18:42, 9 February 2020 (UTC)

@Aram: ckbwiki already has many bots (about 20) - so I suggest you follow your community norms about when and how to use a bot. — xaosflux ^Talk 18:45, 9 February 2020 (UTC)

@Xaosflux: You are right, but most of them are inactive. just i see IAB, AlaaBot and another one. However, THANKS! ⇒ AramTalk 18:59, 9 February 2020 (UTC)

@Aram: to put it simply, the only people that are likely to complain about you editing quickly and repetitively at ckbwiki are the editors of ckbwiki. In general, if you want to do something over and over that really doesn't need anyone to check it (such as updating templates, fixing layouts, etc) I recommend you make a bot account and have it flagged. If you edits will be "content" related (i.e. adding or updating facts to your articles) using a bot flag may be a bad idea, since it will the edits them from people that are not specifically looking for them. — xaosflux ^Talk 19:42, 9 February 2020 (UTC)

@Xaosflux: Don't worry! I'm a very good and active user at ckbwki and i don't think so that a user (on ckbwiki) want to complain to me. I want to use AWB and others to add/delete templates, categoies and files to/from articles, replace old parameter names with the new ones and some other actions quickly on articles. Is a bot's interface like AWB? Because of i have no skills that required to creating a bot. I meant programming languages. How can i see a bot's interface. Thank you for your all answers! ⇒ AramTalk 21:07, 10 February 2020 (UTC)

Programming help - US Census links going dark

Discussion moved to Wikipedia:US Census Migration for easier editing and reference. -- RoySmith (talk) 15:35, 10 February 2020 (UTC)

Automatic archiving

There's something that's wrong with automatic archiving on the page Wikipedia talk:WikiProject U.S. Congress that causes it to work sometimes and leave discussions there for much longer than 180 days sometimes. I'm not really sure what's going wrong with User:Lowercase sigmabot III there. Any ideas as to a fix? Airbornemihir (talk) 20:02, 9 February 2020 (UTC)

@Airbornemihir: "minthreadstoarchive" and "minthreadsleft" are not set, so they assume their default values of 2 and 5, as described at User:Lowercase sigmabot III/Archive HowTo. ST47 (talk) 21:21, 9 February 2020 (UTC)

@ST47: ahhh I see. Thanks! Airbornemihir (talk) 21:57, 9 February 2020 (UTC)

Edit history stats

Hi, is there a place to get page history stats plotted on a graph or in some tabular form? Edits per day, edits per day per user, etc. I could scrape the history and feed it into Splunk but surely this exists some place. Any pointers? --LaserLegs (talk) 00:14, 10 February 2020 (UTC)

@LaserLegs: I think the "Page statistics" link in any article's history page is probably what you're looking for. --Yair rand (talk) 06:55, 10 February 2020 (UTC)

User search by last activity

Is it possible to search for, say, transclusions of {{Semi-retired}} on the user talk pages of editors who have not edited since, say 2018? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:31, 10 February 2020 (UTC)

If you don't mind that it doesn't see deleted or revdelled edits, or logged actions that don't create a revision, quarry:query/42052. —Cryptic 13:53, 10 February 2020 (UTC)

Cite web parameter accessdate

The cite web parameter |accessdate= seems to allow future dates. Why is that? Thanks. Martinevans123 (talk) 15:01, 10 February 2020 (UTC)

Because the world turns and today in one part of the world is already tomorrow in another part of the world. |access-date= should only allow for today plus one day into the future. Are you seeing something different?

Also, questions about the cs1|2 templates are best addressed at Help talk:Citation Style 1.

—Trappist the monk (talk) 15:13, 10 February 2020 (UTC)

Really, it turns!? You must be one of these trendy modern editors. Maybe I'm just too UTC-centric? No, it was just a tomorrow date, thanks. Martinevans123 (talk) 16:05, 10 February 2020 (UTC)

The only mythology here is that somehow the Mediawiki software in incapable of presenting correct time zone information, which leads to bad module writing and impossible citation dates. This is not true. 72.43.99.138 (talk) 14:16, 11 February 2020 (UTC)

Tech News: 2020-07

Latest tech news from the Wikimedia technical community. Please tell other users about these changes. Not all changes will affect you. Translations are available.

Recent changes

• There is a new version of the Wikimedia Commons app for Android. It should fix the failed uploads problem. [5]

Problems

• There was a problem with the new MediaWiki version last week. It deleted some messages by accident. The new version was late because it was stopped to fix things. [6]

Changes later this week

• The MediaWiki action API is used by various tools like bots and gadgets. Some error codes will change. Some parameter values that do not follow the standard will no longer work. [7]
• The new version of MediaWiki will be on test wikis and MediaWiki.org from 11 February. It will be on non-Wikipedia wikis and some Wikipedias from 12 February. It will be on all wikis from 13 February (calendar).

Tech news prepared by Tech News writers and posted by bot • Contribute • Translate • Get help • Give feedback • Subscribe or unsubscribe.

19:09, 10 February 2020 (UTC)

Can't preview

I can't get preview or "show changes" to work. The error message is "Our servers are currently under maintenance ...". Is anyone else experiencing this? SarahSV ^(talk) 05:21, 11 February 2020 (UTC)

SarahSV, I've seen that a fair number of times in the past two weeks. To add to the frustration, it's most frequent when I've a slow or unreliable connection. On occasion preview has worked in Safari when Firefox (which I usually use) gave me nothing but errors; perhaps changing browsers will work for you as well. BlackcurrantTea (talk) 11:18, 11 February 2020 (UTC)

I was getting this recently when looking at my watchlist, which was large. The message seems to be a generic one when a transaction times out because it has taken more than a set period of time. I suppose that recent occurences are due to a new release or setting, like the update which caused outages across Europe. Andrew🐉(talk) 11:56, 11 February 2020 (UTC)

BlackcurrantTea and Andrew, thanks for the replies and tip about other browsers. I did try with others, and it worked, so the problems seemed confined to Firefox. A few minutes ago, it didn't work with another browser either. I lost an edit because I thought it had saved, but I got an error message. The back button and "try again" didn't work. SarahSV ^(talk) 21:29, 11 February 2020 (UTC)

@SarahSV: I don't think it is browser related; that's probably just bad or good luck depending on factors outside your control which made it appear that one browser functioned better than another. There have been reports here of time-outs and related problems for a couple of weeks and someone linked to a WMF blog or somesuch saying a massive denial-of-service attack on Wikipedia was impacting users in various areas. See also #Outage just posted below with "Error Our servers are currently under maintenance or experiencing a technical problem". I looked at two articles you were recently editing and they are massive with, from memory, 5 or 6 seconds of CPU time required to render the page. It would only take a bit of extra pressure on the server or the networks between servers to cause the problems you mention. The "under maintenance" message appears to be a last-resort "we don't know what the problem is but we give up" generic message that is not related to the actual problem. One way to avoid losing stuff is to do the equivalent of Ctrl-A Ctrl-C (select all, copy) before clicking preview or publish. Then, if there is a problem, you should be able to paste into a text editor to save what you had in the edit window. Johnuniq (talk) 22:55, 11 February 2020 (UTC)

Johnuniq, thanks for the information. I do usually save off-wiki and I did this time too, went to press save, then changed my mind and made more edits. That was the edit I lost. Usually, the back button or "try again" retrieves lost text, but this had disappeared entirely. I just tried preview and "show changes" on Firefox on a long article, and it's working again. SarahSV ^(talk) 00:44, 12 February 2020 (UTC)

Gadget/User Script to disable MediaWiki:Bad image list

Wikipedia is not censored and never will be. All articles fall under this disclaimer. By clicking this checkbox you'll only hideing these images from yourself (in all wikiprojects) while still open for other's.

Recently I find out this help page & became really confused. If we are giving advice because they are the half of the userbase, why can't we also provide it as a choiceable option (as a 'Gadget' or an 1 click 'User Script'). We still providing some gadget & most of the user script only because of the user demands. Well, If all of these are against 'POLICY' then a simple info template must be possible, Right ?

---

--Masum The Great (talk) 14:14, 11 February 2020 (UTC)

Related changes and talk pages

Is it possible to view related changes to the subject namespace page when only its talk page is a member of a category (for example Special:Recentchangeslinked/Category:High-importance Astronomy articles)? If so, how? Are there others who would like this to be possible? Utfor (talk) 17:25, 11 February 2020 (UTC)

There is an external website about linked from one or another of our tools pages that can do that for you at the WikiProject level. I'm not sure if there is a similar tool for categories... --Izno (talk) 17:31, 11 February 2020 (UTC)

Could someone please tell me where I can find this? I tried to find it at Wikipedia:Tools, but perhaps I overlooked it. Utfor (talk) 17:56, 11 February 2020 (UTC)

How to search edit summaries

Hi, I posted first on WP:Help desk but got no responses and another editor suggested this board. Is there a way to search edit summaries across Wikipedia? I came across several edits I reverted that said Eklenti yapıldı (plugin done, in Turkish) in the edit summary, and I wanted to check out any similar edits but couldn't figure out how. Schazjmd (talk) 19:46, 11 February 2020 (UTC)

@Schazjmd: you can use a tool such as ths one to search edit summaries. (It isn't very fast). — xaosflux ^Talk 20:59, 11 February 2020 (UTC)

Perfect, thank you Xaosflux! Schazjmd (talk) 21:03, 11 February 2020 (UTC)

Edit request for module and complex template

{{in lang}} has replaced a few hundred individual templates that signified when an external link was in a particular language which is great. What's less great is that now the hundreds of subcategories at Category:Articles with non-English-language external links have all been emptied. Can someone help me fix this? The syntax here is too complicated for me: I can do templates and some Lua but I'm not a programmer. @UnitedStatesian: who started a relevant discussion. ―Justin (koavf)❤T☮C☺M☯ 19:50, 11 February 2020 (UTC)

@Koavf: {{in lang}} does not need to be changed; see the discussion at Template_talk:Link_language#Auto-categories. The empty categories can all be (slowly, as at present, or quickly) deleted as deprecated, as there is a new category structure, populated by the template, that has replaced them. UnitedStatesian (talk) 19:56, 11 February 2020 (UTC)

UnitedStatesian, This is very confusing. Why on Earth did someone make a parallel category system? Why is it named "[x] sources" when they are links and not sources? This seems like a perfect example of needless bureaucracy and busywork when we had a perfectly functional system. ―Justin (koavf)❤T☮C☺M☯ 20:21, 11 February 2020 (UTC)

Pinging Trappist the monk as the creator. ‑‑Trialpears (talk) 20:23, 11 February 2020 (UTC)

Why is it named "[x] sources" when they are links and not sources? It cannot be presumed that article content associated with the {{link language}}, {{<xx> icon}}, and {{<xx>}} templates and now associated with {{in lang|<xx>}} is or ever was a link. Editors have not always used these templates solely with links. The 'sources' naming scheme recognizes that editors at en.wiki use these templates as it suits them regardless of whether the content is plain text, an external link, or even an interwiki link to another language Wikipedia.

—Trappist the monk (talk) 23:02, 11 February 2020 (UTC)

A week ago, I started a discussion trying to resolve the duplicate category trees between {{link language}} and {{in lang}} by redirecting the former to the latter. If successful the old categories could be tagged with {{Db-templatecat}} and deleted. ‑‑Trialpears (talk) 20:05, 11 February 2020 (UTC)

Persistence of check marks when comparing selected revisions

When I mark two versions of an article, usually encompassing those that I haven't read since my last visit, in preparation for clicking on the "Compare revision history" button on an article's "Revision history" page, I'm finding that, after reading the rendered diff and clicking on the back-arrow to come back to the history page, the circles checked don't match what I selected, the two top ones being checked instead. This is intermittent and I can't reliably reproduce it. Does anyone else have this problem? Dhtwiki (talk) 20:08, 11 February 2020 (UTC)

Outage

Error Our servers are currently under maintenance or experiencing a technical problem. Please try again in a few minutes. See the error message at the bottom of this page for more information. If you report this error to the Wikimedia System Administrators, please include the details below. Request from 98.21.227.217 via cp1089 frontend, Varnish XID 898433163 Error: 503, Backend fetch failed at Tue, 11 Feb 2020 21:13:05 GMT

This has been going on for quite some time and I finally got back in.— Vchimpanzee • talk • contributions • 21:34, 11 February 2020 (UTC)

Watchlist format

Hello, I have managed to change my watchlist format by doing something but am unable to restore it to the previous format can some one tell me what I need to do to get back to previous format.

I now get -

  b   01:44 	Anne Hutchinson‎ (diff | hist) (+63‎) InternetArchiveBot (talk | contribs block) (Bluelink 2 books for verifiability. gra) #IABot (v2.0) (GreenC bot) [rollback]
   m    01:44 	RAF Hemswell‎ (diff | hist) (0‎) Neils51 (talk | contribs | block) (→‎After closure: spelling - it's->its) [rollback] (Tag: AutoWikiBrowser)

What I want is what I still get on Commons -

(diff) (hist) File:Blacktoft Sands RSPB reserve with Ship.JPG‎ 14:10 (+410)‎ ‎Pigsonthewing (talk | contribs)‎ ‎Created claim: depicts (d:P180): (d:Q11446)

Keith D (talk) 02:05, 12 February 2020 (UTC)

@Keith D: Do you mean the b (bot) entry? That is a simple option in preferences. The simplest would be to compare Special:Preferences#mw-prefsection-watchlist with commons:Special:Preferences#mw-prefsection-watchlist. The latter apparently is what you want, and the former might be different. Johnuniq (talk) 02:44, 12 February 2020 (UTC)

No it is the actual layout of each entry that changed, possibly after I use Related changes. The time has moved to the front rather than the (diff) (hist) and you get a bunch of changes that you can toggle which is not what I want. I have compared preferences under watchlist and they appear to be the same on both sites. Keith D (talk) 11:42, 12 February 2020 (UTC)

It's "Group changes by page in recent changes and watchlist" at Special:Preferences#mw-prefsection-rc. PrimeHunter (talk) 12:10, 12 February 2020 (UTC)

Many thanks for that. Unchecking resets it to what it was, do not know why preferences were changed by the Recent changes form which should not be modifying preferences. By the way that preference is not available on Commons. Keith D (talk) 17:02, 12 February 2020 (UTC)

Incorrect anchor in Watchlist legend

At Special:Watchlist on desktop there is a box on the right side with a list of abbreviations used on the page. It also links to Help:Watchlist in parenthesis, but it links to an old name of a section, #How to read a watchlist instead of #How to read a watchlist (or Recent Changes). Could somebody fix this? I guess that an interface admin is needed for this, thus @Amorymeltzer:. —⁠andrybak (talk) 12:11, 12 February 2020 (UTC)

Any sysop could make that change, but I happened to be drinking some tea and reading the news, so I took care of it! ~ Amory (u • t • c) 12:29, 12 February 2020 (UTC)

Rendering slow, scripts and notifications in lag

This morning many things are slow to the point of being very annoying and preventing me from doing my work. Examples are (1) After clicking on a page, I have to refresh the screen as it doesn't fully render, (2) scripts run slowly, sometimes so slowly as to be worthless, (3) notifications are in lag - I got a red 1 on my notifications at the top and when I clicked on it, it said I no notifications...a few minutes later I was finally able to click on the notification that had indeed been made earlier, and (4) my interface at the top is rarely complete without being refreshed, sometimes more than once.--Bbb23 (talk) 15:32, 12 February 2020 (UTC)

Need help with a simple script

I'm trying to make a script that simply highlights a string, but I utterly suck at javascript. Basically, if you have something like

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

And I have regex like

(tempor|labore|dolore)

then the above would be rendered as

Lorem ipsum dolor sit amet, consectetur adipiscing elit, sed do eiusmod tempor incididunt ut labore et dolore magna aliqua. Ut enim ad minim veniam, quis nostrud exercitation ullamco laboris nisi ut aliquip ex ea commodo consequat. Duis aute irure dolor in reprehenderit in voluptate velit esse cillum dolore eu fugiat nulla pariatur. Excepteur sint occaecat cupidatat non proident, sunt in culpa qui officia deserunt mollit anim id est laborum.

See also this scriptreq. Headbomb {t · c · p · b} 16:43, 12 February 2020 (UTC)

This is not done "simply". To target a piece of text, but not surrounding text, they need to be in different DOM nodes. To accomplish this, you would have to molest the HTML of the page somehow, perhaps by wrapping each occurence in a span tag with a certain class. I can't say beforehand what the consequences of doing that is, with regard to performance, or other systems expecting a certain HTML structure. Also, the advice you are given at Wikipedia:User scripts/Requests#unreliable.js, a predatory journals/vanity press highlighter is incorrect.

( function() {
	//...
} () );

is an immediately invoked function expression. It does not wait until the page has loaded. It is

$( function() {
	//...
} );

you want. It would be remarkable if the IIFE waited for the page to load, given that it is not just JavaScript, but pure EcmaScript and therefore has no knowledge of – or dependency on – the DOM. If things have worked for you despite that, it has probably worked by accident because the code is not directly in your personal JS page, but on a separate page that you import with importScript. That network request might take several hundred milliseconds, which may or may not be sufficient for the page to load before the script is executed. Nirmos (talk) 18:22, 12 February 2020 (UTC)

@Nirmos: I'll try to update things. It was working probably 75% of the time, with a reload often being enough to get it to work when it didn't. Headbomb {t · c · p · b} 18:54, 12 February 2020 (UTC)

Actually User:Headbomb/unreliable.js was already up to snuff there. Reloads just cleared cached pages I had already visited. Headbomb {t · c · p · b} 18:56, 12 February 2020 (UTC)

I can't see that User:Headbomb/unreliable.js is "up to snuff". Nowhere in the code do I see

$( function() {
	//...
} );

or any equivalent signature. This might be a simplification, but the "thing" immediately after $ or jQuery and its ( can either be a string or a function. When it's a string, it usually means "do something with the element(s) that match(es) this selector". When it's a function, it means "run this function after the page has loaded". I am assuming here that you are confusing the two and think that your

$('.mw-parser-output a.external').each(function() {
	//...
});

waits for the page to load. It does not. Nirmos (talk) 19:26, 12 February 2020 (UTC)

Interwiki "Your notices" seem to be persistent

In recent days, my "Your notices" icon (next to "Your alerts") has had a persistent "82" flag. On clicking, they come from Wikisource (and I'll try to find somewhere to ask the same question over there) and reflect a recent spate of hooks from WS articles I once "created" into Wikidata. Trying to expand the dropdown under "More notices from another wiki" and click the blue dot doesn't reduce the count. Logging in to Wikisource, I blue-dot each individual notification, but on reloading they come right back. I gave this a week after noticing it in case of a database delay. Am I missing something? David Brooks (talk) 18:02, 12 February 2020 (UTC)

For "somewhere" I suggest Meta, they manage all kinds of global stuff including messaging, notices, default profile, subscriptions, blacklist, OAUTH, etc. –84.46.52.187 (talk) 18:13, 12 February 2020 (UTC)

@DavidBrooks: try to clear them using the controls at Special:Notifications. — xaosflux ^Talk 18:32, 12 February 2020 (UTC)

@Xaosflux and 84.46.52.187: I was trying to clear them there (both by a group click and each individual blue dot) but to no avail. However, I then thought about Notifications on my Preferences page on Wikisource and noticed "Connection with Wikidata" was checked. I unchecked and they disappeared. Which is nice, but still obscures the underlying bug of failure to clear. I'll try to find the right place on Meta, and maybe search phabricator. David Brooks (talk) 19:24, 12 February 2020 (UTC)