Wikipedia:Village pump (technical)/Archive 152

Special:DeletedContributions has gone all facebooky, with overlarge entry windows and buttons, and too much blank space between items. This is in MonoBook skin, maybe it was a change that was intended for Vector. --Redrose64 (talk) 23:49, 9 December 2016 (UTC)

Copy of my reply at Wikipedia:Help desk#Deleted contributions:

Large interface boxes like this happen when MediaWiki features are converted to use mw:OOUI. I think MediaWiki's OOUI looks bad. Place the below in your CSS to hide it. You lose the namespace selector for deleted contributions. Once something is converted to OOUI, I don't think you can get the old look back without coding it yourself in JavaScript or something complicated like that. You could add the code to Special:MyPage/skin.css to only hide it in your current skin and still have the option to change skin to see it. PrimeHunter (talk) 13:48, 9 December 2016 (UTC)

.mw-special-DeletedContributions .mw-htmlform-ooui-wrapper {display: none;}

I just opened up Facebook, and it looks nothing like the new Special:DeletedContributions. Legoktm (talk) 01:20, 10 December 2016 (UTC)

I'm just gonna leave this here: Wikipedia:Facebookization —TheDJ (talk • contribs) 18:18, 10 December 2016 (UTC)

In this case, it's not even anything about actual Facebookization. It's just complaining about OOUI's very whitespace-heavy design by comparing it to Facebook, although I'm not seeing much similarity. Anomie⚔ 18:45, 10 December 2016 (UTC)

My point is that it's a forced change to the UI without consultation, without asking me if I wanted it, without providing an opt-out, and most importantly, without finding out if it would make things more difficult for me.

Facebook change their user interface frequently, they usually force it on you with no way of using the old version even if you knew where to find things but can do so no longer.

Periodically various interface pages in the MonoBook skin look less like MonoBook and more like something else (perhaps it's like Vector, which I don't use because of my poor eyesight), you're forced to use the New Look, and there's no way to switch it back even if your eyesight means that the new version is less accessible.

See the parallel? --Redrose64 (talk) 20:55, 10 December 2016 (UTC)

There's ongoing work to improve the density and other accessibility/usability issues in phab:T136790. I've noted (and agreed with) the substance of your concerns at phab:T134525#2862102 regarding that particular specialpage. Quiddity (talk) 21:52, 10 December 2016 (UTC)

Making Facebook UI comparisons isn't helpful, but neither is dismissing concerns about changes solely because they are expressed in such a manner. I dislike the OOUI layout because it has excessive whitespace, takes up too much room, offers no benefit as far as I can see, and no reason for it has been given, nor could I find it, and is therefore an apparently unnecessary change. Apparently it is part of a UI standardisation project, but why couldn't it have standardised on something more like what we already had? From my viewpoint the changes seem as much like whitespace faddism as anything else, given that I could find no particular reasons. And while I appreciate Quiddity's comments in favour of reducing the excesses of OOUI forms, that doesn't change my underlying reasons or opinions. BethNaught (talk) 22:27, 10 December 2016 (UTC)

XKCD is timely ;)

OOUI reminds me of New Coke, when taste tests said the new one was better but everyone wanted the old one anyway. If you just showed me the old and new interfaces in isolation, I'd say sure, new one looks better - much cleaner and more modern. But actually using the "updated" interfaces tends to be a very quick lesson in the importance of information density to a heavily text-based project. (I can't tell you how tempted I was to write this post in huge font with way too many weirdly-spaced line breaks...) Opabinia regalis (talk) 04:03, 11 December 2016 (UTC)

That comic is pertinent to the topic, but isn't an argument in favour of change for change's sake, which as I wrote, is what this seems to me as far as I am concerned. I also disagree that OOUI is intrinsically better, as it takes up ridiculous amounts of space. For your "cleaner" I say bloated, and modernity ought to be irrelevant—the question ought to be, how well does it work? I agree with you that the OOUI design is out of step with an editing community which tends to like information density. That is something UI designers seem incapable of realising without being prodded six ways from Sunday. BethNaught (talk) 10:00, 11 December 2016 (UTC)

Furthermore, when you have a 400% increase in the height of a form with two options under OOUI, how monstrous is the watchlist options panel going to look? BethNaught (talk) 10:13, 11 December 2016 (UTC)

On the two tasks I've already added some thoughts on the issue and any possibilities to address the concerns brought up here. We're taking the feedback serious, but I'd also like to take a moment to speak up on this number. 400% increase sounds really huge, no question. The challenge here is, that just because a certain old interface is well-known and works on my own computer, it doesn't mean to not be critically questioned or reluctant to changing needs for all times. Let's imagine a mobile user on his smartphone in portrait mode, where pinch-to-zoom is needed several times to accomplish the form. For this person the old state is perhaps 400% too small. Or imagine a user with motor control problems, where pointing on a form checkbox took forever. For this person the old state is perhaps 200% too small. Those are things that we're trying to take into account as well with our new forms. Glad to see your participation on Phabricator. Let's work together to make this a good experience for the majority of users and use cases. --Volker E. (WMF) (talk) 01:46, 12 December 2016 (UTC)

I think your statement makes the assumption that operating administration features on a smartphone should be made standard, when I'm not sure this is the case. By nature of the technology, it is much more efficient to use a device made to handle having multiple windows open at once, faster typing rate (generally speaking), a more precise cursor (mouse instead of finger), etc. In addition, by trying to cover both modes of design with one layout, Wikimedia does a disservice to both. Killiondude (talk) 16:26, 12 December 2016 (UTC)

@Volker E. (WMF): So it's a change to benefit users of mobile devices. That's good, provided it's kept within the bounds of mobile. Wikipedia has had several skins for some years now; they are alternative appearances which are intentionally different. When Mobile view was introduced, it was given its own skin which was again intentionally different from any of the desktop skins - and rightly so, since the available space is so much more limited. So I don't see why a "mobile" appearance is being forced upon desktop users - if I want to see what a page looks like in the mobile skin, I click the "mobile view" link at the bottom - or insert a ".m" into the URL.

If the ultimate intention is to get rid of all skins except for the mobile one, please say so now. --Redrose64 (talk) 23:56, 12 December 2016 (UTC)

Not my project, but:

I've not heard of plans to get rid of any skins, much less all skins except the mobile site (which, I gather, isn't actually a skin). I have heard devs shrug their shoulders and say that they don't really mind having four skins so long as it's understood that they're not responsible for supporting all of them. The long-term process appears to be merging Vector and mobile rather than replacing one with the other (only I suppose that desktop-only people like me would never notice when it goes the other way, would we? We would only notice when "ours" is changed).

And, finally, there probably is some good reason to reduce (but not necessarily eradicate) some of the cosmetic differences between the various options. From the POV of a reader, we probably don't want people to think that Wikipedia is two different sites – the dense gray blur that you get on a desktop and an incomplete thing that you get on a tablet. From the POV of the editor, we have a constant problem with people formatting pages on the mistaken belief that if it looks good on my computer (screen size/resolution/width/font size/whatever), then it should be good for everyone. (We see this belief in action when people hard-code the number of columns in a list, for example.) If we can reduce some of those problems by reducing some of the visual differences, then we'll probably have done some good for the world. Whatamidoing (WMF) (talk) 22:17, 13 December 2016 (UTC)

Who says they are being ignored ? Since we are getting anecdotal, that reminds me of when we once showed some total outsiders (who wondered if Wikipedia ever made visual changes at all) what Wikipedia looked like and worked 5 years ago (now 10 years). They couldn't phantom that they had ever used monobook or the OLD WikiEditor, even though they must have used it. But those people don't post on village pumps. No one is ignoring feedback (if it was actually communicated BEYOND this village pump), and the watchlist is not yet converterd. OOUI and the pages that use OOUI get follow up changes ALL THE TIME, based on user feedback just like this. Please consider the space outside of the echo chamber. —TheDJ (talk • contribs) 14:01, 11 December 2016 (UTC)

You yourself have written that Design do not appear to have taken steps to address recurrent criticisms associated with OOUI conversion. I don't see why you're bringing up showing Monobook to outsiders: yes, I use Vector since that was the default when I joined, but Monobook must have something going for it as it still has a 10% user share among very active editors. I wouldn't have minded OOUI if you could opt out of it or skin it to have the older compact layout. Also, the wikitext editor is not a valid analogy to this situation because it added new, useful features.

As for your "echo chamber", I do in fact post at Phabricator and have made a good-faith suggestion about the problem. In any case, I never said they were being ignored, my comment about "dismissing" was about yours and Legoktm's initial comment in this thread. BethNaught (talk) 14:45, 11 December 2016 (UTC)

I think I need to copy-paste what's in mw:MediaWiki UI#The controls are too big!: "One of the goals of this project is to make the site more accessible[1] and for mobile and desktop to have consistent designs. As a result, certain things may appear bigger than you are used to on desktop. Certain forms with lots of controls may take up much more space than they used to." I don't agree or disagree. I just quote it. :)Ladsgroup^overleg 14:35, 11 December 2016 (UTC)

I really agree about another part "One of the goals of this project is to build a Living style guide that developers can use as a reference to build things quickly, in many cases without needing a designer. We want to encourage the same API use and same markup. In short we want people to care about consistency more. There is a Phabricator project where you can raise concerns about the UI and help make the software better. Using standard markup will make it easier for you to override the styling and propose improved versions. We see this as a moving target. Like software, the design is never done.". Let's have consistent UI everywhere, fixing OOUI is easy. :)Ladsgroup^overleg 14:44, 11 December 2016 (UTC)

Does this mean that it is in principle possible to write user CSS or JS reducing OOUI forms to a more sensible size, or merely that form designers can override it if they wish? BethNaught (talk) 14:52, 11 December 2016 (UTC)

I'm pretty sure sizes used in OOUI can be overwritten by user common.css or mediawiki:common.css :)Ladsgroup^overleg 17:29, 11 December 2016 (UTC)

{{Infobox}} has |title= (atop box) and |above= (inside box). All fine. But today in mobile view, the title text shows downgraded, eg compared to above-text, and is left-aligned. IOW: presented misformed. See earlier talks. A bug? -DePiep (talk) 23:17, 8 December 2016 (UTC) (and when in need, I ping Edokter -DePiep (talk) 00:49, 9 December 2016 (UTC))

Possibly related discussion about Infobox television. – Jonesey95 (talk) 02:42, 9 December 2016 (UTC)

That television topic is not related. -DePiep (talk) 11:23, 10 December 2016 (UTC)

It's a bug. See the infobox title in gold, by mobile view: . -DePiep (talk) 23:50, 13 December 2016 (UTC)

I started the discussion in mediawiki talk:common.css but maybe it's not public enough. The new changes in colors just got deployed in English Wikipedia (as part of 1.29.0-wmf.5) 324549 324161 and 324534 changed:

• Search results colors
• "You've got new message" bar.
• TOC border and background.
• Border and background of thumbnails.
• Same for galleries.
• Catlinks (Categories box)
• Wikitables

Changes for Catlinks, galleries, thumbnails and TOC was changing from #f9f9f9 as background to #f8f9fa and #aaa as border to #a2a9b1. So to keep consistency in colors I think it's very important that all #f9f9f9 and #aaa in common.css should change too. (Mostly background of infoboxes) This change is almost unnoticeable (You too didn't notice it, did you?).

Beside that I have proposal to change some more stuff in Common.css to make it more aligned with Wikimedia color palette (phab:M82). This is a part of UI standardization to give users better experience. So these changes are a little bit noticeable:

• #1e90ff to #36c (The left border of ambox-notice and similar ones)
• #fbfbfb to #f8f9fa (background of mboxes)
• #b22222 to #d33 #b32424 (The left border of ambox-speedy and similar ones)
• #fee to #fee7e6 (Background of ambox-speedy and similar ones)
• #f28500 to #ac6600 (The left border of ambox-content and similar ones)
• #f4c430 to #fc3 (The left border of ambox-style and similar ones)
• #bba to #a2a9b1 (The left border of ambox-protection and similar ones)

The color palette colors are widely used everywhere. For example in OOjs-ui (so most buttons you see in the interface). Even color of the disambig icon is aligned. What do you think? :)Ladsgroup^overleg 22:29, 8 December 2016 (UTC)

Ambox-content in the Wikimedia UI colors is too dark and doesn't seem to indicate (at all?) the gradation. And the red too bright--too in your face. The others seem fine. --Izno (talk) 22:53, 8 December 2016 (UTC)

Regarding the red, We can go with #b32424 (another color from the color palette) instead of #d33. @Izno:: What do you think? :)Ladsgroup^overleg 23:16, 8 December 2016 (UTC)

That's much closer. --Izno (talk) 01:33, 9 December 2016 (UTC)

Changed in the proposal. :)Ladsgroup^overleg 02:13, 9 December 2016 (UTC)

Can someone tell me what the new color is for wikitable headers? Imzadi 1979 → 13:58, 9 December 2016 (UTC)

The header is #eaecf0 (changed from #f2f2f2) and other parts are #f8f9fa (changed from #f9f9f9) The border is #a2a9b1 (changed from #aaa) HTH. :)Ladsgroup^overleg 10:01, 10 December 2016 (UTC)

@Ladsgroup: There seem to be at least 5,000 pages using hardcoded values for these (only counting pages using the header colour). Are they fine as they are? Jc86035 (talk) Use {{re|Jc86035}}
to reply to me 11:19, 10 December 2016 (UTC)

We can make a bot to fix it. It's a piece of cake. :)Ladsgroup^overleg 11:27, 10 December 2016 (UTC)

@Ladsgroup: The old table colour #f9f9f9 is also hardcoded into infoboxes, route diagram templates, and a few hundred BSicons. I'm mostly concerned about the latter, because the first two could be fixed by modifying just a few templates/modules and CSS. Jc86035 (talk) Use {{re|Jc86035}}
to reply to me 01:59, 13 December 2016 (UTC)

@Jc86035: We already started that inconsistency by changing TOC and ctalinks backgrounds, I can fix the svg files too. It's added in my to-do list now. Is it okay? :)Ladsgroup^overleg 08:39, 13 December 2016 (UTC)

@Ladsgroup: (pinging Useddenim, Sameboat) I suppose it would be if you also modified Module:Routemap and Template:BS-map at the same time across the various wikis where they're used (if this change is over all WMF sites). Not sure really. Jc86035 (talk) Use {{re|Jc86035}}
to reply to me 12:51, 13 December 2016 (UTC)

No objection, as I can't see any difference between #f9f9f9 and #f8f9fa. (Perhaps I need to recalibrate my monitors?). Useddenim (talk) 13:04, 13 December 2016 (UTC)

Why do the SVG files used in RDTs need changing? They should all have a transparent background already; if any do not, please identify them so that they may be cleaned properly. --Redrose64 (talk) 23:50, 13 December 2016 (UTC)

@Redrose64: Masks and formations for overlaying (e.g.  (BRIDGElr) and  (MSTR3)), as well as elevated corners like  (hSTRc1) and  (ÜWt1), use the colour. Jc86035 (talk) Use {{re|Jc86035}}
to reply to me 06:51, 14 December 2016 (UTC)

If there is no objection until the next 48 hours. I will go on and request from an admin to change them (except ambox.content) :)Ladsgroup^overleg 10:01, 10 December 2016 (UTC)

Looked at the list of Ambox#types. No change for the Move color #9932cc? In the M82 palette I could not find an applicable one (and no purple at all).

Given the descriptions in M82, I'd say ambox-speedy should be #d33 'destructive action', not 'button'. Note that this is for the lefthand vertical bar only, not an area background (which would look attacking indeed). -DePiep (talk) 11:11, 10 December 2016 (UTC)

Hmm, User:Izno: What do you think? :)Ladsgroup^overleg 11:27, 10 December 2016 (UTC)

@Ladsgroup and DePiep: I think when the red that was chosen way back during the ambox design discussions, the bright-bright red was generally also disagreed to. I'm aware that semantically, destructive action is probably more appropriate, but the bright red really is going to have people cough a lung up here (even if I hadn't already opposed that specific change). --Izno (talk) 18:49, 12 December 2016 (UTC)

No objection from me. -DePiep (talk) 18:56, 12 December 2016 (UTC)

I've been getting this message when I try to edit on wikipedia:

Sorry! We could not process your edit due to a loss of session data. Please try saving your changes again. If it still does not work, try logging out and logging back in.

They first appeared on December third, and they continue to reappear whenever I try to make an edit. It's not just in English Wikipedia either. The edits I've been making are standard edits that I've been making for several years and in this particular instance it was at Croatia at the 2016 Summer Olympics. I have tried doing a test edit on my User page which worked and I have also logged out and logged back in again, which also didn't work. --MorrisIV (talk) 03:21, 14 December 2016 (UTC)

I had a bunch of this a few days ago - deleted all cookies and it cleared up. — xaosflux ^Talk 03:23, 14 December 2016 (UTC)

Wow! That was fast. Thank you for solving the problem. --MorrisIV (talk) 03:28, 14 December 2016 (UTC)

Also see phab:T151770 --AKlapper (WMF) (talk) 10:50, 14 December 2016 (UTC)

Hello. Is there a way to have the down border of a cell of a table with more "bold" line? Xaris333 (talk) 20:01, 13 December 2016 (UTC)

You can user 'border-bottom' css property of this cell. See also this. Ruslik_Zero 20:42, 13 December 2016 (UTC)

Thanks! Xaris333 (talk) 15:27, 14 December 2016 (UTC)

I was reading Help:Sorting but I didn't find a solution. How to sort ordinal form of number with one digit and with two digits? I can not use 01, 02 etc.

I can sort correctly something 2 column but I can't sort correctly something 2 column

Xaris333 (talk) 20:30, 12 December 2016 (UTC)

Like this:Tvx1 20:47, 12 December 2016 (UTC)

Yes. But is there an easiest way? Xaris333 (talk) 21:24, 12 December 2016 (UTC)

{{ntsh}} -- GreenC 21:55, 12 December 2016 (UTC)

You should prefer the "less easy" way because it produces HTML without producing a whole bunch of gibberish for re-consumers. --Izno (talk) 22:14, 12 December 2016 (UTC)

Thank you all! Xaris333 (talk) 19:59, 13 December 2016 (UTC)

Actually, it's even easier, you can use data-sort-type="number" on the table header cells. See Help:Sorting#Configuring the sorting. Matma Rex talk 21:33, 14 December 2016 (UTC)

Is it just my setup, or does the "your notices" notification take like five seconds to come up on other people's machines? --jpgordon^{𝄢𝄆 𝄐𝄇} 01:36, 8 December 2016 (UTC)

Special:Notifications often takes a long time to open for me in Firefox but not Safari. I see gray and white stripes. What web browser are you using? Whatamidoing (WMF) (talk) 23:30, 8 December 2016 (UTC)

Firefox. But I just tried it on Safari, and it behaves just the same. Chrome too. (This is MacOS, btw.) --jpgordon^{𝄢𝄆 𝄐𝄇} 15:51, 9 December 2016 (UTC)

Are you having problems loading this page today? Whatamidoing (WMF) (talk) 19:49, 14 December 2016 (UTC)

Seems noticeably faster on Safari. --jpgordon^{𝄢𝄆 𝄐𝄇} 20:50, 14 December 2016 (UTC)

Oh god yes. The process of reading notifications lately has been something like

1. See notification marker.
2. Click icon.
3. Whistle Dixie.
4. Make coffee.
5. Read the newspaper.
6. Water the plants.
7. Watch them grow for a little while.
8. Return to computer.
9. Twiddle thumbs a few times.
10. Read notification.

OK, maybe I'm exaggerating a little ;) But I'm pretty sure this wins the award for least responsive interface element of any website I regularly use. I am also using Firefox on a Mac; on Linux it's much faster. It is not faster to use Special:Notifications vs. using the dropdown, though. Opabinia regalis (talk) 21:06, 14 December 2016 (UTC)

I think you skipped the "get impatient and re-load several times in the hope that it will actually work" part.

The devs' current thinking is that this is probably due to a user script or gadget. Catrope's promised to look at log details for me, in the hope of figuring it out (assuming the script that they fixed yesterday wasn't the sole cause of my problems), but if anyone feels like blanking their scripts, re-loading everything, and reporting back, then that might be helpful. Whatamidoing (WMF) (talk) 00:42, 15 December 2016 (UTC)

The mobile skin is nice, however the editor come with it have a serious problem. Whenever I try to spend an hour or so to do some serious edit via mobile editor, those mobile browsers would remove the editing page from cache due to lack of memory. If I was editing with the regular editor, then those edit content are stored and will be automatically recovered once I went back to the editing interface. However, since the mobile editing interface utilized some dynamic element, whenever I went back to the page after some info searching, my previously typed text would all be lost. So how to disable the mobile editing interface? C933103 (talk) 10:23, 9 December 2016 (UTC)

I don't know if this will solve your actual problem, but you can switch to wikitext on the mobile site. Look for the gear-shaped settings icon next to the Save button.

The desktop wikitext editors (e.g., the 2010 mw:WikiEditor) do not save edits. Instead, your desktop web browser saves it (or doesn't). An actual auto-save system will become part of the visual editor soon. However, I don't know whether that will also work in the mobile site's visual system. Whatamidoing (WMF) (talk) 21:52, 12 December 2016 (UTC)

@Whyatamidoing (WMF): sorry for not enough specificity, what I was talking about is always the wikitext editing on the mobile site.

Yes, it is the mobile/desktop browser that save edits instead of the site/editor. However, what I am talking about is that, with the desktop mode wikitext editor, the browser can save those edits, which is not the case if I am using the mobile skin, probably because the mobile site make the editor as if some form of section on the page instead of another page.C933103 (talk) 06:55, 15 December 2016 (UTC)

Originally asked on 9 December at the Help Desk, where I got no response (except for a talk page message telling me I might have better luck here): Is there a way to set my preferences that I'm overlooking, or any other method, so that I do not automatically follow redirects to their target pages? Basically, I'd like wikipedia to treat every redirect I click as though the link has &redirect=no after it, if possible. AddWittyNameHere (talk) 22:33, 14 December 2016 (UTC)

@AddWittyNameHere: Not that I know of, but it's pretty easy to write a user script to do that. In fact, I just wrote one: User:Writ Keeper/Scripts/noAutoRedir.js. Add importScript("User:Writ Keeper/Scripts/noAutoRedir.js"); in a new line to your common.js or vector.js page, if you wanna use it. Writ Keeper ⚇♔ 22:59, 14 December 2016 (UTC)

Redirects are generated internally, unlike HTTP redirects which are handled by the browser. There is no preference to disable it. Conceivably you could write some personal JS to add redirect=no to every link on every page. --Unready (talk) 23:01, 14 December 2016 (UTC)

This appears to be exactly what Writ Keeper's script does. --Golbez (talk) 23:39, 14 December 2016 (UTC)

not *exactly*; my script identifies which links are redirects by virtue of the "mw-redirect" css class, and so it doesn't touch non-redirect links. Also, it reformulates the URL into an index.php call; for some reason, a straight /wiki/<pagename>?redirect=no fails when there's a special character (like ?) in the page name, even when that special character is appropriately URI-encoded. perhaps I should file a bug report...Writ Keeper ⚇♔ 23:43, 14 December 2016 (UTC)

Indeed, I just tried it. /wiki/%3F?redirect=no is a perfectly-formed relative URL, but ?redirect=no is ignored. I tried it on a different MW installation, which probably has different http-rewrite rules, and got the same result. It's not definitive, but it's probably MW itself, not the web server, that's throwing the query string away. Bummer. --Unready (talk) 00:44, 15 December 2016 (UTC)

Writ Keeper Thank you so much! Makes a lot of my work a fair bit easier. (It doesn't seem to work with the quicksearch, just so you know. Not an issue to me, that's not what I wanted it for anyway, but figured you might want to know). AddWittyNameHere (talk) 01:25, 15 December 2016 (UTC)

You're quite welcome. :D Writing little scripts like this is one of my more favorite things to do on Wikipedia. Writ Keeper ⚇♔ 08:19, 15 December 2016 (UTC)

During a recent session in which someone reverted a large series of edits I had made after a discussion I wasn't made aware of during the two hours in which it occurred (but that's another story), I found my pages were hanging and shut down Firefox, which I prefer to use for editing.

When I restarted it, I was logged out and couldn't log back on. I keep getting this message above the login box saying "There seems to be a problem with your login session; this action has been canceled as a precaution against session hijacking. Go back to the previous page, reload that page and then try again.

I did that and still got the same thing. I restarted the computer. Still happens. But only in Firefox; I'm typing this on Edge.

Can anyone help me with this? Daniel Case (talk) 03:43, 7 December 2016 (UTC)

I've been unable to login using Firefox for months now - again, Edge works perfectly - for me it appears to be related to Norton 360 (Firefox works when Norton isn't installed).Nigel Ish (talk) 19:40, 7 December 2016 (UTC)

I had no problems until yesterday (Chrome also works perfectly as well).

I should add that this applies to all Wikimedia sites, as I have SUL. Daniel Case (talk) 20:27, 7 December 2016 (UTC)

It appears that Firefox may have resolved this issue, which appears was caused by an update. Mine is suddenly working again this evening. Skinsmoke (talk) 17:42, 15 December 2016 (UTC)

I just applied the update, too. No dice. Daniel Case (talk) 19:31, 15 December 2016 (UTC)

Am currently expanding an article on my one of my sandboxes. It is work in progress and needs a lot more work before going live. B-bot has informed that my uploads are orphaned and will be deleted after seven days. But this close to Christmas I will never achieve this deadline. It was only started on 2016-12-12. How do I say slow down?--Aspro (talk) 20:36, 14 December 2016 (UTC)

You cannot have a non-free image that is not in an article. If the bot didn't find it, then some user will. Just let them be deleted and request undelete when the page moves to article space. BTW File:Grandstand.png is way to big, WP:Image resolution guideline is 100,000 pixels Ronhjones  ^(Talk) —Preceding undated comment added 21:36, 14 December 2016 (UTC)

Can you reduce the size for me?--Aspro (talk) 22:04, 14 December 2016 (UTC)

Like to point out: The time to clear OTRS images on Wikimedia Commons is currently several months. Just 7 days on WP though, is proving impractical short ! A bot may be able to work 24/7 , week-in-week-out and not break out in a sweat nor need to sleep but I and many other editors can't. A Seven days cut off, must be a figure pulled out of thin air. Am simply asking... how-to-slow-these-bots-down . Some WP bots are are like skynet coming true. Who is in control here User:Ronhjones, editors (who are sacrificing their free time to improve articles) or da bots? Do you recall the first episodes of Dr Who, where the intrepid heroes came across some lovable little cyborg mutants who's favorite phrase was Annihilate ... Annihilate... Annihilate. What is the difference between Daleks and some of the the modern day WP bots?--Aspro (talk) 15:43, 15 December 2016 (UTC)

It's nothing to do with the bot. The bot is just applying a speedy template - the same template that any editor can place on finding a non-free image which is not in use (I've done hundreds over the years). The seven days used to be NONE, after much discussion some years ago, certain types of speedy deletions were given a seven days grace (now so called dated speedy deletions). This is not a bot issue, but a speedy delete issue. The relevant item can be found at Wikipedia:Criteria_for_speedy_deletion#F5._Orphaned_non-free_use_images. You could suggest a longer time there, I very much doubt if it will happen. Ronhjones  ^(Talk) 22:57, 15 December 2016 (UTC)

I've never actually met anyone who has actually done it.

How did you go about it?

What problems did you run into?

How did you solve them? The Transhumanist 22:21, 15 December 2016 (UTC)

A mirror of what ? the website, the content, the images, the servers ? There are many parts that you can filter and probably no one other that Wikimedia itself does everything. —TheDJ (talk • contribs) 09:29, 16 December 2016 (UTC)

James Forrester (Product Manager, Editing department, Wikimedia Foundation) --19:31, 14 December 2016 (UTC)

At ro:wiki we have a problem we face for many years; one person filled thousands of article talk pages with low quality articles links to various newspapers (example here), many of them with defamatory content, continuing to do so after a policy change was enforced. That user was blocked indefenetely and got permanently banned from editing ever again , and all his contributions can be reverted thanks to ro:wiki policies.

We still have a problem in removing his contributions in the talk space page. Is there any possibility to mass remove all his contributions into the article talk page's space? Thanks in advance!Ionutzmovie (talk) 00:02, 16 December 2016 (UTC)

Ionutzmovie can you point to a few of the "difficult" use cases? (e.g. Is it where this is a page creation, or where this user was not the last editor, etc?). — xaosflux ^Talk 00:15, 16 December 2016 (UTC)

If mass is the problem, you could create a bot that reverts the edits which can be reverted, lists the pages which can't be reverted and tags pages where he's the only editor for deletion. Jo-Jo Eumerus (talk, contributions) 00:17, 16 December 2016 (UTC)

So there isn't any page like Special:Nuke which can do this task?Ionutzmovie (talk) 00:19, 16 December 2016 (UTC)

There is not a Special:Nuke for "reverts"; userscripts (e.g. User:Writ Keeper/Scripts/massRollback.js to click all the rollback links for a user could be used, and admins can use their (markbotedits) acess to hide these from recent changes. — xaosflux ^Talk 00:26, 16 December 2016 (UTC)

Sorry to say this, but I think this is just a matter of personal revenge. Two years ago, an admin blocked me for "trolling" because I dared to answer his accusations, his decision was reverted and since then - war. Then I was blocked for three months because (along other things) answering to accusations is considered to be "personal attacks". Recently I was blocked for an article that I've made and last edited almost one year ago, minutes after answering to an irritated admin. And the most recent blow was to block me forever (ro:Discuție Utilizator:Ark25/2016#Blocat_2), for saving two articles from deletion and trying to save two more (even though I was invited to save articles by the same admin who started this thread), using the excuse that, by saving them, I broke my promise not to add references from newspapers in Wikipedia articles - I thought such an exception will be considered common sense and I even said I won't make such exceptions anymore if they don't like them. And they are probably not going to stop here - delete all the articles that I've created, revert all my edits, try to block me at Wikia also, I won't be surprised even if they will even sue me. Now, those links I've added into the talk pages are (media) resources, useful for further developing the articles (example) and lists (example), including the "Controversies" sections (those called defamatory). Those links are from the same newspapers cited (by others) in tens of thousands of Wikipedia articles (example). Also, I've stopped to add such links when they managed to get a consensus against my actions (it took them years to get there). So it's not really true that sentence: "continuing to do so after a policy change was enforced". But every time I'm asking them to support their accusations with evidence, they refuse to do so and in return, they launch even more attacks. And after that I get get blocked for "trolling", for "personal attacks", for "believing in conspiracy theories", and so on. I was even accused that, having my own opinion while respecting consensuns means "wikilawyering" - so, be careful what you think when editing Romanian Wikipedia.

Anyways, I offer to clean those talk pages myself, if they allow me to create an user just for that. After cleaning, block that user, no problem. I believe they won't answer to anything I have to say (I've asked them recently to allow me to make an user just for doing re-categorizations, disambiguation notes and correcting misspellings and to block it at the first edit that won't fit in those restrictions). So I'm just saying it. —  Ark25  (talk) 15:21, 16 December 2016 (UTC)

Hi everyone! I was interested ScriptInstaller script, but it doesn't work on kkwiki (only shown Must be installed manually message every .js page), how to fix it? Please help me (script page loaded by common.js). Script author currently inactive --Arystanbek (talk) 08:35, 15 December 2016 (UTC)

Try replacing with kkwiki namespace names and lengths here:

((path.toLowerCase().substring(0, 5) == "user:") || (path.toLowerCase().substring(0,10) == 'wikipedia:'))

Also replace "en-wiki" by "kk-wiki" in "If this is not a valid path to an en-wiki .js script". It's just a comment but it becomes misleading if the first change is made. PrimeHunter (talk) 10:45, 15 December 2016 (UTC)

@PrimeHunter: I was tried it this revision, but doesn't work, "en-wiki" is used only comment line. --Arystanbek (talk) 11:19, 15 December 2016 (UTC)

Try also replacing the namespace lengths as I said. The kk name for "user:" is 9 characters so it becomes:

((path.toLowerCase().substring(0, 9) == "қатысушы:") || (path.toLowerCase().substring(0,10) == 'уикипедия:'))

PrimeHunter (talk) 11:26, 15 December 2016 (UTC)

@PrimeHunter: I'm so grateful for your help! . --Arystanbek (talk) 11:46, 15 December 2016 (UTC)

@PrimeHunter: I'm sorry, how to fix these three bugs:

1. here installer link doesn't work. Shown Must be installed manually message;

2. Show installed script list link not shown;

3. already installed scripts shown install link, not shown installed message. --Arystanbek (talk) 00:58, 16 December 2016 (UTC)

I haven't used the script and am not a JavaScript programmer. PrimeHunter (talk) 15:51, 16 December 2016 (UTC)

Whenever I click on the watchlist star, the page name does not display correctly. It just says

The page "'''<nowiki>$1</nowiki>'''" has been added to your [[{{ns:Special}}:Watchlist|watchlist]], which will list edits to this page and its associated {{#ifeq:{{NAMESPACE:$1}}|{{TALKSPACE:$1}}|content|talk}} page.

. GeoffreyT2000 (talk, contribs) 05:58, 16 December 2016 (UTC)

The message is from MediaWiki:Addedwatchtext - which has not been changed in a long time, must be Thursday.. — xaosflux ^Talk 06:02, 16 December 2016 (UTC)

Able to duplicate. Not duplicating on meta: or testwiki: — xaosflux ^Talk 06:04, 16 December 2016 (UTC)

Duplicated on testwiki, get a failure on dewiki - seems to be related to using a parser function. — xaosflux ^Talk 06:09, 16 December 2016 (UTC)

I've updated our local message to make this be usable, need to file a phab bug. — xaosflux ^Talk 06:20, 16 December 2016 (UTC)

phab:T153403 opened. — xaosflux ^Talk 06:25, 16 December 2016 (UTC)

@GeoffreyT2000: looks like our localization markup is no longer supported, partial support may come back if phab:T45512 ever gets worked on. — xaosflux ^Talk 18:40, 16 December 2016 (UTC)

• Note: the phab ticket was "won't fix"'ed. The messages should be displaying in a usable format now, additional discussion can be had at: MediaWiki_talk:Addedwatchtext. — xaosflux ^Talk 20:06, 16 December 2016 (UTC)

In anonymous and some participants disappear top panel. Please tell me how to return?. the problem is browser mozile. --Дагиров Умар (talk) 12:45, 16 December 2016 (UTC)

Error fixed in MediaWiki:Gadget-collapserefs.js. --Дагиров Умар (talk) 22:06, 16 December 2016 (UTC)

When I view a category such as Category:Wikipedia files requiring renaming while being logged out, members such as File:Blc from bn.jpg are not shown. Doing a null edit only temporarily fixes the problem and the problem appears again the next time I visit the category while being logged out. GeoffreyT2000 (talk, contribs) 02:06, 15 December 2016 (UTC)

The example has now been removed from the category but I added File:HAFIZ HAMZA SHAHBAZ.jpeg instead and got the same result for many minutes. It works now but there can apparently be a delay in updating the cached version of category pages for unregistered users. I tested a new page with categories at Special:Newpages and got the same result. Nightjet was displayed in Category:Night trains when logged in but not logged out. It works now and some other new pages didn't have the issue at all when tested within a minute of creation. PrimeHunter (talk) 11:41, 15 December 2016 (UTC)

If I'm not mistaken, logged-in editors are shown a non-cached version of the page, whereas cached versions are displayed to non-logged in readers; this is likely done for performance reasons. -FASTILY 01:19, 17 December 2016 (UTC)

Now, let me start off by saying, I'm jet lagged, and probably am missing something obvious.

I did check this--the bot doesn't appear blocked.

I've been running very simple PERL scripts as User:Joe's Null Bot for some years, these work off the PERL MediaWiki API (e.g., .)

Since early November (and I was literally on Antarctica away from internets in the meantime), it appears that these scripts have been failing, unable to connect to en.wikipedia.org port 443. Can't connect to that port, I hear it cry, error 500. My only environmental upgrade that likely to be relevant was an upgrade to OSX Sierra. The PERL APIs don't appear to have changed in the last year or more, so whatever is wrong is probably on my end, but as I said, I'm jet lagged and not seeing it. Ideas? --joe decker^talk 01:06, 6 December 2016 (UTC)

Hmmm. Given what 443 is, I have one idea. Back in a few minutes. --joe decker^talk 01:09, 6 December 2016 (UTC)

Okay, what's the correct way to populate the hashref passed to API->new for an HTTPS connection? I take it that simply putting https in the URI is not sufficient. --joe decker^talk 01:13, 6 December 2016 (UTC)

Just a guess, for something to try, I've run into problems before with SSL and expired CA certificates. Maybe in the OS upgrade it lost the certificate. Though there is a status code for that 435 .. 500 is a generic which is more a question than answer. -- GreenC 02:18, 6 December 2016 (UTC)

Hmmm. ENWIKI's cert looks good, the only thing that upgraded was my own box, which is my back-room iMac where the scripts run. Weird. --joe decker^talk 02:37, 6 December 2016 (UTC)

I seem to recall an issue with root CA's and macOS Sierra. Can you check if the cert looks good from your box using Safari? I don't know what browser you normally use, but IIRC Chrome was using its own list of root CA's. Rchard2scout (talk) 08:30, 6 December 2016 (UTC)

Both Chrome and Safari report ENWIKI's certs as valid.

Slightly more caught up on sleep, I'll look into this more in the next few hours. I see that midyear I applied a hack to disable some sort of hostname checking to make SSL work, my guess is that that's breaking now. Have gotten similar results from two machines. (If someone has experience with, or better, an concise example of, all the correct incantations for a login connection to en.wikipedia.org over SSL, I'd appreciate it. --joe decker^talk 23:24, 6 December 2016 (UTC)

Wikipedia:Village_pump_(technical)/Archive_146#MediaWiki::API_and_SSL may provide some background. --joe decker^talk 23:26, 6 December 2016 (UTC)

Normally all you need with the Perl API is the protocol card:

my $bot = MediWiki::Bot->new ({
        assert       => 'bot',
        host         => 'en.wikipedia.org',
        protocol     => 'https',
}) or die "new MediaWiki::Bot failed";

Hawkeye7 (talk) 01:27, 7 December 2016 (UTC)

{{ping}Hawkeye7}} Cool, I was using the lower level interface, but it's probably not that hard to cut over.

That having been said, it still doesn't work.

2: 500 Can't verify SSL peers without knowing which Certificate Authorities to trust : error occurred when accessing https://en.wikipedia.org/w/api.php after 6 attempt(s) at afcpend.pl line 86. --joe decker^talk 16:30, 7 December 2016 (UTC)

Let me try putting the $ENV{PERL_LWP_SSL_VERIFY_HOSTNAME} = 0; fix back in. --joe decker^talk 16:32, 7 December 2016 (UTC)

And nope.

2: 500 Can't connect to en.wikipedia.org:443 : error occurred when accessing https://en.wikipedia.org/w/api.php after 6 attempt(s) at afcpend.pl line 86. --joe decker^talk 16:33, 7 December 2016 (UTC)

Have dumped code to User:Joe_Decker/afcpend.pl so that you can laugh at the kludgery. --joe decker^talk 16:39, 7 December 2016 (UTC)

• What platform are you on? I had a similar problem on the laptop, involving the version of Perl Apple had installed. You can try this:

 sudo cpan
 install LWP::UserAgent Mozilla::CA

Hawkeye7 (talk) 09:26, 9 December 2016 (UTC)

OSX, I will give that a shot today when I'm back at the house. That could totally be the issue, thanks! The laptop I'm doing most of my testing on is OSX 10.11.6 --joe decker^talk 16:08, 9 December 2016 (UTC)

Well, the good news is that it changed something :) Now I'm just consistently getting the "can't connect" error, which is sadly shy of good hints as to what's going on. Humph. --joe decker^talk 02:34, 11 December 2016 (UTC)

Still, however, no progress. --joe decker^talk 22:36, 15 December 2016 (UTC)

Couple ideas: MediaWiki::Bot->new() has a debug option. Maybe an errant .perlwikipedia-$botUserName-cookies file? -- GreenC 01:37, 16 December 2016 (UTC)

Good ideas. Debug (set to 2, if I'm reading this right) more or less repeated the error message I'm getting ("500 Can't connect....").. I don't see a cookies file, but I'm not actually sure where (besides beside my application) I should be looking. (I did know to use -a on my ls commands, tho'.) Anywhere else I should be looking for those? I have another thread to pull on, the build test for ::Bot is generating errors, so I can follow up there, too. --joe decker^talk 03:33, 16 December 2016 (UTC)

The cookie probably doesn't exist but could try a search from a root directory just in case: find . -print | grep perlwikipedia -- GreenC 15:32, 16 December 2016 (UTC)

Yep, nothing. --joe decker^talk 16:32, 17 December 2016 (UTC)

User:Theopolisme is fairly inactive (last edit May). He mde User:Theo's Little Bot. Of late the bot has not been behaving very well on one of it's tasks (Task 1 - reduction of non-free images in Category:Wikipedia non-free file size reduction requests. It typically starts at 06:00 and will drop out usually within a minute of two (although sometimes one is lucky and it runs for half an hour occasionally). Messages on talk pages and github failed to contact user. User:Diannaa and I both sent e-mails, and Diannaa did get a reply - He is very busy elsewhere, and hopes to maybe look over Xmas. In view of the important work it does, Dianna suggested I ask here if there was someone who could possibly take the bot over? Ronhjones  ^(Talk) 22:53, 12 December 2016 (UTC)

If nobody has volunteered, then it might be good to ask at Wikipedia:Bot requests as well. -FASTILY 01:24, 17 December 2016 (UTC)

Some discussion of this topic has been going on concerning WikiWork factors at the Bot requests page already-- see Wikipedia:Bot requests#Update WikiWork factors. I would very much appreciate someone reclaiming this bot, I found WikiWork factors useful. Icebob99 (talk) 00:45, 18 December 2016 (UTC)

Please forgive me if this is the wrong place to ask this question (and please direct me to the proper place to ask), but...

Does anybody know if there is a script or an automated tool that can help standardize citations? I'm working on an article in which some footnotes use citation templates and others are simply typed out (e.g., [https://www.dissentmagazine.org/article/women-black-lives-matter-interview-marcia-chatelain "Women and Black Lives Matter: An Interview with Marcia Chatelain"]. ''Dissent Magazine''. Retrieved 2016-11-24.). As part of a drive toward bringing the article to good article status, I'd like all the footnotes to use the citation templates.

I've looked, but most of the citation tools appear to be aimed toward helping editors add new citations, not formatting the existing citations in an article. If you can think of anything that might help, please let me know. Thank you. — Malik Shabazz ^Talk/_Stalk 03:14, 16 December 2016 (UTC)

@Malik Shabazz:, Help:Citation tools. I don't know if there is one for this task though I agree it would be useful. -- GreenC 17:23, 17 December 2016 (UTC)

reFill works but you have to deconstruct the citation to a barelink first. James Foley (journalist) is a good example article. -- GreenC 17:38, 17 December 2016 (UTC)

I don't think such tool will exist. I believe, it would be nightmare to have a "formatter" for all possible cases (extracting right parts). --Edgars2007 (talk/contribs) 18:09, 17 December 2016 (UTC)

Agreed a nightmare/impossible when full-auto on all formats; it might work for some typical formats when run in semi-auto mode (like reFill). -- GreenC 01:08, 18 December 2016 (UTC)

Thank you both, Green Cardamom and Edgars2007. reFill is a new one for me. I'll give it a try. — Malik Shabazz ^Talk/_Stalk 02:10, 18 December 2016 (UTC)

You've had it up for well over a month. You've blown past your goal. You have raised way the hell more than is needed for hosting or even salaries. **TAKE IT DOWN ALREADY**.

--Rpresser 19:49, 16 December 2016 (UTC)

This doesnt seem the right forum for this topic. —TheDJ (talk • contribs) 21:48, 16 December 2016 (UTC)

See the response here. wbm1058 (talk) 05:10, 17 December 2016 (UTC)

@Rpresser: I don't see the fundraiser banner, and I don't think you have to, either. There's a gadget to prevent its display. Preferences/Gadget/Browsing → Check "Suppress display of fundraiser banners" — Maile (talk) 02:21, 18 December 2016 (UTC)

My question is two fold. First the specific one... I am doing some complex searching with things like

intitle:/Infobox/ insource:/foo/ -insource:/bar/

Is there a way to add a condition that basically says "don't show me any sub pages"? What I want to avoid is any page that has a "/" in the title. So "Template:Foo" is fine... but I don't want to get "Template:Foo/<anything>".

Then the broader question. Is there a good place to read up on search syntax other than just Help:Searching? --Zackmann08 (^{Talk to me}/_{What I been doing}) —Preceding undated comment added 07:34, 14 December 2016 (UTC)

mw:Help:CirrusSearch --AKlapper (WMF) (talk) 10:52, 14 December 2016 (UTC)

You can probably do something like -intitle:/\//, where the backslash escapes the slash, so that the slash is not evaluated by the machine as terminating the regex. --Izno (talk) 13:15, 14 December 2016 (UTC)

@Izno: Unfortunately that doesnt work. -- John of Reading (talk) 13:47, 14 December 2016 (UTC)

@John of Reading: I had a sneaking suspicion it would not. --Izno (talk) 13:48, 14 December 2016 (UTC)

Maybe something from Help:Searching/Draft will be interesting. Template:Search link documentation seems to be a copy of draft page. --Edgars2007 (talk/contribs) 15:49, 14 December 2016 (UTC)

I actually wasn't even aware of that page until today. There is some more official documentation (as in, the Search Team actively updates it) at mw:Help:CirrusSearch. With regard to this specific question, I doubt what Zackmann08 is trying to do is possible at present with search. --Dan Garry, Wikimedia Foundation (talk) 17:41, 14 December 2016 (UTC)

Is it in your team's plans for the future? "Search the Wikipedia: namespace, but don't clutter up my results with a jillion AFD pages" would be very helpful. WhatamIdoing (talk) 18:59, 14 December 2016 (UTC)

I don't want to speak for your team and give them work, but a search parameter to eliminate subpages from results shouldn't be too terribly hard to implement. ^demon^{[omg plz]} 23:08, 17 December 2016 (UTC)

-intitle: AfD? — Cpiral§Cpiral 06:54, 18 December 2016 (UTC)

I don't want to type out -prefix:Wikipedia:Articles for deletion all the time. There's currently some 370,000 of them, so it's a lot of potential clutter. Plus, if we can exclude sub-pages, then I won't have to also add -prefix:Wikipedia:Requests for or any of the other things. (And just imagine being able to search templates but not their /doc pages by clicking a little tickbox...) WhatamIdoing (talk) 07:20, 18 December 2016 (UTC)

This is how I solve complex searching. -- GreenC 16:00, 14 December 2016 (UTC)

Very cool. Bookmarked as desirable project. — Cpiral§Cpiral 07:03, 18 December 2016 (UTC)

Subpages are always searched. Sorry. There may be a way, if you described more, but please

• try -intitle: doc -intitle: sandbox to refine.
• don't use insource:/regex/ unless you've read the docs mentioned and can provide a filter for it. You can probably use plain ol' insource.
• get involved with Search docs and Search talk pages before asking for a feature on phabricator. Someone will be happy to work up any possible query, and even discuss and explain. This helps the poor ol' docs too. — Cpiral§Cpiral 06:54, 18 December 2016 (UTC)

A while ago I complained on this board that the software was logging me off for no reason. When I log in, I check the box to stay logged in for 365 days. When it used to be 30 days, it worked fine. After the change, it started logging me off. The last time I posted here it was happening regularly but no more than once a day and not always every day. No one proposed a solution.

It's now gotten much worse. It's logging me off multiple times on the same day, sometimes within minutes of the last one. It's very irritating because it seems to happen more frequently when I'm running a CheckUser, which means of course it tells me I don't have permission, so I log in and continue.

If I don't fare any better here than I did last time, I am hoping someone can tell me how to file a bug report with Phabricator (I've never done it). Thanks.--Bbb23 (talk) 00:19, 18 December 2016 (UTC)

phabricator. There's a welcome box up top that tells you how to do it. You should be able to log on to Phabricator with your Wikimedia unified account. — Maile (talk) 02:26, 18 December 2016 (UTC)

I created a bug report. How do I know if any action was taken on it? Am I notified? On Wikipedia or only on Wikimedia? Thanks.--Bbb23 (talk) 03:27, 18 December 2016 (UTC)

Go back to Phab: and click on the wrench icon in the upper right corner of the screen. That's your settings. You'll find a section about e-mail notifications in there.

Have you done all the usual WP:BYPASS things to make sure that it's not being caused by a corrupted cookie or other local problem? Whatamidoing (WMF) (talk) 04:42, 18 December 2016 (UTC)

Emails are sent from no-reply [at] phabricator.wikimedia.org so check your spambox. --Redrose64 (talk) 11:32, 18 December 2016 (UTC)

I started getting notifications later. They work fine. I've commented at the task to answer the questions asked me. As I mentioned at the outset, I brought this up here once before. See here. As you'll note, Whatamidoing (WMF), I followed the BYPASS procedure back then. I see no need to repeat it now as it didn't work then.--Bbb23 (talk) 15:10, 18 December 2016 (UTC)

20:33, 19 December 2016 (UTC)

For the WatchlistBase and WatchlistGreenIndicators gadgets I'm getting console warnings on every page load that look like "Gadget "WatchlistBase" styles loaded twice. Migrate to type=general. See <https://phabricator.wikimedia.org/T42284>." The documentation at mw:ResourceLoader/Migration guide (users)#Gadget type says that this warning is given for gadgets that load both styles and JavaScript. However, when I checked MediaWiki:Gadgets-definition it seems that these two gadgets load only styles. Does anyone know why the warning is appearing? — Mr. Stradivarius ^{♪ talk ♪} 03:40, 18 December 2016 (UTC)

I can only suggest that this is related to the dependencies. Ruslik_Zero 20:42, 19 December 2016 (UTC)

Because of phab:T150045, you will soon get a confirmation dialog when you click the "mark all as visited" button on your watchlist. You can see how this works now on Meta. I find this annoying and disruptive to my workflow; lack of confirmation never caused me a problem. Can I bypass it with some script or will I have to wait for an opt-out to be provided like I asked for at the phab task? 🎄BethNaught (talk)🎄 08:08, 15 December 2016 (UTC)

I would guess that a script can be written in the same vein as the auto-"purge" script that got built when they rolled out the POST semantics rather than the GET semantics for purge. --Izno (talk) 13:01, 15 December 2016 (UTC)

A JS override has been provided by Sn1per at phab:T150045#2876153. 🎄BethNaught (talk)🎄 17:15, 15 December 2016 (UTC)

• "Are you sure you want to reset unseen watchlist changes by marking all pages as visited?" AKA "Are you sure you want to take a shit and can't hold it any longer?" AKA "I love to annoy you by wasting more of your free time". Seriously??????????????? That's a real brain fart!--TMCk (talk) 00:26, 16 December 2016 (UTC)
• To be honest BethNaught, while that solution works ok, it looks pretty ugly considering that the popup appears, lingers a second and then disappears. I much prefer my fix, which replaces the watchlist reset button with one that does the same thing but without a confirmation dialog. As an added "feature", the page updates visited pages without reloading.

var newButton = '<button id="tv-watchlist-resetbutton" class="mw-ui-button">Mark all pages as visited</button>';
var reloadButton = '<button id="tv-watchlist-reloadbutton" class="mw-ui-button" onClick="window.location.reload();">Reload the page</button>';
$("#mw-watchlist-resetbutton").replaceWith(newButton);
$("#tv-watchlist-resetbutton").click( function() { 
		$.ajax( {
    	url: mw.util.wikiScript( 'api' ),
    	type: 'POST',
    	dataType: 'json',
    	data: {
        	format: 'json',
        	action: 'setnotificationtimestamp',
        	entirewatchlist: 1,
    	    token: mw.user.tokens.get( 'csrfToken' ),
  		}
	} ).done(
		function() {
            $('.mw-changeslist-line-watched').removeClass('mw-changeslist-line-watched').addClass('mw-changeslist-line-not-watched');
            $("#tv-watchlist-resetbutton").replaceWith(reloadButton);
        });
} );

Cheers -- The Voidwalker ^Whispers 01:10, 16 December 2016 (UTC)

Actually, this seems broken. After I click an affirmative confirmation, the magical javascript bar runs to 100%, and then disappears. As does the original "mark all as visited". There are other issues with the implementation; I've filed phab:T153389. --Izno (talk) 01:06, 16 December 2016 (UTC)

The disappearing button was part of the design because the old form submission-based reset loads new watchlist items, since it reloads the page, but the JS reset does not (at least it has not been implemented yet). If you were to push the button more than once, new watchlist changes that would not be currently displayed on the watchlist would be marked as visited, but no visual change would occur. The button was designed to disappear so that you wouldn't do this by accident and end up marking new, unseen and undisplayed watchlist changes as visited. (see phab:T153389#2879181 and relevant comment on gerrit at ) Cheers, Sn1per (talk) 01:25, 16 December 2016 (UTC)

I've updated my script to account for that. (It replaces the "Mark all pages as visited" button with one that reloads the page). -- The Voidwalker ^Whispers 02:02, 16 December 2016 (UTC)

Can the popup at least not take so damn long to pop up? It seems like it has some animated grow-in effect, which takes time, rather than just appearing. I wouldn't mind the confirmation prompt if it were prompt. DMacks (talk) 03:50, 16 December 2016 (UTC)

How long is it taking to show up for you? It's less than a second on my computer, but if it's significantly slower that's something we need to fix. Legoktm (talk) 07:57, 16 December 2016 (UTC)

Nearly a second from first clicking the button until the popup is fully done appearing. What is gained by any animation at all? That's also nonstandard compared to the other popup features in any other WM context. DMacks (talk) 14:47, 16 December 2016 (UTC)

Worst Xmas present I ever got. Same behavior if you want to purge the copyvio or duplicate lists on Commons. Drives me nuts. Reminds me of Data Becker Commodore C64 software, they added are you sure and are you really sure. Was a running gag in the community for quite a while.... --Hedwig in Washington ^(TALK) 04:47, 16 December 2016 (UTC)

Hi, the main reason we added in a confirmation box was that pressing that button is a destructive action that is entirely impossible to undo. There are JavaScript hacks to work around it, but is pressing "enter" to confirm the dialog to inconveniencing? (serious question). I'm open to removing it and reverting back to the status quo, but I would have thought a confirmation step for actions that cannot be undone would be a reasonable thing. Legoktm (talk) 07:57, 16 December 2016 (UTC)

Yes, please revert. There's always the original user script for people who find it a problem. 🎄BethNaught (talk)🎄 08:23, 16 December 2016 (UTC)

Yes pretty please. It is disrupting workflow. Please for purging as well. People how purge need the purge, no need for a confirmation. Thanks! --Hedwig in Washington ^(TALK) 08:54, 16 December 2016 (UTC)

With many +++ from de:WP (e.g. de:Wikipedia:Fragen zur Wikipedia#Beobachtungslistenerliedigungssoftwareterror. It is annoying and takes additional time. Even if you press this button by accident your watchlist is still there and applicable.

And +1 for Hedwig in Washington's wish to change this for purging as well. Keep it simple, please. NNW (talk) 09:29, 16 December 2016 (UTC)

The purge confirmation is a separate issue involving GET vs POST. You can by-pass the GET confirmation with JS, but the devs are disinclined to change it back. --Unready (talk) 09:46, 16 December 2016 (UTC)

No, unfortunately I cannot by-pass it because I am not able to handle JS. If you someone else is I would be pleased to get know what to do to get rid of these confirmations. NNW (talk) 10:37, 16 December 2016 (UTC)

The purge confirmation is a separate issue. It's been discussed before. You can bring it up again in a separate thread, but it's unlikely the outcome will be different. --Unready (talk) 21:42, 16 December 2016 (UTC)

Opinions will be divided, but for me it is extremely annoying to be treated as an idiot and forced to make superfluous clicks every day. An opt-out is a must for this one. --Jossi (talk) 11:20, 16 December 2016 (UTC)

Criticism has also been raised at Commons. 🎄BethNaught (talk)🎄 11:49, 16 December 2016 (UTC)

@Legoktm: I don't know whether to respond here or the ticket I filed (where I think Sn1pre unhelpfully struck out my comment on whether the disappearing button was a bug).

My personal opinion is that the general behavior is great, but the fact the button disappears is basically a shit design. Fomafix's comment on the commit "may be irritating" is not born out by my use of the button--it's irritating not to have that there, which would be more consistent with old behavior (even if it doesn't appear to do anything!). His comment re "This functional change can be done by a separate change to allow a specific documentation and discussion." is absolutely what I would expect. --Izno (talk) 12:59, 16 December 2016 (UTC)

I'm one of those who benefit, since my workflow never needs the "Mark all as visited" button, and every few months I hit it by accident and cannot undo. Jim.henderson (talk) 14:53, 16 December 2016 (UTC)

I've had difficulty getting the "reason given" section to work for this tag on my Wikicomons Discussion page here: https://commons.wikimedia.org/wiki/User_talk:Ericmachmer (Edit: not my wikipedia talk page). Even copying and pasting the tag without modification does not seem to result in the "your reasons here" section appearing. Instead the tag always displays "no reasons given". (My reasons are still in the code for the tag on my page, but are not rendered...and they have been reproduced in plain text below the tag.) Thanks for your time. Ericmachmer (talk) 23:44, 19 December 2016 (UTC)

You have an = in your reason (several, actually). The reason should be an unnamed parameter. Typically, you would use {{=}} in place of = in parameter text. Alternatively, instead of reason= (which is incorrect in any case), you could change it to 1=. --Unready (talk) 00:26, 20 December 2016 (UTC)

Thank you! — Preceding unsigned comment added by Ericmachmer (talk • contribs) 02:46, 20 December 2016 (UTC)

I've been having email issues for about a week now, I'd say, where I'm only sometimes getting emails for pages I'm watching instead of getting emails pretty much any time something is changed on a page I'm watching. Anyone else? Amaury (talk | contribs) 17:14, 20 December 2016 (UTC)

@Amaury: Is this the same as #And yet another problem above? --Redrose64 (talk) 17:32, 20 December 2016 (UTC)

@Redrose64: Possibly? I'm not sure. Were the watch list changes about a week ago? Amaury (talk | contribs) 17:42, 20 December 2016 (UTC)

I'm trying to construct an API query in the API sandbox. I start by choosing "query" in the "action" drop-down list, and format "json". At this point, I see in the video, "An introduction to the API by Roan Kattouw at the San Francisco Hackathon January 2012" the next step is to choose a query type in the "what to query" drop-down menu that appears below "query". However, it appears that Roan is demonstrating an old and deprecated version of the API sandbox, as I don't see any "what to query" drop-down menu in the current version. Unfortunately, this is a puzzle now because there is no video "An introduction to the new API sandbox at the San Francisco Hackathon January 2015". Can anyone help? wbm1058 (talk) 17:38, 20 December 2016 (UTC) The video is mw:File:The MediaWiki Web API and How to use it - San Francisco Wikipedia Hackathon 2012.ogv. – wbm1058 (talk) 17:47, 20 December 2016 (UTC)

Yeah, that video looks pretty dated. After selecting query as your action, you should be able to click on the "action=query" in the left bar where you can now refine how you're querying. What you'll want to be setting are the props, which are best documented over here. ^demon^{[omg plz]} 19:42, 20 December 2016 (UTC)

When I saw the new sandbox, I was confused like that too. That needs to be made a bit more clear.—^cyberpower_{Merry Christmas:Unknown} 19:47, 20 December 2016 (UTC)

So I was patrolling a page earlier, but when I clicked on the "mark page as patrolled" button, instead it loads up the Media Viewer, showing the article's infobox images. This is the first time I encountered this problem on the desktop version of Wikipedia, although occassionally I've had similar problems on the mobile version where when I open a section and an image is at the top of the section, if I press anywhere, the Media Viewer will load even if I wasn't pressing on the image. I'm not sure if it's a problem with me or my phone (as in maybe I accidentally clicked on the image and I didn't realize it; this commonly happens if I open the section and scroll down and the image doesn't load up immediately), or if it's an actual bug, though. Narutolovehinata5 ^tccsdnew 05:01, 21 December 2016 (UTC)

I noticed that Template:Search deletion discussions doesn't work ("An error has occurred while searching: Search request is longer than the maximum allowed length. (485 > 300)") and from a look at Template talk:Search deletion discussions, hasn't worked since July or earlier. Researching this, I found that the creator had built several similar search boxes: User:Stmrlbs/search that search various message boards using a syntax that separates each page with a pipe (WP:Articles for deletion|WP:Categories for discussion) but it seems this pipe syntax isn't supported. It seems that this must have worked when it was first implemented. Was this broken by the switch to CirrusSearch perhaps? How should this be fixed? wbm1058 (talk) 04:56, 17 December 2016 (UTC)

Template:Search prefixes is the template that assumes multiple pages to search can be piped together. wbm1058 (talk) 16:25, 17 December 2016 (UTC)

I'll start looking at the templates, update you. — Cpiral§Cpiral 10:43, 18 December 2016 (UTC)

Due to some extremely long searches that were causing performance issues and consistently returning no results, a maximum length for search queries was implemented around a year and a half ago. The query length is generous; it was chosen based on analysing search logs to only affect a minuscule number of searches. I suggest that the author try to alter his search boxes to fit his queries into the maximum length. :-) --Dan Garry, Wikimedia Foundation (talk) 16:06, 19 December 2016 (UTC)

I see: phabricator T107947 – wbm1058 (talk) 02:09, 20 December 2016 (UTC)

Dan, the maximum allowed length isn't the main issue, the question is why this doesn't work when pasted into the search box:

dog prefix:WP:Articles for deletion|WP:Categories for discussion

when these do produce results:

dog prefix:WP:Articles for deletion

dog prefix:WP:Categories for discussion

In other words, did Template:Search deletion discussions ever work, or was it broken from day 1 when it was created in 2009? Is there a way to search both Articles for deletion and Categories for discussion subpages at the same time, and get a combined results list? If not then we should probably put Template:Search deletion discussions up for deletion. The author hasn't been active since February 2012. wbm1058 (talk) 05:12, 20 December 2016 (UTC)

Prefix never took pipes, so... (Prefix never took negation -prefix: either.) Prefix works the same in CirrusSearch as it did with MWSearch (2014 transition). (See mw:help:cirrusSearch#Notes and references, note 9). — Cpiral§Cpiral 05:02, 21 December 2016 (UTC)

Hello.There are here technical suggestions designed to serve all the Wikimedia projects.please submit requests for apply them or express opinions on the talk page here.I do not want to wait until November 2017.Thank you --ديفيد عادل وهبة خليل 2 (talk) 14:07, 21 December 2016 (UTC)

I happened to run across a reversion just now for which I received no notification. I checked my Preferences, and I see it is set to notify me about reverts "using the undo or rollback tool". Now it is true, Twinkle is not one of those tools, but I want to see those edits treated the same way.

I mean, I already feel like Twinkle is like an Air Force fighter, honed to project the power of a single warrior against thousands of peasants on the ground. It is in keeping with that design not to let the rabble know when they're being targeted. But, well, that's not the way I want to see Wikipedia work. Someone with a fancy tool designed to do quick reverts ought to be leaving as much notice as J. Random User who sees the Undo button on the manual display.

How did this come to happen, and is there an easy fix? Wnt (talk) 00:21, 17 December 2016 (UTC)

This is a known problem. Opinion at WT:Echo#Twinkle reverts is that Twinkle doesn't revert - instead, it goes to the selected previous version, opens that in edit mode and saves. --Redrose64 (talk) 09:40, 17 December 2016 (UTC)

I think the most satisfying solution here would be for submitted edits to be checked to see if they are identical with previous states (in theory, some file could hold checksums for all the edits to an article so that the software would only generate a new checksum for the newest edit and quickly go through the list; checksums could be generated based on non-whitespace characters only so a reversion with a space added would still trip it). Notifications would then be generated for everyone who submitted an edit since then. But it should also be possible to add the "&undo=xxxxxx" text in the Twinkle program - it must already know what edit it is undoing. Wnt (talk) 14:02, 17 December 2016 (UTC)

All revisions have a sha1 which can be used for that kind of things. Cenarium (talk) 20:46, 17 December 2016 (UTC)

I have filed a bug report for this, phab:T153567. Cenarium (talk) 19:40, 18 December 2016 (UTC)

This will be fixed in the first week of January. Cenarium (talk) 14:35, 21 December 2016 (UTC)

Is it possible to add a "CSS image crop" template to an infobox? For example, at Mound Landing, Mississippi I added the 1862 map as an image crop. I'd like to add the same map (with different coordinates) to Norfolk, Mississippi, but if I add the map to the body of the article it will make the quotation look squished. I tried adding it as an "image_skyline" to the infobox, but it didn't work. Is there a work-around? Thanks! Magnolia677 (talk) 03:14, 21 December 2016 (UTC)

I added the CSS crop from Mound Landing to the Norfolk page - and it seems to work OK...Jokulhlaup (talk) 09:17, 21 December 2016 (UTC)

@Jokulhlaup: - Thanks for making the change to Norfolk, Mississippi, but it didn't work. I use Firefox and the text is running up the left side of the image. Magnolia677 (talk) 12:10, 21 December 2016 (UTC)

Got it. I moved the text into the CSS and it works fine now. Thanks for your help! Magnolia677 (talk) 12:31, 21 December 2016 (UTC)

Glad it worked, seems to be a tricky thing to get right for all browsers...Jokulhlaup (talk) 14:51, 21 December 2016 (UTC)

Is there any way to find which articles have the same image twice, like a category or tool? At my local wiki robots were used to add infoboxes into articles, but many remained with the same image twice, like here.Ionutzmovie (talk) 00:11, 21 December 2016 (UTC)

This would be easy with a regex of a database dump, but looking at the example page, the infobox just uses {{Infocaseta Biografie}} .. where does Infocaseta Biografie get information to use the image? Must be Wikidata (yes), which complicates things. In this case a bot would have to check every article using Infocaseta, check Wikidata for image name, see if it exists in the article. It would also have to check for double image names in the wikitext. I would suggest a bot request at rowiki. -- GreenC 22:16, 21 December 2016 (UTC)

I use the ORES beta feature. On Special:Contributions/Gulumeemee, I tried to search for edits needing review by checking "only show edits needing review", but it caused a database error. This doesn't happen with the other checkboxes. I tried it for contributions of a few other users, but the searches also caused database errors. I remember doing the search a few weeks ago, and it didn't cause any errors. Is this a bug? Gulumeemee (talk) 10:52, 22 December 2016 (UTC)

I just went to the history of an article to thank someone for an edit and the "thank" feature isn't there. Is this a bug, or has it been disabled? --Anthonyhcole (talk · contribs · email) 23:41, 21 December 2016 (UTC)

Works for me - which page is this, which revision? --Redrose64 (talk) 00:02, 22 December 2016 (UTC)

Also works for me. Please always post an example when you report a perceived problem. I don't know your case but you cannot thank IP's or yourself. PrimeHunter (talk) 00:48, 22 December 2016 (UTC)

It's back now, thanks. I should have taken a screenshot, I suppose. I noticed it on Maree Smith and then checked a few more histories. --Anthonyhcole (talk · contribs · email) 00:54, 22 December 2016 (UTC)

I've seen it temporarily not work sometimes. Might be a rendering issue somewhere. -- GreenC 16:19, 22 December 2016 (UTC)

As a small side-project, I wanted to make a year-end tool that quantified an editor's content impact. I have a few ideas, but I was interested in parsing a user's edits from the last year to determine how many bytes were added in text (if possible, non-citation text). If further possible, it'd be nice to see whether those edits stuck (whether they were immediately reverted or modified, or to what extent they remain in the current article). I know there have been prior efforts on these fronts, so if someone could point me towards the right projects/tools, I'd appreciate it. czar 20:01, 22 December 2016 (UTC)

I remembered that dashboard.wikiedu.org does most of this (tracks words added & page views on specific articles in a specific time period), but open to other ideas if you have them. Just ping me  czar 20:46, 22 December 2016 (UTC)

Notifications takes from 30 seconds upwards to display. Is there a massive coding inefficiency? All the best: Rich Farmbrough, 12:05, 21 December 2016 (UTC).

Is this just Wikipedia:Village pump (technical)/Archive 152#"Your notices" again? --Unready (talk) 23:25, 21 December 2016 (UTC)

Feel free to follow phab:T153011#2870224 and provide any relevant output in the console of your web browser's developer tools (see Firefox ≥24, Internet Explorer, Google Chrome, Apple Safari for more info) in a comment in that Phabricator task. Thanks! --AKlapper (WMF) (talk) 10:16, 22 December 2016 (UTC)

Rich, what browser are you using? Most of us with this problem are using Firefox, and the solutions seem to be fixing/turning off user scripts/gadgets. Whatamidoing (WMF) (talk) 21:44, 22 December 2016 (UTC)

Firefox, for sure. Chrome is still slower than I would expect, but only a couple of seconds. I have a total of around 870 notifications, I don't know if this is a lot or not many, but I did wonder if that was related. Also I am registered on almost every project, another possible cause of slowness. All the best: Rich Farmbrough, 22:00, 22 December 2016 (UTC).

I have more accounts and more notifications in this account than in my volunteer account, but I use Safari for this account and it's much faster.

We're collecting information at phab:T153011; please feel free to add yours. Kaldari added some details about a script that he fixed, which cleared up some of the problems. Whatamidoing (WMF) (talk) 22:27, 22 December 2016 (UTC)

Snottywong's vote counter tool that's supposed to count how many people voted a certain way in an AFD is not working (at least for me, even though I was able to use it just a few days ago). I would post this on Snottywong's own talk page but he is retired. Everymorning (talk) 22:45, 22 December 2016 (UTC)

It seems to be working for me. correctly shows 4 !votes for deletion. When you say it "is not working" -- what did you try? What happened? - Jarry1250 ^{[Vacation needed]} 23:59, 22 December 2016 (UTC)

Never mind, it's working for me now. Everymorning (talk) 04:12, 23 December 2016 (UTC)

(This thread was reposted after being archived without an answer. --joe decker^talk 22:37, 15 December 2016 (UTC))

Were there signficant changes to the Wikimedia server configuration for ENWIKI that would affect login APIs deployed in November?

(This is related to Wikipedia:Village_pump_(technical)#API_PERL_question.._correct_specification_of_HTTPS_connections, but here I just need a yes/no.) --joe decker^talk 16:48, 8 December 2016 (UTC)

Joe, has the bot been blocked during the last month or so? A change to how blocked accounts are handled is the only change I have found in recent Tech News announcements that seems to be potentially related. Whatamidoing (WMF) (talk) 00:00, 16 December 2016 (UTC)

Nope, but that's a good thought. Not seeing any items in the block log and it still seems to have a bot flag. Thanks for the idea...--joe decker^talk 01:16, 16 December 2016 (UTC)

Two-factor authentication was enabled for admins in November - maybe that has something to do with it? — Mr. Stradivarius ^{♪ talk ♪} 01:18, 16 December 2016 (UTC)

Joe's Null Bot shouldn't be an admin :) Don't think it is, anyway.  :) I think I'm seeing the build test fail in a similar way, more soon. --joe decker^talk 03:26, 16 December 2016 (UTC)

Yeah, whatever is going on appears to fail the build tests, so it's either on my side (totally possible, but a little weird since it went out of service while I was away), or the server side on *both* ENWIKI and test.wikipedia. --joe decker^talk 07:51, 17 December 2016 (UTC)

@Joe Decker: Could you post an example of the query you're trying to make, the HTTP response code, and the body of the HTTP response? -FASTILY 08:13, 17 December 2016 (UTC)

@Fastily: Sure thing. A full sample script (save for password redaction) is at User:Joe Decker/afcpend.pl, the relevant API call is below, the relevant output is 2: 500 Can't connect to en.wikipedia.org:443 : error occurred when accessing https://en.wikipedia.org/w/api.php after 6 attempt(s) at afcpend.pl line 82.

 my $mw = MediaWiki::Bot->new({
    assert => 'bot',
    host => 'en.wikipedia.org',
    protocol => 'https',
    operator => "Joe's Null Bot",
}) or die "new mwbot fail";

...

 while (1) { 
    if ($mw->login( { username => "Joe's Null Bot", password => 'REDACTED' } )) {
      last;
    }
    .... error handling ...
  }

The build test "make test" for MediaWiki-Bot-5.006003 fails with a similar error, failing attempting to connect to test.wikipedia.org. The test recommends a particular mailing list for reporting problems with the component or tests, email there has failed to produce a response. --joe decker^talk 16:48, 17 December 2016 (UTC)

My Perl scripts are working fine. I am running MediaWiki::Bot v5.006003. The code you posted is the same as mine. I rebuilt the bot module and did "make test". It "Failed test 'lock history'" and "Failed test 'The API error text was returned'", but it appears to have connected ok to Wikipedia.

Have you tried the debug setting? ie. debug => 2.

Have you tried logging in from the MediaWiki::API interface? Bgwhite (talk) 07:56, 19 December 2016 (UTC)

First, yeah, that does leave me to suspect that there's something systemic in my particular operating system or environment (I'm getting the same results on both machines, but both are MacOSX, and it's possible I hit an update. Glad to know that the build tests are working on the same version I'm using for someone.

Yes, I've tried debug => 2. It produces no output other than more or less reiterating the 500 "can't connect" information.

Yes, actually, I started this process with the ::API interface, which is what I traditionally ran on. Conditional access errors referred to in the other thread led me to try out the ::Bot interface instead while I've been debugging this, and the results are similar.

To summarize what I think I know so far, from what you're saying -- it isn't specific to a server (test.wikipedia and en.wikipedia), it isn't specific to the version of ::Bot or what have you, and it isn't precisely specific to a single machine of mine. Which means it's likely something about my code or my environment -- MacOSX or something in my internet connection being the only candidates that come to mind. This also is consistent with the fact that the code had been working, and seemed to fail while I was literally half a planet away from the it, without internet access.

(And yeah, I hear some of you thinking this: When I have a little free time, it would probably make sense to just redo this functionality on labs or what have you in, perhaps Python. I originally used Perl because I knew it and it had to only be a few lines of code, right? I'm not suggesting Python is better suited or not, but actually hosting this outside of my environment would be superior for several reasons which are likely obvious. I just need to have some time to get around to it.) --joe decker^talk 16:07, 19 December 2016 (UTC)

I'd suggest moving it to WMFLabs. Both MediaWiki::API and MediaWiki::Bot are installed on the system. One can install Perl modules in their home directory to use system wide. I've got my Perl programs causing havoc and mayhem at labs. Bgwhite (talk) 23:02, 19 December 2016 (UTC)

I look into getting set up there. Thank you, that's a very sensible next step. --joe decker^talk 21:18, 21 December 2016 (UTC)

Starting the process, but I'll admit that I'm a little lost as to the environment that I've found so far. --joe decker^talk 06:03, 23 December 2016 (UTC)

Hi all. Please see this post at WP:AN. It appears that something has changed so that not only do expired BLP PRODs not appear in the requisite category, but they also don't appear in WP:PRODSUM, which is where admins generally patrol the queue of expired PRODs. Therefore, at the moment we have 85 unsourced BLPs which are tagged for deletion; some are still within the 7 days, but many are not and there is no way of differentiating; Category:BLP_articles_proposed_for_deletion_by_days_left is not working (for example Nicole Mandich has expired but is still showing as 4 days left). I'm unsure where the issue lies - the PROD BLP templates haven't been edited for a while, and this seems to be a recent issue. Black Kite (talk) 19:40, 22 December 2016 (UTC)

@Black Kite: This is a side-effect of Joe Decker (talk · contribs)'s problem with User:Joe's Null Bot - see the top VPT thread.

Does anyone else have a neat way to null-edit every member of a category? -- John of Reading (talk) 20:11, 22 December 2016 (UTC)

Yes, I can do a one time run - which category? — xaosflux ^Talk 20:18, 22 December 2016 (UTC)

The only tool in my tool box is AWB, so I've just used that to null-edit everything in Category:BLP articles proposed for deletion by days left. I hope I don't have to do that every day, though. -- John of Reading (talk) 20:20, 22 December 2016 (UTC)

Thanks for that. My question is why is it only affecting BLPPRODs and not normal PRODs which are still appearing in WP:PRODSUM as normal? Black Kite (talk) 20:43, 22 December 2016 (UTC)

I"m guessing that the answer to your question has to do with how the code that produces PRODSUM gets information about what PRODs and BLPPRODs are out there in general. Probably gets a list of one from one category, one from another, and those categories are probably populated differently by the PROD and BLPPROD templates.

FYI, I'm trying to get stuff running over at WMF labs, but I still can't figure out enough of the environment to get anywhere useful. In time, I'm sure. --joe decker^talk 06:07, 23 December 2016 (UTC)

Original announcement by Yurik (WMF)

Commons Datasets (Tabular and map data) has been launched on Commons, available from all wikis.

TLDR: One data store. Use everywhere. Upload table data to Commons, with localization, and use it to create wiki tables, lists, or use directly in graphs. Works for GeoJSON maps too. Must be licensed as CC0. Try this per-state GDP map demo, and select multiple years {{Graph:US Map state highlight}}:

This graph was using the legacy Graph extension, which is no longer supported. It needs to be converted to the new Chart extension.

See or edit source data.

Data can now be stored as *.tab and *.map pages in the data namespace on Commons. That data may contain localization, so a table cell could be in multiple languages. And that data is accessible from any wikis, by Lua scripts, Graphs, and Maps.

Lua lets you generate wiki tables from the data by filtering, converting, mixing, and formatting the raw data. Lua also lets you generate lists. Or any wiki markup.

Graphs can use both .tab and .map directly to visualize the data and let users interact with it. The GDP demo above uses a map from Commons, and colors each segment with the data based on a data table.

Kartographer can use the .map data as an extra layer on top of the base map. This way we can show endangered species' habitat.

Demo

This graph was using the legacy Graph extension, which is no longer supported. It needs to be converted to the new Chart extension.

See or edit raw graph data.

• Raw data example c:Data:Ncei.noaa.gov/weather/New York City.tab
• Interactive Weather data {{Graph:Weather monthly history}}
• Same data in Weather template User:Yurik/WeatherDemo
• Interactive GDP map {{Graph:US Map state highlight}}
• Endangered Jemez Mountains salamander habitat
• Population history {{Graph:Population history}}
• Line chart {{Graph:Lines}}

Getting started

• Try creating a page at data:Sandbox/<user>.tab on Commons. Don't forget the .tab extension, or it won't work.
• Try using some data with the Line chart graph template

A thorough guide is needed, help is welcome!

Documentation links

• mw:Help:Tabular Data
• mw:Help:Map Data

If you find a bug, create Phabricator ticket with #tabular-data tag, or comment on the documentation talk pages.

FAQ

• Relation to Wikidata: Wikidata is about "facts" (small pieces of information). Commons datasets store "blobs" - large amounts of data like the historical weather or the outline of the state of New York.

TODOs

• Add a nice "table editor" - editing JSON by hand is cruel. phab:T134618
• "What links here" should track data usage across wikis. Will allow quicker auto-refresh of the pages too. phab:T153966
• Support data redirects. phab:T153598
• Mega epic: Support external data feeds.

— JJMC89 (T·C) 21:09, 22 December 2016 (UTC)

User:JJMC89 this is very cool and opens a lot of potential. Do you know if it's possible to import Wikipedia API data (such as JSON result) into a table without resorting to a bot? That could open the full API to Lua on Wikipedia. A first step in external data feeds? -- GreenC 17:36, 23 December 2016 (UTC)

I just cross-posted the announcement. Yurik may be able to answer your question. — JJMC89 (T·C) 18:24, 23 December 2016 (UTC)

@Green Cardamom: Lua (sadly) still has no access to MW API. I thought there was a task, but I couldn't find it, so created phab:T154042. Datasets can only be populated by direct editing, until we come up with the "external data feed" system that allows community to subscribe to reputable data sources. Which means - only user and bot edits for now. There has been some talk about creating a bot that copies the results of SPARQL queries into datasets, and those queries can be configured via the corresponding talk pages. I think something similar can be done for MW API bot. P.S. @JJMC89: thanks for posting! --Yurik (talk) 19:41, 23 December 2016 (UTC)

Thanks for opening the Phab I'm following it. Yeah it wouldn't be difficult to write a cron bot on Tools that dumps API data into space where Lua can access it but it would be brittle and custom work. -- GreenC 20:13, 23 December 2016 (UTC)

I think I know why my contributions page looks different but I want it back to its previous appearance please - can someone help? Please? I tried to translate a deletion discussion from the Swedish Wikipedia using the internal-tool and now I have these three icons at the top of my Contributions page - New Page / Upload media / Translation. I want my old (just my) Contributions page back. I find the icons to be distracting - please can someone tell me how to make them go away? If I want to do an upload, I go and do it. If I want to translate, I'll use an external tool next time. I don't need an icon for a long list of "New Pages" (Wanted Pages) that I'll never use. Thanks, Shearonink (talk) 22:24, 26 December 2016 (UTC)

@Shearonink: This is probably the Content Translation tool, which you can turn off by going to the Beta tab, unchecking it, and clicking Save. Sam Walton (talk) 22:29, 26 December 2016 (UTC)

Bless you kind stranger, I just found those three large icons to be so distracting. Heh, it would be nice if "Did you upload this gadget by accident? Would you like to turn this feature off?" instructions were included with the gadget for those of us who are techno-challenged. Shearonink (talk) 22:33, 26 December 2016 (UTC)

A while ago I have created a few articles in pages which were redirect. After a few weeks, those articles are still not indexed by search engines like google. I have tried google's crawling request site https://www.google.com/webmasters/tools/submit-url and that still does not work. Is there sone setting on wikipedia for former-redirect pages that prevented search engibe crawlers from crawling those oages and how to make those crawlers crawl those pages? C933103 (talk) 05:23, 27 December 2016 (UTC)

@C933103: There might be such a thing, but we can't tell as you need to link to the page in question. Jo-Jo Eumerus (talk, contributions) 10:20, 27 December 2016 (UTC)

We should probably add this issue to Phabricator, if it isn't there already. Do you have any specific examples? I don't think it's an issue on our end, I know the WMF has contacts with Google. Cenarium (talk) 10:37, 27 December 2016 (UTC)

Are the pages in question yet marked as reviewed? They're no indexed until a page patroller has reviewed them. --Izno (talk) 10:54, 27 December 2016 (UTC)

@Jo-Jo Eumerus, Cenarium, and Izno: Xinjiang Time, Longest flights, Ultra long-haul C933103 (talk) 11:00, 27 December 2016 (UTC)

Xinjiang Time was created as a redirect to Time in China#Xinjiang by C933103 on 9 December 2016 and changed to an article on 17 December. My Google search for "Xinjiang Time" finds Time in China as the second hit two and says it was cached on 26 December. Thincat (talk) 11:15, 27 December 2016 (UTC)

I can find the last two pages on Google just fine. The first does not display. Might have something to do with it being unpatrolled - didn't we recently change unpatrolled pages so that they don't index? Anyhow, patrolled it. Jo-Jo Eumerus (talk, contributions) 11:21, 27 December 2016 (UTC)

Jo-Jo Eumerus - Yes: NEWER pages that are NOT patrolled are NOINDEX'ed. — xaosflux ^Talk 14:19, 27 December 2016 (UTC)

For the past few months, the page Endogeneity (econometrics) has had an equation not displaying properly, citing the problem

 Failed to parse (MathML with SVG or PNG fallback (recommended for modern browsers and accessibility tools): Invalid response ("Math extension cannot connect to Restbase.") from server "/mathoid/local/v1/":): E(z_{i}u_{i})={\frac {\gamma _{2}}{1-\gamma _{1}\gamma _{2}}}E(u_{i}u_{i}) 

The syntax was not problematic, and the old versions of the article containing the exact same code displayed the equation properly. Emptying the cache apparently solves the problem, but it does not give us a satisfying explanation, or a way to solve it in the future. When I write this message, the problem seems to have been fixed permanently, but since we don't know how, starting a discussion seemed appropriate. The page chemical equilibrium apparently had similar issues earlier this week. I don't understand the way equations are translated in images enough to make any kind of progress alone. Luis Goslin (talk) 16:16, 19 December 2016 (UTC)

User:Physikerwelt is really good at figuring out these questions. Whatamidoing (WMF) (talk) 23:04, 20 December 2016 (UTC)

@Whatamidoing (WMF): I had a look at the linked pages, but I could not see anything suspicous. I have no idea what to do about a temporary problem, since I can not access the log files. For the future it would be good to capture a screenshot and record time and date so that someone with access to the WMF infrastrucutre (i.e. the services team) can have a look at the log files. I feel sorry that I can not provide better help at the moment.--Physikerwelt (talk) 08:36, 22 December 2016 (UTC)

Thanks for the reply, User:Physikerwelt. I understand the difficulty of dealing with temporary problems. I mainly posted this because it was typed before I noticed the problem was somehow resolved. I did take a screenshot, though. How and to whom can I send it to help ? I uploaded it on my user page, if someone wants to take a look. Luis Goslin (talk) 17:02, 22 December 2016 (UTC)

Services is User:GWicke's team. Whatamidoing (WMF) (talk) 21:36, 22 December 2016 (UTC)

A similar error can be observed at the moment on the converse nonimplication article. If not, I uploaded another screenshot on my user page. Luis Goslin (talk) 01:42, 28 December 2016 (UTC)

On Android tablet, when zooming in to a picture displayed in Media Viewer, the effect is that the caption enlarges to take up more and more space, until only a sliver of the actual picture is left visible. What is required is for the picture only to zoom, if that is possible. 31.49.181.199 (talk) 04:14, 28 December 2016 (UTC)

Thanks for this note. User:Matma Rex probably knows whether this has been reported previously. Whatamidoing (WMF) (talk) 04:17, 28 December 2016 (UTC)

I've never actually worked on the media viewer. I suspect the pinch zoom behavior depends on the browser you're using. Tasks T77620 and T77423 look vaguely related. Matma Rex talk 10:51, 28 December 2016 (UTC)

Hi. I don't know exactly what happened but for some reason the columns setup has gone haywire. I tried to fix but I really am no good at tghis sort of thing and I can't manually retrofit the entire article, literally from A to Z. So would some kindly Yuletide/Chanukah/Kwanzaa spirit help. Thanks. P.S. -- Happy New Year to all. Quis separabit? 17:59, 28 December 2016 (UTC)

Sorry, I see nothing wrong in this article. Ruslik_Zero 19:57, 28 December 2016 (UTC)

@Ruslik_Zero -- thanks but could you just take a look at letters "C" and "D", especially the columns. Thanks. Quis separabit? 21:28, 28 December 2016 (UTC)

(edit conflict) It looks like the columns autospace to the width of the screen versus the length of entries in each column. Mkdw^talk 21:29, 28 December 2016 (UTC)

@Ruslik_Zero -- Thanks. But the list has been there for years and I only just noticed it. Is there any way to fix it? Quis separabit? 21:51, 28 December 2016 (UTC)

Primefac has made some changes to {{Multicol}} as part of a merge with {{Col-begin}}/{{Col-break}}/{{Col-end}}}. Might try asking them. Mkdw^talk 23:04, 28 December 2016 (UTC)

Using such templates causes an accessibility problem, in that each time you use {{col-break}} or similar, you are terminating the list earlier than its true finish, and starting another one. It is better to use a columnar system that emits a single list, and moves the problem of where to break the columns from the editor to the browser, such as {{div col}}/{{div col end}}; you can see this in use at Wikipedia:Meetup/UK#Oxford. --Redrose64 (talk) 23:57, 28 December 2016 (UTC)

As Redrose64 has stated, it's not about the template but its usage. If you put four items in the first column and twelve items in the second, it's going to show it like that. As suggested, use {{div col}} if the list changes often and you want to keep the columns even. Primefac (talk) 00:06, 29 December 2016 (UTC)

Hello. I have found with a query a list of articles . Is there a way to add to the article of these items a specific category? Not by hand. Xaris333 (talk) 00:10, 29 December 2016 (UTC)

Hi all

I'm looking for ideas for a technical solution to allow Visual Editor to be used on Wikiproject pages.

Having VE enabled would make editing pages easier much in the same way as it often makes articles easier to edit, especially for things like tables which are incredibly useful for organising work within a Wikiproject.

However I have not been able to find a solution to enabling VE for Wikiprojects as they share the Wikipedia: namespace with many other kinds of pages, some of which break when someone edits a page with VE. There seem to be two options, perhaps there are more?:

1. Implement VE in a non standard way for Wikipedia: pages e.g in the same way Wikidata does it, using a small pencil icon within Source Editor
2. Move all Wikiproject pages to their own namespace allowing VE and any other tools to implemented that would help Wikiprojects.

Thanks

--John Cummings (talk) 17:38, 29 December 2016 (UTC)

No thanks, and no thanks. This has been discussed often. --Redrose64 (talk) 17:59, 29 December 2016 (UTC)

The current version of VE isn't really suitable to edit WikiProject pages, because WikiProject pages usually contain a lot of code. (((The Quixotic Potato))) (talk) 19:28, 29 December 2016 (UTC)

I have received THIS message over 20 times in the last several days. Though each time I have confirmed, in a day or so, it comes back. I don't believe the issue is on my end, as I have received WP mail recently. How do I make the notifications stop and get someone to actually fix the problem? Thanks! SusunW (talk) 16:19, 23 December 2016 (UTC)

THIS is not a message but a link which displays different things for different users. Special:EmailUser/SusunW says: "This user has not specified a valid email address", and I see no mail link at User:SusunW. Please clarify what the mails say, what you see at Special:ConfirmEmail, what you do and what happens when you do it. Is the problem that you get mails without doing anything, or that you cannot confirm your email address? PrimeHunter (talk) 18:30, 23 December 2016 (UTC)

The message says "Confirm email address You must validate your email address in order to use email features. Click the button below to send a confirmation email to your address. Then, follow the instructions in the email. To check whether you have already confirmed, please see your preferences. A confirmation code has already been emailed to you; if you recently created your account, you may wish to wait a few minutes for it to arrive before trying to request a new code."

Then I get something like this in my e-mail: "Hello SusunW, Welcome to Wikipedia! You've joined the English-language version of the free encyclopedia that anyone can edit. To confirm your email address, please open this URL in your browser: https://en.wikipedia.org/wiki/Special:ConfirmEmail/<long string of characters here> This link expires at 12:41, 30 December 2016 (UTC)."

Which when I press on the link, I get a message in WP that says "Your email address has now been confirmed."

In a few hours, I will get yet another message that says confirm your e-mail and the whole process repeats. My e-mail works fine. I've had the account since the late 1980s. This happens every single time someone e-mails me for the first time with WP, so it isn't my system, I am fairly sure it is a WP issue. SusunW (talk) 18:47, 23 December 2016 (UTC)

I think your email address can get unconfirmed if BounceHandler detects that emails to you don't get delivered. This can an issue on the side of Wikipedia (if it's detecting this wrong for some reason), or on the side of your email provider (if they're in fact bouncing some emails but not other). Can you file a Phabricator task? Someone with access to server logs should be able to see what is happening. Matma Rex talk 01:10, 25 December 2016 (UTC)

Actually, I have access to them myself, I nearly forgot. @SusunW: Your email provider is refusing to accept some automated email from Wikipedia (all of the bounced emails I see are Echo email notifications in HTML format). You might have run out of space in your inbox (try deleting some unneeded emails), or you might have enabled some weird overeager security/spam filters (if you changed any settings recently, try changing them back), or it might dislike HTML email (try switching "Email format" to "Plain text" in Special:Preferences#mw-prefsection-echo), or it might be some general problem with your provider that affects all of its users. But this is really not my area of expertise. If nothing I suggested above helps, please file a task, so that someone who knows more about MediaWiki email stuff will investigate. Matma Rex talk 01:26, 25 December 2016 (UTC)

Matma Rex Thanks! Totally weird but I have copies of identical e-mails in both my inbox and my spam folder. Go figure. At least now I know what to look for. Marked them all as not spam and edited the preferences. Hopefully that fixes the problem. Thanks again. SusunW (talk) 01:53, 25 December 2016 (UTC)

A user has contacted OTRS and reported this issue. I'm not familiar with such points, hence please address it here:

Ticket#: 2016081810005346

"I just tried to convert 'Naive Bayes classifier' wikipedia web to pdf form.

There is small problem that a formula of left column overlapped right column as follow. [Image available at the interface]

Thank you for providing nice service."

Thanks. --Mhhossein ^talk 08:36, 30 December 2016 (UTC)

@Mhhossein: I remember reading somewhere that the Create PDF feature is very underdeveloped, with markup like tables just not rendering properly. Sam Walton (talk) 09:14, 30 December 2016 (UTC)

{{Bookcreatorstatus}} displays a warning in several pages. PrimeHunter (talk) 10:32, 30 December 2016 (UTC)

@Mhhossein: the requester may find that the "printable view" (https://en.wikipedia.org/w/index.php?title=Naive_Bayes_classifier&printable=yes) will meet their needs as a workaround. — xaosflux ^Talk 13:07, 30 December 2016 (UTC)

Thank you all. --Mhhossein ^talk 13:15, 30 December 2016 (UTC)

This project in Phabricator might already have an existing bug report. If not, please feel free to file one. --AKlapper (WMF) (talk) 17:45, 30 December 2016 (UTC)

Dear fellow Wikipedian village pumpers,

VeblenBot is a bot which supports the peer review process, used by thousand sof wikipedians and with generally 30-50 active reviews. The whole process relies on the bot to catalogue new reviews and remove old or closed ones. The bot is not supported and frequently fails for months at a time. We really need a replacement bot and I lack the skills to create or maintain one, but I hope an editor here may be able to help out. We need a committed editor as we have had (at PR) several unfulfilled promises in the past. I have also placed a request at "Bot requests" in the hope that someone can help. Further details are available from the "technical details" peer review page. Thank you for your help! --Tom (LT) (talk) 13:00, 30 December 2016 (UTC)

Probably best to have one discussion at WP:BOTREQ, but, in short, Perl enthusiasts please enquire. - Jarry1250 ^{[Vacation needed]} 22:02, 30 December 2016 (UTC)

The bot that archives my user talk page is overdue for a visit. I set it up years ago and have not touched the settings since. I'd appreciate it someone familiar with bot archiving setups could look over my settings to see if it has somehow become obsolete or broken. Roger (Dodger67) (talk) 08:13, 31 December 2016 (UTC)

On 27 December 2016 it archived 2 threads . It seems to work fine. You probably need to change the settings from 60 days to something like 30 days. (((The Quixotic Potato))) (talk) 08:41, 31 December 2016 (UTC)

@Dodger67: I've updated your settings to have the bot visit your page more frequently. Feel free to play around with the settings and/or ping me if you have any questions. -FASTILY 10:36, 31 December 2016 (UTC)

Thanks Fastily and The Quixotic Potato The posting rate is significantly higher now than it was when I initially set the archiving. Having 80 or sometimes even more "open" topics on the page is excessive, so halving the archive "age" makes sense. Roger (Dodger67) (talk) 14:24, 31 December 2016 (UTC)

I just saw a very interesting article at where the NSA had a secret Wiki based on Wikipedia content. They have an actual Wikipedia mirror, and then they have a page they add their own stuff to. Anyway, for the article Anna Politkovskaya, it contains text very like 2007-2010ish versions of ours, but it is not exactly the same as any of the revisions I have looked at. There is a sentence "She was the daughter of Soviet Ukrainian diplomats posted at the United Nations", for example. I've tried posted at the United Nations in Wikiblame and get nothing, but its output seems to suggest that it only tries a few random revisions rather than all of them. Is that true? Please, someone do us a favor and find exactly when this text was lifted from Wikipedia. I don't know that will shed any light on anything but my nose is twitching. Wnt (talk) 21:36, 29 December 2016 (UTC)

@Wnt: Weird. Unfortunately I am unable to understand the Russian language. UPDATE: you cannot use Special:Export to get all the revisions if there are more than 1000; it only shows the oldest 1000 revisions. You have to use the API or a dump to get them all.

I checked archive.org and this revision contains a very similar sentence:

Politkovskaya was born Anna Mazepa in New York City in 1958 to Soviet Ukrainian parents, both of whom served as diplomats to the United Nations.

Here is another revision:

Politkovskaya was born Anna Mazepa in New York City in 1958, where her Soviet Ukrainian parents were diplomats at the United Nations.

But look what I found in an even earlier revision:

Politkovskaya was born Anna Mazepa in New York City in 1958, where her Soviet Ukrainian parents were diplomats at the United Nations (according to various sources her father was a high-ranking KGB officer[citation needed]).

Here is a website where someone who copypasted the article dated 13.10.2006 08:15:11 (she was killed ~6 days before that).

Note that page 18 of this PDF gives the following information about her father:

Mr. Stepan F. Mazepa

UN Secretariat Building

Room 3580 A

UNITED NATIONS, New York

There is some ambiguity about her name and place of birth (it is an old tradecraft trick to take advantage of the fact that there are different ways to transliterate certain foreign names, and changing someones place of birth makes it more difficult to track that person):

Politkovskaya was born Anna Mazepa in New York City in 1958, the daughter of Stepan F. Mazepa from Kostobobriv (Kostobobrov), Ukraine.

Some sources say that her birth name was actually Hanna Mazeppa.<ref>Halyna Mazepa: My fondest Ukrainian memories are of Katerynoslav, day.kiev.ua</ref>

Other sources state that she was born in Chernihiv (Chernigov) region of Ukraine.<ref>Biography, annapolitkovskayafund.com</ref>

Her parents, Soviet diplomats at the United Nations, were Ukrainian.<ref>Anna Politkovskaya, notablebiographies.com</ref> Politkovskaya spent most of her childhood in Moscow; she graduated from Moscow State University'...

Not sure what this means... It is possible that that part of the article was written by someone on that secret wiki. Happy new year and remember... Putin khuilo! (((The Quixotic Potato))) (talk) 08:52, 31 December 2016 (UTC)

The article export feature sounds useful. If I understand correctly you have actually used it to go through all the history and not found that phrase. To me what is relevant about this is that I was thinking one of two things might be true: a) the NSA was using Wikitext with little revision, which would be very funny for all of us, or b) they have someone rewriting Wiki content (and the "posted to", for example, is indeed a very professional phrase. What the rewriting tells us is actually quite significant: the NSA, being willing to hire high-paid people with classified security clearances to rewrite any source text they could find in the world, went to Wikipedia to start their revisions! In other words, we are the best, not just in terms of free content, open content, public content, but in terms of any database the government can find to work with.

The article content seems like more of a distraction so far. I was only interested in it as an example and I don't see how any omission would indicate NSA editing. If we had an exact time of the download we could look to see if someone made an amateur-hour mistake of editing the content when they took it, but I'm not expecting that to be likely, and we don't have the time to work with. Wnt (talk) 12:18, 31 December 2016 (UTC)

@Wnt: Yeah I was basically using the export feature as a poor man's WikiBlame. It creates a big textfile with all the revisions, and I searched through that. I have tried to date the screenshot by using Google reverse image search and Tineye reverse image search, but neither of them could find anything. They have clearly spend a lot of time and effort rewriting it, because every section contains a classification level. U = unclassified, noforn = no foreigners, et cetera, see Classified information in the United States and Sensitive Compartmented Information for information on how they mark how secret which information is. (((The Quixotic Potato))) (talk) 12:40, 31 December 2016 (UTC) p.s. Of course Wikipedia is the best, that shouldn't surprise anyone!

I don't know if this is relevant but the export facility says it is limited to 1000 revisions but Anna Politkovskaya has 2047 revisions.

Good point, it is. I am surprised that they would make a completely independent fork, which gets outdated quickly, instead of a layer on top of Wikipedia. (((The Quixotic Potato))) (talk) 13:15, 31 December 2016 (UTC)

So is there a way to get the export to "continue from" some point? It disturbs me that Wikipedia has basically allowed this feature to break, especially when so far I see no alternative -- for the average user. Yeah, if you own a supercomputer and can download all the dumps back to the beginning, or can program a web crawler to request every revision one by one, you can do this, but I don't think Wikipedia should be designed only for the Googles -- and NSAs -- of the world to use and not for anybody else. Wnt (talk) 20:21, 31 December 2016 (UTC)

Yep, I figured it out, you cannot use Special:Export to get all the revisions if there are more than 1000; it only shows the oldest 1000 revisions. You have to use the API or a dump to get them all. I'll write a script for this task. (((The Quixotic Potato))) (talk) 20:29, 31 December 2016 (UTC)

Hi! Some people dislike infoboxes. Would it be a good idea to add an option to hide them to Special:Preferences? Basically all it would need to do is add the CSS code .infobox {display:none;}. This may be helpful in reducing the amount of discussions about infoboxes (but maybe I am too hopeful). Not everyone is familiar with CSS, and checking a checkbox and clicking the Save button is more userfriendly than telling them to go to Special:MyPage/common.css and add that line there. (((The Quixotic Potato))) (talk) 23:46, 30 December 2016 (UTC)

If they really dislike them enough, they could add the style to their personal CSS themselves. --Unready (talk) 00:28, 31 December 2016 (UTC)

True, but those who are in favor of userinfoboxes are generally not nerdy enough to make that recommendation, and those who oppose them are generally not nerdy enough to know that this is even an option. (((The Quixotic Potato))) (talk) 00:36, 31 December 2016 (UTC)

Please avoid extending the infobox wars which have nothing to do with what someone wants to see in their browser. To put it another way, adding a gadget is not going to happen without evidence that "some people" exist, and even then it won't be considered unless there are many people. Mentioning infoboxes and userboxes shows some confusion. Johnuniq (talk) 02:02, 31 December 2016 (UTC)

Well, I was trying to reduce drama, not extend it. TBH IDGAF, I don't have a dog in this race, I have never added a infobox or deleted one. I have corrected the mistake, thank you for pointing it out. Seems like this is unlikely to happen. Oh well, I tried. (((The Quixotic Potato))) (talk) 02:33, 31 December 2016 (UTC)

@Johnuniq: I am kinda curious why you wrote: "the infobox wars which have nothing to do with what someone wants to see in their browser". What do you mean? I know very little about the infobox wars, but it seems to be a logical assumption that is is about what someone wants to see in their browser (or the opposite, what someone does not want to see). I wish there was a Wikipedia History book. If you mean that these conflicts are always more about interpersonal relations than the actual 'topic' of the dispute then I understand that. (((The Quixotic Potato))) (talk) 02:35, 31 December 2016 (UTC)

Then they can either learn to do it through personal CSS, or their technical luddism is itself an indication of how ludicrously worthless this grudge against infoboxes is. Andy Dingley (talk) 02:51, 31 December 2016 (UTC)

Hehe. ok. (((The Quixotic Potato))) (talk) 02:56, 31 December 2016 (UTC)

People using the Wikipedia mobile app see infoboxes in a collapsed state but they can be expanded. Also, cleanup templates and other unwelcome material are hidden. I rather like this arrangement. Thincat (talk) 13:15, 31 December 2016 (UTC)

On behalf of the Luddites, I should say that I don't generally approve of hiding and unhiding stuff on the page. If you have to download it, you might as well see it. If it is large and optional, we ought to have some other page. We have a whole category system, and there's no reason why we can't have other dedicated clickable navigation pages for excessive infobox material. There is also a category system, which badly needs to be upgraded to the early 2000s with things like non-fixed page structure and length. And I still get junk hits every now and then, even though it's not supposed to happen, on web searches where every possible wrong answer is indelibly linked to whatever it was you were trying to find because of an infobox. Wnt (talk) 23:50, 31 December 2016 (UTC)

If you have to download it, you might as well see it. Well, then the gadget idea is useless, because something hidden by CSS is still downloaded. Using the CSS suggested, there isn't even an option to display it later, if desired. --Unready (talk) 23:54, 31 December 2016 (UTC)

I have no strong opinions one way or another, but I’ve read enough of the infobox discussions to have noticed quite a few arguing against them on behalf of the general reader, ostensibly at least. So nothing that only affects logged-in users is likely to answer all the objections.—Odysseus1479 07:17, 1 January 2017 (UTC)

Anyone know if sock suspects get notifications when an SPI is opened? A few weeks back I opened a report on someone and they showed up to respond, which I thought was very odd, because I hadn't notified them, and it's extraordinarily rare for sock suspects to respond spontaneously to SPIs unless they've watchlisted the SPI or something. So I was curious if anything like the {{checkuser}} template might be generating the notification the way {{u}} might. My hope is to get a confident, "Yes, template X generates a ping" or "No, they shouldn't have been pinged--I think they were lying to you" or something of that sort, if possible. Thanks! Cyphoidbomb (talk) 18:08, 30 December 2016 (UTC)

User gets pinged, if you include in one edit your signature (with timestamp) and link to userpage of particular user (which can be done with {{checkuser}}, {{u}}, {{ping}}, plain link [[User:Edgars2007]] etc.). --Edgars2007 (talk/contribs) 18:27, 30 December 2016 (UTC)

I don't think {{checkuser}} generates a ping, but let's find out: Cyphoidbomb (talk⁺ · tag · contribs · deleted contribs · logs · filter log · block user · block log · CA · CheckUser(log) · investigate · cuwiki · SI). If not, then more likely they're watching the page and were notified by some other means. Ivanvector (^Talk/_Edits) 18:29, 30 December 2016 (UTC)

Yeah, I think I got pinged there, Ivanvector. Thanks for helping me reason that through. Thanks also, Edgars. Cyphoidbomb (talk) 18:32, 30 December 2016 (UTC)

I see. Well, that's good to know then. Ivanvector (^Talk/_Edits) 18:33, 30 December 2016 (UTC)

{{checkuser}} includes a link to {{user}}, so that presumably generates the ping. --David Biddulph (talk) 18:39, 30 December 2016 (UTC)

Templates don't generate "pings" (I don't like that word, in my line of work it has a very different and somewhat older meaning). What matters is that there is a link to a user page, it doesn't matter if that link is template-generated, nor if one template transcludes another. I have explained this many times, on this page and at WT:Echo, and I would put the explanation into Wikipedia:FAQ/Technical if I could make it suitably concise yet comprehensive and understandable. --Redrose64 (talk) 20:34, 30 December 2016 (UTC)

Sure they do (def 3). --Izno (talk) 21:34, 30 December 2016 (UTC)

It's a management buzzword that originated when an engineer's boss overheard the engineer using that term, and misunderstood what they were talking about. Happens every day. --Redrose64 (talk) 22:05, 30 December 2016 (UTC)

You forgot to wikilink engineer. --Unready (talk) 00:21, 31 December 2016 (UTC)

You can modify the relevant templates to use {{noping}}. Legoktm (talk) 09:47, 1 January 2017 (UTC)

Is there any way to transclude a specific revision of a template, rather than just defaulting to the current revision?

I am discussing the effect of some changes on navboxes, and it would be handy to be able to illustrate their effects by transcluding a series of specific revisions. --BrownHairedGirl (talk) • (contribs) 16:49, 1 January 2017 (UTC)

@BrownHairedGirl:, short answer: no; long answer: not currently (phab:T70399). — xaosflux ^Talk 17:24, 1 January 2017 (UTC)

Thanks, Xaosflux. I can't be the first person to need this, so I hope that the functionality gets added. --BrownHairedGirl (talk) • (contribs) 17:28, 1 January 2017 (UTC)

The {{vanchor}} template states "This template should not be used within section headings. Doing so will result in broken links in edit summaries, as well as duplicate anchors" in a comment added in 2015 by an IP editor. On the other hand, {{anchor}} seems to assume that it will be used in section headings, though it does say that some may prefer using bare html. I don't see any reason why there should be any difference in the use of these templates. If so, it seems the documentation of the two templates should be aligned. Am I correct or am I missing something here? in replying, please {{ping}} me as I don't plan to watchlist this page. Thank you. YBG (talk) 05:54, 1 January 2017 (UTC)

Without going over the doc, I would say that use of {{vanchor}} for section headings is redundant. MediaWiki inserts a visible anchor when text is marked as a section heading. On the other hand, there may be several reasons for using an editor-inserted non-visible anchor. 204.19.162.34 (talk) 15:05, 1 January 2017 (UTC)

(edit conflict) The comment about duplicate anchors is presumably referring to how if you have a heading like == {{vanchor|something}} ==, there will be two anchors for "#something" in the page. On the other hand, == {{vanchor|something}} something else == will have an anchor for the whole "something something else" and an anchor for just the "something", so no duplicate in that case.

The comment about broken links in edit summaries is referring to the fact that templates aren't expanded when generating the automatic edit summary for editing a section, so the automatic edit summary would wind up as something like /* Section name{{anchor|foo}} */

HTH. Anomie⚔ 15:08, 1 January 2017 (UTC)

I think I have found the answer to my question at WP:SPECIFICLINK, which suggests inserting {{anchor}} before the section heading, which avoids the need for either {{anchor}} or {{vanchor}} in section heading. YBG (talk) 23:05, 1 January 2017 (UTC)

I was searching earlier for something on Wikipedia and couldn't find it so went to Google. Followed the link back to Wikipedia and found I was logged out (however I am still logged in on regular pages). I then noticed that the address was "en.a0.wiki" (as in https://en.a0.wiki/wiki/Main_Page) and not "en.wikipedia.org". What is the a0? Also it says I'm centrally logged in and to reload the page to apply my settings, which didn't do a thing. Best thing was I got the IPv6 address (http://whatismyipaddress.com/ip/2400:f100:2:0:4c45:aeff:fef4:4a58) which says I'm in Hong Kong. As it's −32 °C (−26 °F) I wouldn't mind being there rather than here. I realise this is a really trivial question. CambridgeBayWeather, Uqaqtuq (talk), Sunasuttuq 08:51, 31 December 2016 (UTC)

Wikipedia is at wikipedia.org. en.a0.wiki is a live mirror not listed at meta:Live mirrors. .wiki is a top-level domain from 2014. https://www.whois.com/whois/a0.wiki says a0.wiki is registered to "Zhou Shi Peng" in China and not the Wikimedia Foundation like https://www.whois.com/whois/wikipedia.org. Do not enter your password at a0.wiki or other mirrors. They may claim to be the real Wikipedia but that is at wikipedia.org. PrimeHunter (talk) 10:59, 31 December 2016 (UTC)

Please explain how you "followed the link back". It is quite conceivable that companies like Google could start working with China or another country and start providing links to a local censored server in place of Wikipedia links - I don't think this happened, but we should rule it in or out. Wnt (talk) 12:55, 31 December 2016 (UTC)

I had no intention of logging in there especially as it was clear I was still logged in here. I was looking for some information on Wikipedia on logging into WP:AWB that I knew I had seen but couldn't find. I searched for "awb two factor authentication" and got this. I checked the 3rd and 4th result then realised that the 5th was the one I remembered. Clicked on it before I saw the a0. CambridgeBayWeather, Uqaqtuq (talk), Sunasuttuq 13:18, 31 December 2016 (UTC)

CambridgeBayWeather in case you never found your answer - see this page: Wikipedia:Using AWB with 2FA — xaosflux ^Talk 20:45, 31 December 2016 (UTC)

Xaosflux. Thanks I did eventually find that but now I can't remember what I did to get it running. CambridgeBayWeather, Uqaqtuq (talk), Sunasuttuq 06:41, 1 January 2017 (UTC)

A major factor in this is that the site is using Wikipedia's trademarked logo (image located at https://en.a0.wiki/static/images/project-logos/enwiki.png - archive of image and Wikipedia's trademarked name in identifying itself in browser history as "Wikipedia, the free encyclopedia" and at the top of the page (archived image). These are clear legal no-nos, giving rise to confusion about site identity, and certainly could give the impression that they are trying to mislead people logging in, possibly damaging Wikipedia's security either directly or if their server were to be compromised. Certainly they are vulnerable to action here, though I would hope this is merely an accident of someone copying an archive without much thought. See also WP:Wikipedia logos. Wnt (talk) 23:26, 31 December 2016 (UTC)

If it looks like a trademark violation, contacting WMF Legal is welcome. Thanks! --AKlapper (WMF) (talk) 10:04, 2 January 2017 (UTC)

Recently, when editing some pages (talk pages, user pages, project pages), I've had this "Signing" checkbox show up under the edit window much of the time. When it does, it will also change my edits to some degree, e.g adding a sig where it thinks there should be one, adding newline characters at wrong places (this page is transcluded in a link URL on my main userpage), and showing me a dialog box asking if I want to save despite not having signed (which I do every time it asks). How can I disable this? עוד מישהו Od Mishehu 15:37, 2 January 2017 (UTC)

I'm not seeing a "Signing" checkbox anywhere. I wonder if you're somehow pulling in a user script that's doing this, although I don't see anything that looks likely in User:Od Mishehu/common.js. Anomie⚔ 17:52, 2 January 2017 (UTC)

@Od Mishehu: - try turning off your global signing script in meta:User:Od_Mishehu/global.js. — xaosflux ^Talk 18:14, 2 January 2017 (UTC)

Yes, thank you. עוד מישהו Od Mishehu 22:38, 2 January 2017 (UTC)

Section headings of Wikipedia:Citing sources contain numerous span tags such as <span>id="Adding the citation"</span> <span>id="Inline reference"</span> <span>id="Footnotes and references"</span>. What is the purpose of these tags? Is there a list of these codes somewhere? Why does the same id appear multiple times in the page? Is there some place where this is documented? I have looked everywhere I can think of. Thanks. Comfr (talk) 21:44, 1 January 2017 (UTC)

The syntax is <span id="Adding the citation">...</span>. It creates an anchor by the given name so you can make the link Wikipedia:Citing sources#Adding the citation. See WP:ANCHOR. PrimeHunter (talk) 21:53, 1 January 2017 (UTC)

Thanks PrimeHunter. Comfr (talk) 23:47, 1 January 2017 (UTC)

@Comfr: The same id must not appear multiple times in the page - ids must be unique within a given page. On a closely-related matter, I have started a thread at Wikipedia talk:Manual of Style/Linking#Bad advice on anchors. --Redrose64 🌹 (talk) 00:55, 3 January 2017 (UTC)

File:Reina restaurant Istanbul.JPG is currently on the Main Page's "In the News" section. However, go to File:Reina_restaurant_Istanbul.JPG#File_usage and you'll see nothing about it being used anywhere on this wiki except for 2017 Istanbul nightclub attack and Wikipedia:Main Page history/2017 January 1. Why is the current Main Page missing? This state of things would be expected if the Main Page were linking to a redirect to the file, but neither Special:WhatLinksHere/File:Reina restaurant Istanbul.JPG nor c:Special:WhatLinksHere/File:Reina restaurant Istanbul.JPG mentions any files that redirect to this one. Nyttend (talk) 01:26, 4 January 2017 (UTC)

Since Main Page and {{In the news}} actually display File:Reina restaurant Istanbul (cropped).JPG. They use link= to link File:Reina restaurant Istanbul.JPG instead of the displayed version. PrimeHunter (talk) 02:05, 4 January 2017 (UTC)

"The Crash (Mad Men)", a page I created, was added on December 31st, but Tool Labs reports that it was receiving views before then. How is this possible? Thanks, AndrewOne (talk) 04:52, 3 January 2017 (UTC)

Views of non-existing pages are counted and stored in a database, e.g. when somebody clicks a red link. But it appears the linked tool will only display data if the page currently exists. See User:West.andrew.g/Popular redlinks (last updated in July). By the way, please add the page to The Crash. PrimeHunter (talk) 11:26, 3 January 2017 (UTC)

People do sometimes search for topics that don't exist yet, or click on redlinks that exist in other articles (e.g. because they think maybe they're going to buckle down and start the article, but then change their mind.) So those redlink "views" would still count toward the pageview total. Bearcat (talk) 17:18, 4 January 2017 (UTC)

It seems not functional since few days. Only the user bar is visible. I use Chromium with almost no addons. --Rezonansowy_akaRezy (talk | contribs) 21:28, 2 January 2017 (UTC)

Loads for me. Try to disable it in Special:Preferences#mw-prefsection-betafeatures, clear your cookies and cache, log back in, reenable and try again. — xaosflux ^Talk 22:11, 2 January 2017 (UTC)

@Xaosflux: Hmm, still doesn't work. It may be a conflict with some of scripts loaded from my common.js or global.js. --Rezonansowy_akaRezy (talk | contribs) 15:14, 3 January 2017 (UTC)

User scripts/gadgets are the most common problem, especially if they assume (without checking) that all pages contain a content area. You can differentiate between common.js problems (and enwiki gadgets) and global.js problems by seeing whether it loads for you at another Wikipedia. Go to, e.g., w:ht:Special:ContentTranslation and see what happens. (The Haitian Creole Wikipedia is particularly convenient for this kind of testing, because it has no local gadgets.) If it opens there, then the problem isn't your global scripts.

If the problem is only with local scripts, then I believe you could go to another Wikipedia to do the translation work. You can go to, say, the German Wikipedia to translate a Spanish Wikipedia into English. Whatamidoing (WMF) (talk) 18:02, 4 January 2017 (UTC)

Background: from the Help Desk I found the page Simon Nwakacha which is currently tagged G11 (for good reason). I suspected the topic could be salvageable, hence set off for a web search.

Right now, a DuckDuckGo search for "Simon Nwakacha" does not turn up the Wikipedia article, which is the expected behaviour (pages are noindex-ed until a page patroller approves them). However, it does display a box with the current (or recent) contents of the page (see DuckDuckGo#Overview). This is probably not intended at least from Wikipedia's part.

If DDG just fetches the content of the WP page if the search string returns something else than a 404 error, there is probably not much to be done from our side (except maybe send them a polite request). However, I guess it could be due to a misconfiguration of Wikipedia leading to one of DDG's robots accessing supposedly no-indexed pages, so I am mentioning it there. I see no such behaviour from Google or Yahoo, FWIW. Tigraan^{Click here to contact me} 17:49, 4 January 2017 (UTC)

Reported: . MER-C 03:00, 5 January 2017 (UTC)

Several times over the last two or three days, I've had problems when following links to commons:. The usual error message is:

Wikimedia 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 213.120.234.154 via cp3037 frontend, Varnish XID 874052594
Upstream caches: cp3037 int
Error: 404, Requested domainname does not exist on this server at Wed, 04 Jan 2017 10:54:50 GMT

Is this a reported problem? --Redrose64 🌹 (talk) 11:05, 4 January 2017 (UTC)

The HTTP 404 referred to in my first post here is intermittent, and today it happened for my third visit to Commons. My second visit was OK; for my first visit, see above. Firefox 50.1.0 --Redrose64 🌹 (talk) 13:18, 4 January 2017 (UTC)

@Redrose64: Not just Commons, if I'm not mistaken English Wikipedia was just down for about four minutes and I don't think it was just me based on the pause in the main recent changes list... EoRdE6^{(Come Talk to Me!)} 23:22, 4 January 2017 (UTC)

It also affected meta: for several minutes, but it was not the same problem as before - the returned message was somewhat like this, which is very different from the 404 message. --Redrose64 🌹 (talk) 00:00, 5 January 2017 (UTC)

This was wikitech:Incident_documentation/20170104-MonologSpi. — This, that and the other (talk) 05:45, 5 January 2017 (UTC)

At my recent nomination of a variant of this idea at another venue, two commentators whose "home wikis" are in other languages stated their belief that the necessary CSS exists for displaying lock icons according to the protection status of a page. Can anyone here confirm that this is the case? Samsara 10:56, 2 January 2017 (UTC)

"already done with css" was said by a Dutch editor who may have misunderstood a Dutch Wikipedia feature. nl:MediaWiki:Gadget-ProtectionTemplates.js is a default Dutch gadget enabled for IP's. It checks for protection and automatically displays a protection template. Like all gadgets it requires JavaScript in the browser, changes the page after the rest has loaded, creates an overhead, and doesn't run in Mobile. A feature in MediaWiki itself would be better. PrimeHunter (talk) 11:57, 2 January 2017 (UTC)

(e/c) As far as I remember, not pure CSS, but JS. It is possible to retrieve the protection level using wgRestrictionEdit variable in javascript. Based on that, you could automatically add an indicator. And several wiki's indeed do this (and so does the mobile skin btw). The only problem is that it's not visible for people who don't use Javascript. We could also in the core of the software, add an option to automatically output a page indicator or something I guess. —TheDJ (talk • contribs) 12:05, 2 January 2017 (UTC)

That indeed would be (and was) my proposal, but it came in at no. 50 out of the 250+ proposals, and only the first 10 to 20 will actually get actively worked on (at least that's the current and official pronouncement and held true for last year's outcome), so it would need another public push to actually get implemented. Samsara 16:33, 2 January 2017 (UTC)

While I don't necessarily oppose this in principle, it would have to be per-page overridable and highly configurable, as there are cases in which protection templates are not used or desirable, such as user pages. Any discussion about enabling this also needs to consider the precise criteria for an automatic template, or mandate a follow-up discussion to do so. BethNaught (talk) 16:40, 2 January 2017 (UTC)

I see per-namespace configurability as a given, since protection and PC are already configured on this basis. As for the other issue, I would like to see a case where a page was protected, a clerk (for lack of an official title) came along to place the template that was originally omitted, and the clerk's change was then reverted. Does it happen in actual practice? Samsara 17:00, 2 January 2017 (UTC)

I've reverted such a "clerk's change" on several occasions. I don't have any recent examples to hand, but here is an example of where I reverted the addition of a prot icon that had been added by the protector of the page. --Redrose64 🌹 (talk) 15:20, 3 January 2017 (UTC)

...your edit summary mentioning that automatic display is already being used in template space. Samsara 18:53, 3 January 2017 (UTC)

It's not fully automatic. It relies on the presence of either {{collapsible option}} or {{documentation}}, both of which contain code to detect the current prot level and its expiry. These don't use either CSS or JS to obtain that data - they use some Lua code that queries the page information. So in the absence of both of those templates, another means of displaying a padlock should be used. --Redrose64 🌹 (talk) 23:30, 3 January 2017 (UTC)

Yeah, but Lua code is server side. So, CSS or no CSS, there's apparently a very easy way to get there. Samsara 03:48, 4 January 2017 (UTC)

See also Wikipedia:Bot requests#Add protection templates to recently protected articles. --Redrose64 🌹 (talk) 23:30, 3 January 2017 (UTC)

I was later wondering if maybe that the code used for protection templates in {{collapsible option}} and {{documentation}} could be used in a template used in articles, such as {{Reflist}}, but that template is used on some non-articles, and thus would cause pp templates to appear on non-articles, such as user pages, and some users might not want pp templates showing on pages in their userspace. —MRD2014 (talk • contribs) 00:20, 5 January 2017 (UTC)

Evaluation can be made conditional on namespace. Samsara 07:57, 5 January 2017 (UTC)

For the better part of the entire past year, the Untagged Uncategorized Articles tool on the WMFlabs Toolserver has been cluttered with a set of phantom articles, mostly deleted titles which don't even exist at all but also including a smaller number of pages which are properly categorized. However, nothing I can do to the pages in question makes them drop from the list: restoring and redeleting or vice versa doesn't work, creating a new temporary page with Category:Temporary maintenance holdings on it doesn't work, null-edits don't work, nothing works. I've asked on multiple occasions, both here and on the Toolserver side, for assistance in getting this fixed, but no matter where I ask the buck gets passed as a problem at the other end — if I ask here it just gets dismissed as a problem with Toolserver's database-replication copy, while if I ask over there they can't find a problem and it gets dismissed as corruption in the data that en is passing to Toolserver — so nothing ever actually gets done to fix it. The tool's official maintainer, User:JaGa, hasn't been able to fix it either — and he hasn't been on Wikipedia at all since September, so approaching him again isn't an option.

The titles in question are: Andreas Nödl, Axiom Landbase Pvt. Ltd., BatissForever, Bhaiyato The Hobbit, Bingham Road railway station, Byjus classes, Carrhotus malayanus, Changed people, Changed person, Cyclone Victor(1986), Dee Sterling, Divegrass, Divisor theory, Eddie Mao, Egger Island, Escalo Frio, Faithlessly, Fimbriata, Hyperbolic Geometry:Poincaré half plane model, James Steinkamp III, January 2 in India, January 3 in India, Kris Reeder, Levent Karahan, Marius Kižys, Maxipad Detention, Mercurialsoft inc, Oscar Möller, Peter Johnston (negotiator), Roadjammers, Roofing Hub, Taxation in Armenia, Thunder2D, Tirupati college of education, Tractors India Pvt. Ltd. (TIPL) and Đuli.

The last time a similar problem occurred, for the record, there appeared to be a number of ghost metadata files for the offending pages, which had them coded as articlespace pages even though they were actually in project or user or template space (see User talk:JaGa/Archive 17#New toolserver problem for a bit more background if needed.) I'm wondering if that may also be the problem here, but I have no idea how to figure that out. And, for the record, some (though not all) of them are still turning up on Wikipedia:Database reports/Untagged stubs as well, even though that page's most recent run was months after the pages were deleted, so I suspect the issue is indeed on this side or that wouldn't be happening either.

Accordingly, I wanted to ask if anybody is willing to actually look into this in depth, and do a thorough and holistic investigation on both en and Toolserver to figure out what's actually happening and how to actually fix it, instead of just passing the buck as "their problem, not ours, the end". This really needs to be fixed, because I can't just keep working around the problem into infinity — I really need somebody to actually solve this. Thanks. Bearcat (talk) 17:08, 4 January 2017 (UTC)

All of those except for Đuli appear to be instances of phab:T138967: they either have a page table row in the Tool Labs database despite not existing here or they exist here but with a different page_id (you can check the page_id with the API). Đuli has page_id 45504597 in both places, although I see the Tool Labs database is missing some rows in the categorylinks table for that page. Anomie⚔ 01:58, 5 January 2017 (UTC)

Bearcat, as you note, deleting and undeleting doesn't work, but what about deleting and not undeleting? I created a placeholder at egger island, used Special:MergeHistory to move the contents there, and deleted Egger Island; it's still deleted. Does the page still show up in the report? It did just now when I viewed it, but it also still shows up as blue in this discussion section, so I can't rule out a simple caching situation. Nyttend (talk) 05:02, 5 January 2017 (UTC)

Unfortunately, no, that didn't clear that title off the list. Bearcat (talk) 17:40, 5 January 2017 (UTC)

I just noticed this, but don’t know how long it has been like this. The left margin within the watchlist, where the “m”, ”b”, etc. flags go, is far too wide for this purpose. It’s wide enough for the whole of January above it to fit in it, and seems gratuitously to be taking up space. It does not matter which flags are displayed (and I only ever see those for minor and bot edits). The screenshot at the right shows the problem. This is with Safari on OS X, with e.g. all settings (such as browser zoom) at defaults and the default vector skin. It looks nothing like the screenshot at Help:Watchlist which allocates no space for that column.--JohnBlackburne^words_deeds 15:21, 5 January 2017 (UTC)

JohnBlackburne Is this only occurring when you set Group changes by page in recent changes and watchlist... on ? — xaosflux ^Talk 17:55, 5 January 2017 (UTC)

Yes, that fixes it. Odd, I don’t remember ever setting that, not even sure what it does. Suddenly disabling it I can see Wikidata changes, which I turned on a while back but never saw. I did look on the Watchlist tab of preferences for anything that might make a difference, but did not think to look at any other tab. Thanks.--JohnBlackburne^words_deeds 18:10, 5 January 2017 (UTC)

It puts all changes for a page that appear on your watchlist in the same spot (pages sorted by most recent change) rather than sorting globally by most recent change. This can be combined with another preference ("enhanced recent changes) which collapses the set of diffs with Javascript. Indeed, Wikidata changes do not appear with this option turned on--this is phab:T46874. --Izno (talk) 18:16, 5 January 2017 (UTC)

So for it to work you probably need 'Expand watchlist to show all changes, not just the most recent' enabled on the Watchlist tab. Otherwise you won’t see anything, except a wide margin and no Wikidata changes. Seems rather poorly designed, with multiple interrelated options spread over two tabs, and a jarring design change for no obvious reason enabled by one of them.--JohnBlackburne^words_deeds 18:26, 5 January 2017 (UTC)

I often refer to sources that publish less than once a month, like "Winter/Spring 1983". Unfortunately, the |date= parameter does not handle these sorts of formats. Yes, I know I can manually reformat it into a form that will work, but isn't that what computers are for? Is there any technical reason the template does not handle these? Maury Markowitz (talk) 14:45, 28 December 2016 (UTC)

See this help page for valid formats. In your case, something like this should work:

"Article about something". Title of Magazine. Winter–Spring 1983.

Will that work for you? If not, you can always hack it by putting the information into |issue=, but there are strict constructionists around here who will frown on that workaround. – Jonesey95 (talk) 16:01, 28 December 2016 (UTC)

That will work, technically, but means a) that I am not using the original format, and b) have to figure out how to type that character. Is there any reason / can't be used if that dash can? Maury Markowitz (talk) 20:45, 29 December 2016 (UTC)

The cite templates follow MOS:DATERANGE. – Jonesey95 (talk) 21:08, 29 December 2016 (UTC)

In the case of issue months or numbers, it‘s not necessarily a range, often rather a combination or notionally double issue. Here, for example, it might indicate the magazine’s having previously changed its publishing schedule from quarterly to semiannual. Maybe a distinction without a difference, though.—Odysseus1479 06:59, 1 January 2017 (UTC)

Slashes are also problematic because of the possible ambiguity. Does "Winter/Sprng" mean [during] "Winter or Spring" or "Winter & Spring"? A point also made @ MOS:SLASH. 64.134.64.162 (talk) 19:28, 5 January 2017 (UTC)

@Maury Markowitz: See Wikipedia:How to make dashes. Personally, I use the CharInsert edit tool, it's between the edit box and the edit summary box. --Redrose64 (talk) 22:14, 29 December 2016 (UTC)

I don't want to be that guy, but this is clearly broken, why not just fix it? Maury Markowitz (talk) 00:04, 30 December 2016 (UTC)

@Maury Markowitz: You have made valid points but will need a wp:RfC to allow other real-world dates (unless date in parameter "issue="). Also common "2016-05" has been rejected as May. The problem seems to be people think wp:MOS format needs to be enforced by templates, but that elevates MOS wp:guideline as if a wp:policy and bullies people to comply as abandon real-world professional dates. Early January tends a poor time for RfC, when many educated people, busy post-holidays, have less time to debate issues. Try RfC later date. There are intelligent people here but...busy. -Wikid77 (talk) 06:54, 5 January 2017 (UTC)

AFAIK, citation dates do not need verbatim formatting, so stating "Winter–Spring 1983" rather than "Winter/Spring 1983" is not material. 72.43.99.146 (talk) 15:48, 31 December 2016 (UTC)

So what? There is no reason not to support this commonly used format. The wikipedia should be supporting its editors, not straightjacketing them. Maury Markowitz (talk) 13:51, 5 January 2017 (UTC)

The correct venue for requesting changes to the cs1|2 templates is at Help talk:Citation Style 1. Since cs1|2 obeys the date format requirements established by MOS:DATEFORMAT, if and when the MOS is changed to allow a new publication date format, then cs1|2 can be modified to support it.

—Trappist the monk (talk) 14:15, 5 January 2017 (UTC)

Why using the unit conversion template to convert 370cm and 390 cm to inch result in same inch number? 370 cm (150 in) 390 cm (150 in) C933103 (talk) 11:19, 4 January 2017 (UTC)

Rounding, a controllable feature explained in detail in the {{convert}} documentation. Converting your values to the first decimal place, 370 cm (145.7 in) and 390 cm (153.5 in), we can see that the conversion template is actually converting correctly. But your values only have two significant figures, so the results are likewise rounded to two sig-figs by default (145.7 or 153.5 to the nearest tens is 150). DMacks (talk) 12:00, 4 January 2017 (UTC)

(edit conflict) See Template:Convert#Rounding: 100 ft is 30 m or 30.5 m or 30.48 m? 370 and 390 have two significant digits. 37 cm and 39 cm both round to 15 in. You can for example round to the nearest integer with |0: 370 cm (146 in), 390 cm (154 in). PrimeHunter (talk) 12:03, 4 January 2017 (UTC)

@PrimeHunter and DMacks: I see.but in this historical section, if you preview this, you can see 370cm for A320 cabin and 390cm for C919 cabin and Mercure fuselage get rounded, however 373cm for DC8 fuselage and 376cm for Boeing fuselages are not rounded. Why is that? Is that because the system incorrectly assumed the last zero being insignificant? Is there more intellectual approach to this?C933103 (talk) 05:09, 6 January 2017 (UTC)

Please use the template talk page to ask about {{convert}}. However, since we started here, the converts you mention are:

• {{Convert|370|cm|in|abbr=on}} → 370 cm (150 in)
• {{Convert|390|cm|in|abbr=on}} → 390 cm (150 in)
• {{Convert|373|cm|in|abbr=on}} → 373 cm (147 in)
• {{Convert|376|cm|in|abbr=on}} → 376 cm (148 in)

The inputs with two significant figures (370 and 390) are giving two significant figures in the output, and the inputs with three significant figures (373 and 376) are giving three significant figures in the output. That is good. In the unlikely event that it is known that 370 means 370±0.5 the desired result could be obtained with either of these:

• {{Convert|370|cm|in|0|abbr=on}} → 370 cm (146 in)
• {{Convert|370|cm|in|sigfig=3|abbr=on}} → 370 cm (146 in)

Johnuniq (talk) 09:32, 6 January 2017 (UTC)

Is there any good reason to use one of common.js or vector.js, or a combination of both? If I understand right, common.js will load extensions in all environments while vector.js will only load them while using the vector skin, but if I only ever use vector anyway, is there any reason I might want to move some/all of my scripts (see User:Ivanvector/vector.js if it helps) over to common.js (which I don't currently have)? Like is there a performance impact or anything? Ivanvector (^Talk/_Edits) 11:44, 6 January 2017 (UTC)

If you ever did use a different skin, what scripts do you load that you would want in all skins? Put those in common. Put ones you want only in vector in vector. There's no performance difference. --Unready (talk) 15:10, 6 January 2017 (UTC)

That's pretty much what I thought. Thanks! Ivanvector (^Talk/_Edits) 15:14, 6 January 2017 (UTC)

Some scripts work in all four current skins, and these are suitable for common.js: but some will only work in one or two, so those should go in vector.js or whichever is appropriate. --Redrose64 🌹 (talk) 16:05, 6 January 2017 (UTC)

Some images from upload.wikimedia.org do not load for me or take a while to load. This has never happened to me before. Conor2004 (talk) 17:08, 6 January 2017 (UTC)

I'm experiencing the same thing Autumn Wind (talk) 17:25, 6 January 2017 (UTC)

Might be related to Wikipedia:Village pump (technical)#Commons server problems - HTTP 404. --Redrose64 🌹 (talk) 20:46, 6 January 2017 (UTC)

Hello, I was wondering if anyone was able to offer insight into why many edits seem to reorder fields in the country infobox. See as an example. I've noticed this for awhile but never thought to ask about it until now. It is however annoying as the new order is a bit nonsensical, and can make future editing a bit more difficult. Thanks, CMD (talk) 09:51, 7 January 2017 (UTC)

@Chipmunkdavis: VisualEditor reorders the parameters based on TemplateData in the template's documentation. It may be a good idea to change the order of the parameters in the TemplateData to match the documentation's blank example. Jc86035 (talk) Use {{re|Jc86035}}
to reply to me 09:58, 7 January 2017 (UTC)

Perfect, thank you very much. CMD (talk) 10:11, 7 January 2017 (UTC)

@Chipmunkdavis: No problem :) Jc86035 (talk) Use {{re|Jc86035}}
to reply to me 13:23, 7 January 2017 (UTC)

I cannot get Draft:Bishal Ruidas to appear in Category:AfC G13 eligible soon submissions. Purging, null editing, and the "forcelinkupdate" API parameter all did not work. GeoffreyT2000 (talk, contribs) 17:09, 7 January 2017 (UTC)

@GeoffreyT2000: Isn't it only five months since you moved it in August? Pppery 17:15, 7 January 2017 (UTC)

Yes, which means that the category will appear on the page if one purges it. When doing a null edit, the category disappears because it thinks that {{REVISIONTIMESTAMP}} refers to the time of the null edit rather than the last saved revision in the page history. The API parameter only updates link tables, not category tables. GeoffreyT2000 (talk, contribs) 17:30, 7 January 2017 (UTC)

Sorry, I though you were talking about the category for actual G13-elligible submissions, not Category:AfC G13 eligible soon submissions, which I didn't know existed. Pppery 17:33, 7 January 2017 (UTC)

I don't have a smartphone, so I've only occasionally seen Wikipedia on one. I've been thinking about the case of a new editor who was first blocked, then indeffed for repeating an action they'd been warned about on their talk page, and who has never edited anything except article space. They don't have e-mail enabled and the mini-orange bar is kind of small and the red numbers even smaller, and it occurs to me that all their edits are marked as mobile and using the mobile interface. What do such users see when they have a talk page message? Is it possible for them not to notice? And how easy is it for them to find their talk page if they do see the message and access the notification? (Since they don't have e-mail enabled, they won't be getting e-mail alerts, presumably, although I believe that's now a default setting for those who do enable e-mail.) Does anyone have a screenshot or any other clarification? Yngvadottir (talk) 06:59, 4 January 2017 (UTC)

An editor gets a nice red indicator at the top red, which is pretty hard to miss. See to the right side —TheDJ (talk • contribs) 08:59, 4 January 2017 (UTC)

I didn't notice it until i vread the comment in which you mentioned it, so it's quite possible that a newcomer (having less of an idea than me about how Wikipedia works) would miss it, too. עוד מישהו Od Mishehu 09:39, 4 January 2017 (UTC)

I do have two iPhones and they also display a red tick like that one. Jo-Jo Eumerus (talk, contributions) 09:40, 4 January 2017 (UTC)

OK, thanks to all. So they get the red numbers but not the little orange "new message" bar. If they then click on the red number, how easy is it for them to reach their talk page? Is it similar to what I experience on desktop view—the notifications screen loads slowly and then displays a list of who's left messages on the talk page, each of which is clickable? Since they are blocked, their talk page is the only place they can now edit, and they are actually required to copy the unblock template, but several of us have been hoping they will make any kind of response there, so the next issue is whether they can easily figure out how to. Yngvadottir (talk) 15:40, 4 January 2017 (UTC)

The difficulty of figuring out how to deal with complex, template-based workflows is one of the reasons that WP:Flow was invented. Anyone who's watched unblock requests knows how many newbies get it right, even on desktop systems. It's a complicated workflow no matter what your device is. So imagine that the editor doesn't just get a message with technical instructions. Imagine that the message has a "Respond" button built into it, maybe with a few tickboxes (e.g., "I want to appeal this block" vs "I don't want to appeal this block right now"), and that this button adds the template, cats the page, and pings the admin – and all the newbie has to do, or even all the newbie can do, is provide the content for the message. Or imagine that it has a timer to remind the blocking admin to check back after a week (a day, a month, a year) to see whether an indef is still necessary (or whether a re-block has been earned).

But that's the idea for the future, and before we can do the fancy stuff with Flow, it has to be able to handle conversations. They're still working on some of that (e.g., making it possible to find comments made in Flow via Special:Search). WhatamIdoing (talk) 18:41, 4 January 2017 (UTC)

I'm afraid that doesn't answer my question. Flow destroys our ability to use talk pages for article collaboration and as a result the community has decisively rejected it. Even if it hadn't, completely reconfiguring all talk pages is clearly not a viable solution right now. So I'm still asking—right now, what is this editor's path to their talk page? If they do whatever passes for clicking on the smartphone version of the red number they see, can they get there to read the messages, and can they then see an edit button of some sort? Because if not, we need to treat editors flagged as using mobile differently, perhaps by feeding them a big and obvious link in warning templates. I'm going to ping in Cullen328 at this point, because he's been exploring both editing interfaces on his smartphone and has experience explaining the results to technical incompetents like me. Yngvadottir (talk) 19:31, 4 January 2017 (UTC)

I know that you have always held firmly to the belief that Flow cannot be used for collaboration. However, as a point of data, I have personally used Flow to collaborate on creating and improving multiple pages on multiple wikis. So it clearly can be done: I have done it, and so have many other people. WhatamIdoing (talk) 19:44, 7 January 2017 (UTC)

Yngvadottir, I use the desktop view by default on my smartphone the vast majority of the time, but occasionally check out the mobile view, and did so just now. In desktop view, I have a highly visible orange bar telling me I have messages, plus a less prominent little bell icon with the number 2 telling me that I have been pinged or thanked. In mobile view, as mentioned above, the orange bar is absent, but the bell icon is more prominent and has a visible red box with the number 2 overlaid. So, if a user understands that they should click that box, they will see a list of pings, notifications of talk page messages and so on. The problem is that some new users may ignore the red box on mobile, but then again, some ignore the orange bar on desktop Cullen³²⁸ Let's discuss it 22:44, 4 January 2017 (UTC)

When one clicks on the red tick, one should have a list of notifications. These link to the pages they are on, i.e the talk pages. Also, plenty of projects work with Flow, not that transition here would be easy. Jo-Jo Eumerus (talk, contributions) 20:15, 4 January 2017 (UTC)

@Jo-Jo Eumerus and Cullen328: Thanks for confirming. It does sound as though the editor has simply not clicked on the message icon, but it's possible they don't know how to respond. Is the path to the talk page as clear on mobile as it is on desktop (I have no idea how the click mechanism works on a phone, and as I say, for me at least, notifications load agonizingly slowly), and is there then an "edit" button? I dimly recall from the early days of the mobile interface hearing that mobile users don't see a tab for article talk pages. I'd like to be sure this editor has just failed to click through and respond, rather than needing instructions on how to do so. There are going to be increasing numbers of new editors using mobile only, so despite my own preferences and limitations, I'm concerned. Yngvadottir (talk) 22:58, 4 January 2017 (UTC)

I found that talk page button at the bottom of each page a while ago. I think it's there under the theory that phones are better for reading than for editing, so a person who accesses Wikipedia on phone is less likely to need the talk page than one who accesses Wikipedia on desktop. Jo-Jo Eumerus (talk, contributions) 23:07, 4 January 2017 (UTC)

The massive problem with that theory, Jo-Jo Eumerus, is that it becomes a self-fulfilling prophesy. It is 100% possible for people to edit productively using a smartphone, and I know that's true because I do so myself. I have written many articles including Good articles, participated in thousands of AfD debates, answered hundreds of questions at the Teahouse, all on my Android smartphone. There are literally billions of Android smartphones in use today, and failure to encourage worldwide smartphone editing is a gigantic mistake. Please read my essay called Smartphone editing for further information on this critical issue. Cullen³²⁸ Let's discuss it 01:12, 5 January 2017 (UTC)

I've lost the cite button and there are a number of scripts (all in one place) that aren't working. The one script that works reliably is DYKcheck. Twinkle, enabled through preferences and not JavaScript, loads intermittently only. Bypassing my cache doesn't fix any of this. Dropping the URL for a page in edit mode into a private browsing session does two things: (a) it shows the cite button (yeah!) but (b) it logs me out. Logging back in within that private environment, the cite button is gone. The problems are the same whether I'm using my laptop or my iPhone. Bandwidth shouldn't be the issue (31 Mbps down and 9.4 up). Given I have the problem across two devices, I'm somewhat reluctant to clear my cache. Any suggestions? Schwede66 22:03, 7 January 2017 (UTC)

You have 'User:AlexTheWhovian/script-redlinks.js' enabled and this throws an error for me. Might be related, or not. —TheDJ (talk • contribs) 22:18, 7 January 2017 (UTC)

		A beer on me!
		TheDJ deserves a beer; problem fixed! Schwede66 22:37, 7 January 2017 (UTC)