The technical section of the village pump is used to discuss technical issues aboutWikipedia. Bugs and feature requests should be made at Bugzilla (see how to report a bug). Bugs with security implications should be reported to securitywikimedia.org or filed under the "Security" product in Bugzilla.
This tends to solve most issues, including improper display of images, user-preferences not loading, and old versions of pages being shown.
No, we will not use JavaScript to set focus on the search box.
This would interfere with usability, accessibility, keyboard navigation and standard forms. See task 3864. There is an accesskey property on it (default to accesskey="f" in English). Logged-in users can enable the "Focus the cursor in the search bar on loading the Main Page" gadget in their preferences.
No, we will not add a spell-checker, or spell-checking bot.
You can use a web browser such as Firefox, which has a spell checker.
If you changed to another skin and cannot change back, use this link.
Alternatively, you can press Tab until the "Save" button is highlighted, and press Enter. Using Mozilla Firefox also seems to solve the problem.
If an image thumbnail is not showing, try purging its image description page.
If the image is from Wikimedia Commons, you might have to purge there too. If it doesn't work, try again before doing anything else. Some ad blockers, proxies, or firewalls block URLs containing /ad/ or ending in common executable suffixes. This can cause some images or articles to not appear.
Currently here's what I have: "You have 8,545 pages on your watchlist (excluding talk pages)."
I know it's possible to edit the raw watchlist, but I'd like the option to purge it as I look through the currently active list. Can a box be added beside each line so that I can check all those I want to get rid of and thus clear the list as I go along each day? -- Brangifer (talk) 05:59, 17 October 2013 (UTC)[reply]
Isn't Special:EditWatchlist what you need (that's the link for "View and edit watchlist")? If doing bulk changes (like deleting all items that match a particular search), you would use Special:EditWatchlist/raw, copy the text into an editor, get it how you want, paste it back into the edit box, and click Update. However, I don't think there's anything realistic you can do once you have a few thousand items, apart from severely pruning almost everything. Johnuniq (talk) 06:12, 17 October 2013 (UTC)[reply]
I think Brangifer is looking for something that can be used directly from the "view relevant changes" screen, i.e. the regular watchlist. I use popups for this - you can mouse over the article link and then click the "un" link from "un|watch" in the "actions" menu. However, maybe someone's written a user script that makes an actual button/checkbox/link on the watchlist page itself. Is anyone aware of such a thing? — Mr. Stradivarius♪ talk ♪06:30, 17 October 2013 (UTC)[reply]
Yes, that's what I'm looking for. I already know about the Special:EditWatchlist, but I'm only interested in pruning the no longer relevant ones which actually keep popping up on my watchlist when I'm looking at recent changes. A box would be nice so everyone has access to it on their own watchlist, without having to install a script. -- Brangifer (talk) 14:25, 17 October 2013 (UTC)[reply]
This may be more work than you want to do, but here's what I do (this on on a Mac, in Firefox, so the keyboard commands may be different for you). Hold down the Command key and click on each of the pages that I want to unwatch. This opens a bunch of tabs in the background. Close the watchlist and position your mouse arrow over the blue star that indicates that the article is watched. Click it, then use Command-W to close the tab (without moving the mouse). Click the star on the next article, Command-W, click, Command-W, etc. You can unwatch 50 articles in a couple of minutes if you get into a rhythm. – Jonesey95 (talk) 15:04, 17 October 2013 (UTC)[reply]
I'm familiar with that method and have used it, but I want something easier. I already use Firefox most of the time and often have over 20 tabs open. -- Brangifer (talk) 06:32, 18 October 2013 (UTC)[reply]
As mentioned on Wikipedia talk:Geonotice#Shut it off?? this was due to a misconfiguration in our new caching datacenter (ulsfo). I've fixed the issue, but it'll take an hour to propagate through DNS. For the time being I've moved traffic for oceania back to eqiad. When DNS is finished propagating, I'll move the traffic back to ulsfo.--Ryan lane (talk) 01:27, 20 October 2013 (UTC)[reply]
Rehman If you disable javascript in your browser, the French version stops working, and comes out as a vertical strip of images like your sandbox currently does -- so there's likely some default javascript at French Wikipedia that's handling that effect, and likely something we don't currently have here. In that case you wouldn't be able to get your template working here no matter how errorless the template code is :) I'll attempt to find the JS code for theirs, and maybe we can get it added here -- cause that does look like an interesting and useful template. equazcion→12:26, 22 Oct 2013 (UTC)
No problem :) It looks like the French common.js has the following (copied it to my userspace): User:Equazcion/DiaporamaFrench.js, which seems to be what's handling those slideshows. I was hoping this was a MediaWiki extension but it rather seems to be something custom they cooked up there (at least from what I could ascertain), so we'd need to translate the code and add it here ourselves.
I'd like to find out if this would actually happen (adding this to the site's default JS) before going ahead with any effort to make it work. Anyone have any thoughts? equazcion→13:29, 22 Oct 2013 (UTC)
This sort of thing has come up before, it might be worth looking in the archives for past discussion. One question that comes to mind is whether the printed output is sensible. Anomie⚔01:18, 23 October 2013 (UTC)[reply]
Well, specifically was wondering what "recent" meant (recent edits, recent authors). Apparently 30, from Whatamidoing's link. NE Ent11:24, 20 October 2013 (UTC)[reply]
There are so many internet images that are meant to contain 60° angles whose angles are a lot less perfect than they could have been. For example, the image in the Wikipedia article Eisenstein prime is vertically stretched as shown in the first image with the blue line having an exact 60° angle from horizontal. This is how to contruct a line with an exact 30° angle from horizontal as shown in the second image. You will see the imperfections in the image better if you view them at full resolution. I don't know if it's possible for scalable vector graphics to add a feature of automatically generating a line with an exact 30° angle from horizontal or vertical using a fractal like piecing together of that line. Furthermore, any image with a line with an exact 30° or 60° from horizontal should automatically scale up to a larger image with a larger line with an exact 30° or 60° from horizontal as well as a line slightly clockwise of that orientation always scaling up to another line slightly clockwise of that orientation and a line slightly counter clockwise of that orientation always scaling up to another line slightly counterclockwise of that orientation. Due to the quantization of possible line orientations for each size, it's possible for a line to be constucted in the first place at an orientation such that there exists no possible smaller line that can be scaled up to a line that size with that orientation. Blackbombchu (talk) 02:52, 20 October 2013 (UTC)[reply]
Your examples are both PNG files, which are not vector graphics, nor are they truly scalable. In true Scalable Vector Graphics, drawing a line at 30° - or any other angle - is very easy:
The density is calculated from population and area values -- the Density parameter is ignored. I've updated the documentation of the template. NE Ent11:51, 20 October 2013 (UTC)[reply]
Oh I know, but calculation is still wrong. It is calculate in round figure. If you divide 4,572,951 (district population) from 3029 (district area) then it will be around 1510.--Prateek MalviyaTalk05:05, 21 October 2013 (UTC)[reply]
Yes, it rounds it, but not to the nearest integer - or even the nearest ten. It first does the exact calculation:
{{#expr:4572951/3029}} → 1509.7230108947
Then it puts that figure through {{Order of magnitude}} and subtracts the result from 1:
{{#expr:1-{{Order of magnitude|1509.7230108947}}}} → -2
This is then used as the second argument to {{rnd}}:
I posted this to the Wikimedia Commons pump and was advised to ask here. I'm trying to use the API to get info on many files, but in some cases, the filenames are long, as they are mostly in unicode and thus need urlescaping. For example, this url requests info for fewer than 50 pages, but as you'll see from the link, it returns the blanket error page. The same error seems to happen if I use cURL to stick the query string in the data part of a "POST" request. But both POST and GET methods start working if I reduce the URL to fewer than 8188 chars (oddly, this limit seems to apply to the urlencoded string, so that what counts is the string length of the urlencoded (%) characters, even if the request is made with non-encoded characters). This hints to me that it is a limitation in the MediaWIki software, rather than Apache etc. I can't find any reference to a maximum url length in the API docs. Can anyone point me in the direction of some API documentation about this? HYanWong (talk) 09:15, 20 October 2013 (UTC)[reply]
There are many places that such a limit could be applied. The limit could be in the caching layer (squid or varnish), or it could be in apache (possibly squid/varnish is re-encoding the URL when forwarding, or possibly it's a rewrite rule re-encoding it), or it could be in various places in PHP, or it could be in MediaWiki. Considering the error received from that particular URL, I'd guess the operative limit is in squid. As noted, doing it as a POST rather than a GET will work fine. Anomie⚔11:29, 20 October 2013 (UTC)[reply]
Thanks for that. Odd that I couldn't get POST working on my end, but I probably messed up something in cURL. I'll probably just stick to GET (as requested on the API page, to allow caching), but check on the title lengths and make multiple GET requests. It looks like there aren't any obvious bits of documentation I can point when commenting my code. Oh well. Thanks again for the help. HYanWong (talk) 13:23, 20 October 2013 (UTC)[reply]
Via the Commons pump, Bawolff kindly stuck a note on mw:API:FAQ. But as I've also asked there, I'm still getting problems (of a slightly different type) with long titles and urlescaped POST data. The query I posted above works fine for me if proper unicode characters are used in the POST data, but try, for instance, using curl http://commons.wikimedia.org/w/api.php --data @filewithURLescapedPOSTDATA.txt to send the following string as POST data to the API: action=query&format=json&prop=imageinfo%7Ccategories&iiprop=url%7Cmime%7Cmediatype&clprop=hidden&cllimit=500&redirects&titles=File%3A2010.+%D0%92%D1%8B%D1%81%D1%82%D0%B0%D0%B2%D0%BA%D0%B0+%D1%86%D0%B2%D0%B5%D1%82%D0%BE%D0%B2+%D0%B2+%D0%94%D0%BE%D0%BD%D0%B5%D1%86%D0%BA%D0%B5+%D0%BD%D0%B0+%D0%B4%D0%B5%D0%BD%D1%8C+%D0%B3%D0%BE%D1%80%D0%BE%D0%B4%D0%B0+01.jpg%7CFile%3A2010.+%D0%92%D1%8B%D1%81%D1%82%D0%B0%D0%B2%D0%BA%D0%B0+%D1%86%D0%B2%D0%B5%D1%82%D0%BE%D0%B2+%D0%B2+%D0%94%D0%BE%D0%BD%D0%B5%D1%86%D0%BA%D0%B5+%D0%BD%D0%B0+%D0%B4%D0%B5%D0%BD%D1%8C+%D0%B3%D0%BE%D1%80%D0%BE%D0%B4%D0%B0+06.jpg%7CFile%3A2010.+%D0%92%D1%8B%D1%81%D1%82%D0%B0%D0%B2%D0%BA%D0%B0+%D1%86%D0%B2%D0%B5%D1%82%D0%BE%D0%B2+%D0%B2+%D0%94%D0%BE%D0%BD%D0%B5%D1%86%D0%BA%D0%B5+%D0%BD%D0%B0+%D0%B4%D0%B5%D0%BD%D1%8C+%D0%B3%D0%BE%D1%80%D0%BE%D0%B4%D0%B0+08.jpg%7CFile%3A2010.+%D0%92%D1%8B%D1%81%D1%82%D0%B0%D0%B2%D0%BA%D0%B0+%D1%86%D0%B2%D0%B5%D1%82%D0%BE%D0%B2+%D0%B2+%D0%94%D0%BE%D0%BD%D0%B5%D1%86%D0%BA%D0%B5+%D0%BD%D0%B0+%D0%B4%D0%B5%D0%BD%D1%8C+%D0%B3%D0%BE%D1%80%D0%BE%D0%B4%D0%B0+09.jpg It works if you remove any one of the titles from the list (doesn't matter which), so I don't think it's to do with a weird character in the unicode escape sequences. What am I doing wrong? HYanWong (talk) 13:14, 21 October 2013 (UTC)[reply]
How odd. I'm getting the same html formatted page as the GET error: "Our servers are currently experiencing a technical problem", with
Request: POST http://commons.wikimedia.org/w/api.php, from 81.155.213.67 via amssq41.esams.wikimedia.org (squid/2.7.STABLE9) to () Error: ERR_INVALID_REQ, errno [No Error] at Mon, 21 Oct 2013 18:09:26 GMT
using curl 7.19.7 (universal-apple-darwin10.0) libcurl/7.19.7 OpenSSL/0.9.8y zlib/1.2.3. I've just tried pasting into a different text editor in case, but get the same. And since it's a POST request, I assume it shouldn't be cached anywhere. But if it's just my setup, I guess it's not worth documenting. HYanWong (talk) 18:14, 21 October 2013 (UTC)[reply]
Update, I've just tried it on a completely different system (curl 7.20.1 (amd64-portbld-freebsd8.0)), based in a different city (so no caching), and it fails with the same error. So it's not just my home setup. HYanWong (talk) 18:27, 21 October 2013 (UTC)[reply]
There seems to be a limit on the size of the post body. If you start randomly deleting characters from the end, at some point you start getting a reply. Since it is squid that is throwing the error, I suspect the limit is somewhere in squid actually. That would mean that if you are logged in you would bypass the problem. —TheDJ (talk • contribs) 18:51, 21 October 2013 (UTC)[reply]
While there is a limit on the size of the post body, that limit (as far as I can tell) is 100M so that's not the problem with this 1K post. The problem is that curl sends an Expect: 100-continue header by default which the squids can't handle. If you give curl an option --header 'Expect:' to suppress this header, the post works fine. BJorsch (WMF) (talk) 19:19, 21 October 2013 (UTC)[reply]
The --header 'Expect:' does the trick. Thanks a lot. I still don't understand why it works for a smaller POST body without that option to curl, but I'm happy to go with this fix. Thanks for all the help. HYanWong (talk) 22:37, 21 October 2013 (UTC)[reply]
Regex newline matching doesn't seem to work on MediaWiki page content
In a userscript attempt, I'm hitting a wall when I need to retrieve a multi-line regex match:
varrequest={action:"query",titles:wgPageName,prop:"revisions|info",intoken:"edit",rvprop:"content",indexpageids:1,dataType:"xml",format:"xml"};$.get(mw.config.get("wgScriptPath")+"/api.php",request,function(response){varcontent=$(response).find('rev').text();varregexObject=newRegExp('== Sample title ==[\s\S]*');varmatch=regexObject.exec(content);});
[\s\S]* is usually supposed to match everything including new lines, but here it's only returning the title line: == Sample title ==. The "match" variable should basically be retrieving the entire page. I'm running this on User:Equazcion/sandbox2, which contains some pasted content from WP:VPP. Does anybody have any idea what I'm missing here? equazcion→17:27, 20 Oct 2013 (UTC)
When using a constructor (as opposed to a regexp literal), you have to double escape, [\\s\\S], since otherwise the single slash is "sucked up" by the built-in string handling ('\s' => 's'). Theopolisme(talk)17:53, 20 October 2013 (UTC)[reply]
Ahh, that works! You rock, thanks Theopolisme :) equazcion→18:09, 20 Oct 2013 (UTC)
ie10 supports columns. my guess is that you activated inadvertently the "Compatibility mode", which cause ie10 to emulate ie7. if this is what really happened, you can notice that the "broken page" icon in the address line turns blue. peace - קיפודנחש (aka kipod) (talk) 18:09, 20 October 2013 (UTC)[reply]
Latest tech news from the Wikimedia technical community. Please inform other users about these changes. Not all changes will affect you. Translations are available.
You can now create and edit references inside media captions. [2]
You now need to press the "delete" key twice to delete a template, reference or image; the first time, they only become selected, to avoid accidental deletion of infoboxes and similar content. [3]
When you resize images, you will now still see them, and their size will also be seen in the center. [4][5]
MediaWiki 1.22wmf22 was added to test wikis on October 17. It will arrive to non-Wikipedia wikis on October 21 and all Wikipedia wikis on October 24 (calendar).
Within the past 24 hours, I've noted that Special:Whatlinkshere is showing spurious links to pages in the article namespace whose titles are identical to those of a template. For example, the article Bone transcludes {{Cleanup}}. Special:Whatlinkshere/Cleanup now includes Bone as the first linking article, even though Bone does not include a link to the mainspace page Cleanup, and also includes several hundred other articles that do not contain links to the mainspace page but do transclude the template. Special:Whatlinkshere/Template:Cleanup also (correctly) includes Bone. Neither Bone nor Template:Cleanup has been edited recently, so this seems like it has to be caused by a software change.
Using the API, I can find that there are entries in the pagelinks table showing links from Bone to both Cleanup and Template:Cleanup. Nonetheless, examining the wikitext of Bone with all templates expanded shows that no such link should exist. Also, although many pages that transclude the template now show such spurious links to mainspace pages, not all of them do.
The same thing can be found in backlinks to pages whose titles correspond to many other commonly-used templates, including Update, Tone, and POV.
This is a tough one. I couldn't see any unusual html output, and I couldn't see any links to mainspace articles like Cleanup in the expanded template wikicode. I've reverted back to the old version of Template:Ambox for now until I can find exactly what went wrong. I did see a discrepancy in the talk page link code between the old template and the Lua version, so it might have something to do with that. At any rate, the spurious links should now disappear after a null edit. — Mr. Stradivarius♪ talk ♪12:06, 21 October 2013 (UTC)[reply]
That should be a documented feature. The reason is the same as why the #ifexists parser function records as a page link: so that when the targeted page changes existence, MediaWiki can know to update the referring page. Also, BTW, using title:getContent() is recorded as a transclusion of the title, and accessing title.fileExists on a File- or Media-namespace title counts as an image usage. BJorsch (WMF) (talk) 02:02, 22 October 2013 (UTC)[reply]
Note the function is "#ifexist" (and #ifexists is a dud); not being a purist, but we don't want people adding "-s" and complain the if-functions are broken now, so: "#ifexist" no ess. -Wikid7704:14, 23 October 2013 (UTC)[reply]
I've now fixed the module and switched back to the Lua version of {{ambox}}. Hopefully it should all be working as it should now, but again, let me know if there are any problems. Also, note that while there should no longer be any article-space links for the |name= parameter, there will still be a link from Special:WhatLinksHere/Foo if the code {{ambox|talk=Foo}} is used. — Mr. Stradivarius♪ talk ♪12:57, 23 October 2013 (UTC)[reply]
We could use template coder eyes at Template talk:Pp-template#Update for template protection, so we can finally get that new pink lock to show up and differentiate template-protected pages. Check over the current sandbox2 version and share your thoughts on whether or not it's ready and what else might need to be done. From what I can tell, this is pretty much the final step in implementing template-protect.
This template is called by {{documentation}} when protection is present on a template, so these changes could potentially affect many pages. Though assuming we do it right, visible changes would only show up for previously indef'd templates that have been downgraded to template-protect. equazcion→14:09, 22 Oct 2013 (UTC)
Rollback button on Contributions special page
I was curious if there's a way to disable the rollback button or hide it from view when on a user's contributions page. I know there is an option in preferences to disable it for the watchlist, just curious if there is a way for the contribs page too. I often view WP on my phone and I've miss-clicked on the RB button on the watchlist in the past and would like to prevent that from possibly happening in the future, albeit on the contribs page. I can't see myself rolling back an edit without viewing it first anyway. Thanks, —dainomite15:36, 22 October 2013 (UTC)[reply]
Oh, Equazcion and Writ Keeper. I implemented the css change first, cleared my cache and nothing changed, the rollback button was still there. Then I implemented the jss change and still nothing. After that I started to think some more and went through my twinkle settings at Wikipedia:Twinkle/Preferences. I saw that under the "Show rollback links on these pages:" section the box marked "Contributions pages of other users" was checked. I unchecked it and saved my preferences and the rollback buttons were still there and went back to my twinkle preferences and noticed the box stayed checked no matter if I unchecked and saved. So I raised the question at WT:TWINKLE to see if that was a bug or something. Any input or comments would be much appreciated. —dainomite02:16, 26 October 2013 (UTC)[reply]
Do you know if you're seeing the Twinkle rollback link or the "real" rollback links? The Twinkle ones are bold and light-blue colored, while the real ones look like ordinary links. You could try going into User:Dainomite/twinkleoptions.js and removing this line: "showRollbackLinks": [], then go back to the twinkle preferences page and try unchecking the box again. equazcion→02:26, 26 Oct 2013 (UTC)
Yes, Commons redirects to the Wikimediafoundation. When I tried to log in, I got the error message: "Login error. There is no user by the name "Maile66". Check your spelling." Maybe I don't have an account with the foundation.— Maile (talk) 17:48, 22 October 2013 (UTC)[reply]
I am also seeing this problem. Sometimes the URL takes you to Commons, sometimes it takes you to Wikimedia foundation. Some sort of load balanced server error? — Amakuru (talk) 17:57, 22 October 2013 (UTC)[reply]
Yes, I get that error if I go to [8]. Another problem: if I try to go to a page in the project namespace, Commons sometimes removes the namespace from the page title, so I end up at "Village pump" instead of "Commons:Village pump". Is anyone else having this problem? --Stefan2 (talk) 18:07, 22 October 2013 (UTC)[reply]
Me too. I was trying to go to the description page of File:Alder Dam.JPG and for a while it would only go to a WMF page only, one that I'm not a member of. Now it seems fixed though, and the description link goes to the Commons. Soranoch (talk) 18:17, 22 October 2013 (UTC)[reply]
Server/Software problems, started about 93 minutes ago. Try to open your Watchlist..... Was completely broken when it started (no activity on Commons for 20 minutes). --Denniss (talk) 18:41, 22 October 2013 (UTC)[reply]
I had to clear my browser's cache to stop it being redirected to Wikimediafoundation when visiting links I visited when it was broken. Colin°Talk18:59, 22 October 2013 (UTC)[reply]
Right now, I have the most trouble going to my contributions in the commons and going to the commons category "Chevrolet vehicles." I'm kind of afraid to clean my cache though, because if I did, it would remove my ability to enter a lot of sites, and some of them have passwords I can't keep track of. ---------User:DanTD (talk) 17:16, 26 October 2013 (UTC)[reply]
About gadget "CharInsert"
Hi,
here, on en.wiki, you have the gadget (extension) CharInsert. It's nice. I also want this extension on my home wiki - on romanian wikipedia, but we don't have there this gadget.
I want to ask: can somebody develope an userscript, so i can use this thing on ro.wiki to? :) Thanks. // XXN (talk) 10:16, 23 October 2013 (UTC)[reply]
It doesn't seem to work for me. I've tried enabling and disabling all the edit toolbar preferences, and no combination seems to make it work. I think there's another custom variable somewhere that creates the new group which you would then be able to insert characters to. VanIsaacWSVexcontribs22:40, 23 October 2013 (UTC)[reply]
i.e., "=" instead of ":". it would be even nicer if you use "window.charinsertCustom", though in most cases this is not mandatory. it seems that in at least one of your tries you actually had it right - i think that if you return to this version and do deep refresh several times, the magic will happen. peace, קיפודנחש (aka kipod) (talk) 23:22, 24 October 2013 (UTC)[reply]
Look in the history. I've tried every permutation of equals and colon. I saw the colon syntax on a user page when I searched for "charinsertCustom", and that's where I couldn't figure out anywhere else to go. VanIsaacWSVexcontribs01:04, 25 October 2013 (UTC)[reply]
please note that i actually linked to one of your good edits. if you go back to this one and refresh enough times, you'll see it's working. since my previous message i tested it, and i can verify it does. peace - קיפודנחש (aka kipod) (talk) 03:20, 25 October 2013 (UTC)[reply]
I don't think that's a gadget or anything else related to Wikipedia. It looks more like a browser add-on that lets you search by clicking words on any page. This is an example, but there are many of these things in existence. equazcion→16:52, 23 Oct 2013 (UTC)
When decades actually begin and end. Or: counting, and what zero means.
I discovered on the 1976 article that all year articles contain incorrect mathematics: The article incorrectly states that 1976 is the 7th year of the 1970's decade. But we all know that counting begins at the number 1. So, even though it looks wrong, the 70's began on January 1st 1971. Therefore ... 1,2,3,4,5,6 ... that means 1976 is the 6th year of the seventies decade. Simple stuff really. There is no "year zero," but there is the 10th year in a series. And that 10th year belongs to the original series, not the next. Meaning 1970 is the final year of the 60's, etc. This incorrect math is seen elsewhere on Wikipedia:
"... the 1000th and last year of the 2nd millennium , 100th and last year of the 20th century, and 1st year of the 2000s decade."
Well, that contradicts itself quite obviously. Because by normal counting rules inherent to the fabric of space-time, the last of a series can't also be the first part of a series. Or: zero can't be 1, and 1 can't be zero. The year 2000 is the last year of the 1990's. The decade begins on January 1st, 2001.
So, I couldn't find where to edit 1976 to correct the math. And when I started looking around for this flaw, I noticed that it's everywhere. If there is any kind of absurd objections to my intention to correct basic math mistakes, perhaps a professional mathematician can step in and provide basic counting lessons. Then perhaps a script could be generated to automatically correct the errors, rather than having to manually edit each one.
I'm surprised this has escaped detection for this long. But obviously, basic math should be correct on an information project that calls itself an "encyclopedia."
I think that would be a prefect example of pedanticism, and would really not serve the encyclopedia well in practice. Although your math may be technically correct, common usage of decades doesn't follow strict math. If we made the correction you suggest, Wikipedia would likely conflict with information in the vast majority of sources. Which would be bad. equazcion→17:01, 23 Oct 2013 (UTC)
(edit conflict) It is inconsistent, but long-standing convention on Wikipedia is that when speaking of decades, we ignore the last digit of the year. So the 1970s runs from 1970 to 1979. But with centuries, the 20th century runs from 1901 to 2000. More at WP:CENTURY. --Redrose64 (talk) 17:01, 23 October 2013 (UTC)[reply]
Actual encyclopedias encode math correctly, there is no reason why Wikipedia would invent byzantine math rules on it's own. And yes, the math is technically correct, and that is what an information source like Wikipedia should be: accurate. Wikipedia should be brave enough to conflict with sources that are plainly wrong. Real encyclopedias (as Wikipedia is trying to be) are the source of information and do not have to follow incorrect math just to be in harmony with "the vast majority of sources."
And: now that it's proven that the "long-standing convention on Wikipedia" is technically incorrect, it's time to fix it. In reality, we don't "ignore the last digit of the year." As I described in my lesson how to count above, the 1970's run from 1971 to the end of 1980. The 20th century runs from 1901 to 2001 (because centuries by definition have 100 years). No words or grammatical inventions will change how basic math works in reality.
So, we are in agreement then about how to count. Numbers begin at 1, and the first unit of the next series of 10 begins with 11. All Wikipedia dates need to be corrected to represent the reality of how numbers actually work, not how the common mind misunderstands them. Common minds will gradually begin to understand how numbers work when this flaw is corrected. An online information source such as Wikipedia is highly degraded when compliance with fuzzy common misunderstandings is held to a higher standard than basic facts.
But this issue is so simple, perhaps it seems complex. Perhaps at this point a professor of mathematics should explain how counting works. Thank you for reading. — Preceding unsigned comment added by 50.136.29.18 (talk) 17:29, 23 October 2013 (UTC)[reply]
And tell all the people who celebrated the start of the millennium on 1 January 2000. The maths isn't wrong but society has decided it want to count 0-9. NtheP (talk) 18:53, 23 October 2013 (UTC)[reply]
This is more of a language convention issue than a mathematical one. The "courage" argument could be made for many things here, and the reason Wikipedia doesn't take such initiative is that its very principle is not to. "Common misunderstandings", when it comes to such conventions, are rather common here for that reason. We have WP:COMMONNAME, for example, which describes our practice of using names commonly associated with topics rather than names that could be said to be the most technically accurate. I'm sure many scientists and professors cringe at some of our article names, since they represent misconceptions that have worked their way into common use, and that they've been trying to dig out of their students' heads for years. Nevertheless, verifiability, not truth is the standard here (more or less). We follow the sources for a given topic, rather than seek to correct them based on our own research. equazcion→18:53, 23 Oct 2013 (UTC)
He was asking if something technical could be done to correct a problem that he already presumed was present. At least that's how I see it. equazcion→19:57, 23 Oct 2013 (UTC)
This is a matter of words and their definitions, not of math. And the current articles have it right. The "first year" means the the first one year time period. Counting years counts the end of those time periods, and age is the amount of time span from the zero point. North8000 (talk) 19:19, 23 October 2013 (UTC)[reply]
Please if you will post the link(s) to the original discussions regarding this technical question. I don't want to waste anybody's time here. I'll just read what's there and not make a fuss. I would talk from an account, but I don't like to argue. Really, this one just seemed like a huge over-ripe piece of low-hanging Wikipedia logic-flaw fruit ready to be cut up and shared with a laugh. Enjoy, and thanks ! 50.136.29.18 (talk) 19:35, 23 October 2013 (UTC)[reply]
I don't think anyone understands what you mean about posting links. I was referring to your original question above, when I said it was originally a technical matter (in case you thought I meant this was referring to a past discussion). equazcion→19:48, 23 Oct 2013 (UTC)
Actually, you are factually wrong. 1976 is the seventh year of the 1970s. It is, however, the sixth year of the 198th decade. Just like 2013 is the 14th year of the 2000s, but the 13th of the third millennium. When you are dealing with ordinals vs. cardinals, the rules are different. VanIsaacWSVexcontribs23:25, 23 October 2013 (UTC)[reply]
Wikimedia Error
I have been trying to make an edit to this page, List of auxiliary Interstate Highways, and have received several timeouts that resulted in the standard Wikimedia Error page that reads...
Our servers are currently experiencing a technical problem. This is probably temporary and should be fixed soon. Please try again in a few minutes.
You may be able to get further information in the #wikipedia channel on the Freenode IRC network.
The Wikimedia Foundation is a non-profit organisation which hosts some of the most popular sites on the Internet, including Wikipedia. It has a constant need to purchase new hardware. If you would like to help, please donate.
If you report this error to the Wikimedia System Administrators, please include the details below.
Error: ERR_READ_TIMEOUT, errno [No Error] at Wed, 23 Oct 2013 18:23:51 GMT
I isolated it to this section of that page, List of auxiliary Interstate Highways#Auxiliary Interstates, by copying and pasting that section to my sandbox. I tried to make a dummy edit, but timed out to the Wikimedia Error page. This happens when I try to preview the edit. After timing out twice, I tried to edit a different page and had no problem on preview. Then I tried again to edit this page and once again timed out. I used IE10 and WIN8 for the initial edit attempts. Then I switched to Firefox24. This time I tried to save the page after edits, but again received the Wikimedia Error page. This time the bottom section read...
If you report this error to the Wikimedia System Administrators, please include the details below.
Seems the edits are getting saved even though an error gets returned, FYI. I made a successful dummy edit, and see one from User:Paine Ellsworth just before mine in the history. equazcion→19:44, 23 Oct 2013 (UTC)
I am working on a faster subtemplate, to be Template:JctI, which can show road junctions in the format of {Jct|state=xx|I|nnn}, but 4x-5x times faster. For example, it can reduce the reformat time, by 14-24 seconds faster, in page section:
The related templates in that section noted by Paine, including Template:Roadlink and Template:Convert, use 35 seconds and 3 seconds, towards the timeout limit of 60 seconds, which left only 22 seconds to format the {Jct} entries in the road table. In rare cases, all templates could reformat in 57 seconds, but a slight delay in the servers would exceed the 60-second timeout and hit the "504 Gateway" error during edit-preview. The draft version of {JctI} is in page User:Wikid77/Template:JctI, and seems to properly format all the {Jct} entries in that page, allowing the whole section to reformat within 40 seconds. More later. -Wikid77 (talk) 23:21, 26 October 2013 (UTC)[reply]
To editor Wikid77: That sounds much better! I also wonder if the entire section, Auxiliary Interstates could be reorganized into two or three subsections, so that subsection PREVIEW would reformat and render even faster? – Paine EllsworthCLIMAX!02:10, 27 October 2013 (UTC)[reply]
Going the wrong way
Recently, I've come across an unusually large number of internal links that lead to the wrong articles....most commonly, a name that leads to the wrong person. (and for some strange reason, the wrong person often turns out to be an athlete or ballplayer!) We have a bot that tells you when your link leads to a disambiguation page. Would it be possible to design a bot that searches for wrong links...maybe by comparing key words in the two articles? Obviously, you couldn't make the results definitive...just provide a suggestion to double-check...but it would be helpful (and prevent a lot of ballplayers from getting undeserved credit). WQUlrich (talk) 19:28, 23 October 2013 (UTC)[reply]
@WQUlrich: I'm guessing that the consensus here is that this is "not a good idea". And, to speculate, I'd guess one reason is a concern over false positives - if the bot decides the link is wrong, and it's not, then the bot is going to have unnecessarily bothered an editor (posting on his/her talk page about a possible error).
Moreover, this is actually a rather hard problem for a bot to understand. First, since articles don't have "key words", the bot will need to be able to create them, for each article. Then, if a link were (say) something like "Smith's father was Joe Whomever Smith", it's not at all clear that one article would have much in common with the other, and thus "key words" wouldn't really help. So you might end up with a bot that was only reasonably confident in its predictions, say 25% of the time, and still had a 10 or 20 percent error rate. And a bot that takes a LOT of time doing its processing.
@John Broughton: That's really what I expected, but I don't know very much about programming so I thought there might be some technique I was unaware of. I guess, practically speaking, if an article gets a lot of use it shouldn't take long for someone to discover the error. If it doesn't get much use (or any) then it's hardly a burning issue. WQUlrich (talk) 19:02, 25 October 2013 (UTC)[reply]
Byte size limit of transclusions
I've learnt recently that when templates are too big, pages that transclude it display a link instead of the template. What is the byte size "border"? (If that makes sense.) -- tnumbermaniacc20:59, 23 October 2013 (UTC)[reply]
Refer to article Saengerfest. A couple of days ago, there were a small list of backlinks under "What Links Here". Right now, there are thousands and thousands of backlinks there, the majority of which seem to be article talk pages. Does anybody know what causes this phenomenon? It's listed at GA since last week, but that listing had not caused this proliferation of backlinks as of a couple of days ago.— Maile (talk) 23:37, 23 October 2013 (UTC)[reply]
(edit conflict) Picking a few random talk pages I expanded all the templates and did a ctrl-f search for "Saengerfest", and WikiProject Texas' GAN list is the only place it showed up (I would've posted that but WOSlinker beat me to it, and had a diff to boot -- how he did that so quickly is beyond me ). Pretty sure that's all that would be needed to produce all those backlinks, no error required. equazcion→23:52, 23 Oct 2013 (UTC)
I sometimes edit Wikipedia on my phone, using the desktop version, with no trouble. However sometimes it redirects me to the mobile site and I don't like it, I prefer the capabilities of the full version. Is there a way to turn redirecting off or a button to "switch to desktop version" within the mobile site?KonveyorBelt04:00, 24 October 2013 (UTC)[reply]
If you dig through the archives you'll find this question has come up about every three or four months and the TL;DR answer is, unfortunately, no. Perhaps someone could add a note about this on the help page for the mobile site (or create such a help page if one does not already exist)? Technical 13 (talk) 04:08, 24 October 2013 (UTC)[reply]
Bullet point behavior on Wikipedia is unintuitive, and can become an intricate procedure when used during discussions. Responding well to a bullet point comment means placing your comment on the very next line with no break in-between, and with, at most, one more asterisk than the line above.
One must take care not to stray from these rules, lest one look like an idiot.
Hello, how are you?
I appear to be a moron, thank you. left a blank line!
Wikipedia can help you pick out brain medicines, as it has for me. too many asterisks!
I have literally never come across an instance where placement of multiple bullet points on a line before a statement was intended, nor could I fathom when that might be beneficial. Can we fix this behavior and make bullet points easier to use, by doing away with these silly requirements? Let's just have multiple asterisks produce a single bullet with the requisite number of indents before it, in all cases. PS. I realize some might say bullets should just not be used in discussions, but they will be regardless. PPS. I also realize this may be some unintended technical issue, but if so I'm hoping we can agree it might be worth fixing? equazcion→04:43, 24 Oct 2013 (UTC)
I believe this has come up before, and I know there is a workaround to avoid this behavior.
Hello, how are you?
I appear to be a moron, thank you. left a blank line!
Wikipedia can help you pick out brain medicines, as it has for me. too many asterisks!
Replacing all of the leading asterisks with colons avoids the issue. Whether or not this should be fixed in the core, is up to the developers to decide (and I will almost bet there have been tickets submitted on Bugzilla about this behavior), and you may want to ping them or post a ticket on Bugzilla... Technical 13 (talk) 04:54, 24 October 2013 (UTC)[reply]
Line breaks between list items (especially those involving bullets, like the last two examples) cause problems for screen reader users like myself. Indenting using a mixture of colons and asterisks (especially when the colons come first) also creates some hideous HTML, which is presented literally by screen readers. Graham8708:38, 24 October 2013 (UTC)[reply]
Were it up to me, I'd have a character that served the same function as : or * that would be active only in talk pages but that didn't create html lists to implement a discussion. Very often, lengthy discussions become wall-of-text-like when one writer follows immediately after another without a line break. This I think is an encumbrance on understanding.
I don't know how to implement such a mechanism because I haven't given it any more thought than what I've described. If such a mechanism is possible, it would be a solution to both Editor Graham87's and Editor Equazcion's issues. And, it would keep other editors from harping on me for adding line breaks to my contributions to talk pages.
Quick guide to the "right" way to reply to talk posts: 1. Never put a gap between posts. 2. Copy the asterisks (*), colons (:) and number marks (#) from the post to which you are replying, then add one symbol of your own.
<p style="margin-left: 1.5em">
Indented without list markup
</p>
I agree with Trappist regarding the need for better markup; the way to indent text without abusing list markup (shown right) is too painful for use in discussions. Another wiki I edit uses tildes for this purpose, but MediaWiki already uses tildes for signing posts. The developers are unlikely to add such markup because the upcoming Flow will automatically indent posts.
Regarding the wall-of-text problem, I have the following in my common.css to resolve this:
This also adds extra space to definition lists in articles, but I like the extra spacing there too.
Part of the problem of multiple bullets is down to MediaWiki not requiring editors to mark the start and end of a list, to simplify markup. MediaWiki tries to figure out where lists start and end by seeing if an asterisk is matched at the same position on the next and previous lines. If there are multiple asterisks not matched on the previous line, MediaWiki takes it to be the start of multiple bulleted lists nested inside each other, hence multiple bullets appear. If the previous line is blank, all of the asterisks start a new list, so all cause a bullet to show. – PartTimeGnome(talk | contribs)20:28, 24 October 2013 (UTC)[reply]
Multi-indents and need for relative ":+" indentation: We should have relative-indentation (of the form ":+" as one more tab stop than the above msg), but there is a common case where a message is followed by a double-indented response, when interjecting a 2nd response above a prior (often lengthy) response:
: This is the original message.
::: This is the 2nd, but interjected, response.
:: This is the first response.
Such cases, of double-indented lines (2 extra colons "::"), are common when someone thinks they have a more-direct response than the immediate reply had stated. In a relative style, it could be ":++" to indent by 2 levels more than the prior msg. For relative indentation as ":+" then the parser would need to be smart about remembering the indentation level of the prior text, and when too complex, just show as relative to the margin (so then ":+++" would be same as ":::"). Of course, if templates had the parser extension for global variables, then we could write a smart template which kept a counter and indented by "{{in+}}" as a template which increments the global variable for indentation in the talk-page. Unfortunately, {:+} will transclude page "+" into a discussion, so that would be even more confusing. At least we have {od}, which pretends to be smart enough to know wherever the last word was displayed, to handle the reverse outdent cases:
{{od}} is not that smart, nor does it pretend to be (what you describe cannot be done in a template). Using it without any parameters produces an outdent marker with a length equivalent to 10 colons of indent. If you want a different length, specify it as a parameter. E.g. {{od|::::}} will make an outdent marker equivalent to 4 colons of indent. – PartTimeGnome(talk | contribs)20:36, 24 October 2013 (UTC)[reply]
Drop-down option for saved edit summaries/subject headers
The drop-down option for saved edit summaries/subject headers etc. has vanished for the second time (previously fixed by unclicking 'Always use a secure connection when logged in' on 'Preferences'), and I can't seem to fix it. Any help welcome. GiantSnowman11:44, 24 October 2013 (UTC)[reply]
How to remove "Add category" link from Main page?
Hello, this is Surya - admin in Tamil Wikipedia. In Tamil Wikipedia's main page the "Categories" link & category addition option is always shown. How to remove that from main page. I searched for it in en.wiki's main page code, but, can't find though. Need help, kindly respond with any markup or CSS changes. Thanks a lot. --SuryaPrakashTalk...13:15, 24 October 2013 (UTC)[reply]
I don't see any category stuff at the bottom of w:ta:Main Page. I was going to suggest this was caused by the Hotcat gadget, but it looks like you don't have that at ta.wikipedia. It looks like w:ta:User:Surya_Prakash.S.A./vector.js (your vector.js) does have w:ta:User:Jayarathina/iwt.js, which seems like it might have something to do with categories (not really sure). I would try removing that and seeing if it makes that stuff disappear from the main page, but this is just a guess. equazcion→13:49, 24 Oct 2013 (UTC)
That's odd. When I checked their gadgets just prior to my reply above, hotcat wasn't there... or at least, I could've sworn it wasn't, and I did the ctrl-f thing and everything. Wondering if I missed it somehow or they added it because of my reply referencing it here. equazcion→17:14, 24 Oct 2013 (UTC)
Nevermind. I'm using Chrome to auto-translate, and it apparently removes those .js filenames from the gadget list. So all I saw was "Viraivuppakuppi", which apparently means "Hotcat". equazcion→17:17, 24 Oct 2013 (UTC)
I'm trying to delete File:Minigun System.gif but I'm not able to. I'm getting an error when I try to do so: "Error deleting file: The file "mwstore://local-multiwrite/local-deleted/o/l/c/olcm49xeacsh8myzkuz294lv1d0n3xv.gif" is in an inconsistent state within the internal storage backends". Anyone know what the problem is? Mark Arsten (talk) 15:30, 24 October 2013 (UTC)[reply]
For the second time today, I attempted to undelete a file and received the following message (or one similar):
Error undeleting file: The file "mwstore://local-multiwrite/local-public/1/16/Emeritus_Professor_Norman_Maclean.jpg" is in an inconsistent state within the internal storage backends
In the other instance, I had access to the file, so ignored it and re-uploaded it. In this case I do not.
Did you try to see it? I tried again and it failed. I also ran into a similar problem with:
Error undeleting file: The file "mwstore://local-multiwrite/local-public/3/33/Jackson_Laboratory_President_and_CEO_Edison_T._Liu,_M.D.jpg" is in an inconsistent state within the internal storage backends
Has the font color of text in the body of Wikipedia pages changed? I think it used to be black, but now all pages appear to have grey text. Is this the case for anyone else? -- ToshioYamaguchi19:53, 24 October 2013 (UTC)[reply]
To expand a bit, I know that the font color used to be the same as the color of my signature (which is black). Now there is a noticeable contrast (at least for me). -- ToshioYamaguchi20:06, 24 October 2013 (UTC)[reply]
In case anyone was wondering, we recently updated Special:PasswordReset's basic design to match login and account creation on all wikis. This was noted above in Tech News, but I thought I'd give it a bit more visibility with a separate thread. Password reset is linked directly from login, and it's one of the most common ways to get there. Having login and password reset look a bit more alike will make it easier on everyone, and especially new people. What we've settled on so far isn't perfect, and ideas for further improvements are being discussed on MediaWiki.org, since this is a part of MediaWiki core. Steven Walling (WMF) • talk19:57, 24 October 2013 (UTC)[reply]
This code comes up in Guided Tours as a way to send the user back to the editing page if they save without making an edit (which fails to advance the tour)
var postEditButtons = [];
if ( mw.config.get( 'wgAction' ) === 'view' && !gt.isPostEdit() ) {
postEditButtons.push( {
name: 'Go back and make an edit',
onclick: function() {
window.location.href = window.location.href +
"?action=edit";
}
} );
}
https://en.wikipedia.org/wiki/MediaWiki:Guidedtour-tour-twa1.js
As you can see, I can either append ?action=edit or &action=edit, either works for only some of the urls I'm dealing with. Any ideas for a fix? Cheers, Ocaasit | c20:28, 24 October 2013 (UTC)[reply]
A URL may be thought of as several portions. The ? separates the main part from the "query string"; and an ampersand & is used to split the query string into two or more parameters. It follows that there can only be one question mark; that an ampersand cannot be used unless a question mark is also present; and that the first ampersand (if present) must be after the question mark. So, at this exact moment, my browser has in the address bar http://en.wikipedia.org/w/index.php?title=Wikipedia:Village_pump_%28technical%29&action=edit§ion=39 - here, http://en.wikipedia.org/w/index.php is the base URL, ? separates that from the query string, which is title=Wikipedia:Village_pump_%28technical%29&action=edit§ion=39 - that in turn has two ampersands, therefore three parameters: title=Wikipedia:Village_pump_%28technical%29action=editsection=39 --Redrose64 (talk) 20:36, 24 October 2013 (UTC)[reply]
(edit conflict) URL format is generally that there's one (and only one) question mark for the first URL parameter, and if there are multiple parameters, the others get ampersands. Post-EC -- Basically what RedRose64 said, only simpler :) equazcion→20:37, 24 Oct 2013 (UTC)
You can do it with mediawiki.Uri (this is already loaded by several modules, so you probably will not be causing anything extra to be downloaded). You should use mw.loader.using to depend on it like this:
mw.loader.using('mediawiki.Uri',function(){// Put the whole tour in here. ...onclick:function(){window.location.href=newmw.Uri().extend({action:'edit'}).toString();}// ...});
The nice part is that this works for any number of URL parameters (you just comma separate them), and it handles all the ? and & business for you. It will also override existing parameters with the same name . Superm401 - Talk20:52, 24 October 2013 (UTC)[reply]
Need invisible space in Wikipedia mixed numbers
In July 2013, there was a discussion to put an   in all mixed numbers, to separate the integer portion from the fraction during a copy/paste, {{frac|21|3|4}} gives 21+3⁄4 and paste-copies as "21 3/4". The archived discussion:
Unfortunately, the extra space in the fraction is excessive (appearing as if "21 .75"), and we need a hidden space. Recall that in other cases, we put hidden text in a span-tag "position absolute" off-screen, but the copy/paste included the off-screen text when copied in any browser. So, I think:
:Can you use  ? It looks like you already have the whole thing wrapped in a "nowrap" class, so you shouldn't necessarily need the "non-breaking" feature of . It looks like you're using   to create a space, but I don't see that on any HTML entities lists (my Google-fu is failing me). What does it do? – Jonesey95 (talk) 22:39, 24 October 2013 (UTC)[reply]
Even a thin-space &thinsp is likely too wide, and might fail on some older browsers. In a decimal numeral, &thinsp does: "21 .75" whereas there should be no spaces inside numerals. Instead, the hidden space would be: 21<span style="position:absolute; top: -9999px"> </span><sup>3</sup>⁄<sub>4</sub> and appear as "213⁄4" (←try copy/paste with that fraction). -Wikid77 (talk) 22:51, 24 October 2013 (UTC)[reply]
(e/c) Now that I've gone and read the thread linked above, which anyone should do before responding here, it looks like people have tried thinsp and other tricks, only to run into copy-paste problems when trying to copy the thinsp version of 21+3⁄4 into a text editor called "Notebook++" on Windows. I tried copying and pasting the output of the thinsp version of the fraction on my Mac, from Firefox 24 into TextWrangler 2.2.1, Outlook 2011, Notes (Mac OS 10.8), Excel 2011, and Word 2011, and it pasted just fine in all of them. With respect, I think the person with Notebook++ is basing the "it's working" test of copy and paste on an less-than-adequate text editor.
Responding to myself again to point out that the existence of the space itself is a WP:MOSNUM issue that appears to be under discussion, both at the thread above and at Template_talk:Frac#Spacing.
I don't know how relevant this is but regarding Notepad++, its default encoding is ANSI, and a LOT of characters common in Wikipedia articles don't render that way. You can switch the encoding easily to UTF-8 (what Wikipedia uses, I think, so anything that shows up here should come out correctly there), and even change the default to that, but some people might not realize. Someone should post a thinsp example fraction here for testing. equazcion→01:37, 25 Oct 2013 (UTC)
Copying and pasting is never going to work across the board. Personally, I see no benefit in trying to make it work. The templates are there for formatting only. Consider 1+2/3 and ; you wil notice when copying that <math>...</math> rendered output cannot be copied at all. So why are we trying to make it possible with templates? — Edokter (talk) — 10:01, 26 October 2013 (UTC)[reply]
"User contributions" has disappeared from my Toolbox on my user page and all other user pages
Well, I still have "User contributions", also "Logs" which should be just below it but is absent from your screen shot.
"Message names" was introduced some months ago, and was announced on this page. It's useful for technically-minded people, since it reveals which of the hundreds of MediaWiki: files contains text like "From Wikipedia, the free encyclopedia". It's not really an everyday thing. You can hide it with some CSS:
How can I get "User contributions" back? (I don't care one way or another about "Message names" but I resent it was just put there with no explanation.) I don't follow Village Pump (technical) and I expect most users don't. It's bad that I have to do something to "hide" "Message names".
Please don't fragment the discussion. It makes the discussion harder to follow (and you'll probably get a reply along the lines of "see previous discussion"). People can see your comments here and would tell you if they knew what was wrong. You can help by trying a few things to narrow the problem down:
It looks probable that you have chosen the preference under gadgets/appearance which says "Add page and user options to drop-down menus on the toolbar." If so, you can turn that option off which will return the links to the toolbox drop-down menu or get used to seeking those options under the new user drop-down menu.—John Cline (talk) 21:02, 24 October 2013 (UTC)[reply]
Thank you! Thank you, John Cline! That was the problem. All fixed now! Sorry I got frustrated trying all that other stuff, but thank you! Soranoch (talk) 21:41, 24 October 2013 (UTC)[reply]
Yes, it looks like you have been changing preferences at Special:Preferences#mw-prefsection-gadgets. "User contributions" changes location when you enable "Add page and user options to drop-down menus on the toolbar", and "Message names" appears when you enable "Add a toolbox link to display the current page with MediaWiki message names replacing their text". PrimeHunter (talk) 21:23, 24 October 2013 (UTC)[reply]
Can't think of anything I changed that would have affected this. Just today, the little triangle widgets in the navigation sidebar that allowed me to expand or collapse the groups have disappeared. Curiously, this only affects Chrome while I'm logged in. When I log out in Chrome, the widgets are there and they are there in IE whether logged in or not. Any ideas on what might be causing this? older ≠ wiser20:48, 24 October 2013 (UTC)[reply]
I'm getting this in Firefox on and off. I can generally hard-refresh to get them back, but it still recurs. equazcion→20:51, 24 Oct 2013 (UTC)
Looks like I only have to hard-refresh once per unique page. I guess the styles server is/was having trouble today, again. equazcion→20:53, 24 Oct 2013 (UTC)
Heh, it had been bugging me most of the day, but I figured it might be one of those things that just comes and goes of its own accord, and so I didn't bother asking. Of course, almost immediately after posting here, the widgets are back. Go figure. older ≠ wiser21:01, 24 October 2013 (UTC)[reply]
Looks like the two users above (User:Bkonrad, and User:Soranoch in the section above) are/were experiencing the same issue. The server that provides certain javascript and css is likely having intermittent trouble. bypassing your cache will fix this some of the time (during times when the server is back up but your browser is still showing its cached page version). I'm experiencing the same issues, FYI. equazcion→21:12, 24 Oct 2013 (UTC)
Mobile Wikipedia overwriting Firefox's desktop request
When using Firefox for Android on my phone, I've sometimes wanted the desktop edition of a page instead of the mobile edition. Tapping "Request desktop site" isn't working. It's being overriden. I know you can click desktop at the bottom of the page, but it takes forever to scroll on pages like Template:AFC statistics. :( -- tnumbermaniacc04:42, 25 October 2013 (UTC)[reply]
The problem you are reporting sounds like a potential issue in the code of the MediaWiki software or the server configuration (and bugzilla:56139 is maybe related, maybe not). If the problem is reproducible, it would be nice if somebody who has this issue could send the software bug to the 'Bugzilla' bug tracker by following the instructions How to report a bug. This is to make developers of the software aware of the issue. If you have done so, please paste the number of the bug report (or the link) here, so others can also inform themselves about the bug's status. Thanks in advance! --AKlapper (WMF) (talk) 08:45, 25 October 2013 (UTC)[reply]
I have the same problem in others websites. The server detect that your are using mobile so it redirect you automatically to the mobile version of the site. Rabah201130 (talk) 09:02, 25 October 2013 (UTC)[reply]
Not knowing exactly what the "Request desktop site" in Firefox does, my guess is it only modifies the URL to try and load the desktop version. MediaWiki will then send you back to the mobile version because it does not detect a cookie indicating you prefer the desktop version. That cookie is only set when you click the Desktop link. — Edokter (talk) — 09:49, 26 October 2013 (UTC)[reply]
How to get interwiki links using a bot
Hi,
I want to know if there is a solution to get the interwiki links using a bot. Before Wikidata, the interwiki links were included in the text of the article. I'm using DotNetWikiBot API. Thanks. Rabah201130 (talk) 09:02, 25 October 2013 (UTC)[reply]
The special pages that list the uncategorized categories, templates and pages have stopped updating across wikis on September 10. Has this been discussed previously somewhere or is there a bug report filed? --Pxos (talk) 09:27, 25 October 2013 (UTC)[reply]
my output shows It has to work when you set the transliteration as ascii, It's not related to the output device because when you set it as ascii, anything will be translitered to latin (e.g. س-->s) :)Ladsgroupبحث16:17, 25 October 2013 (UTC)[reply]
User-config.py Syntax
Good Evening Everyone,
Here is the photo that show the «syntax» I put in «User-config.py» file of my BOT to work on arabic wikipedia.
Can you please tell me about mistakes in it ?
I've copied the following from Village Pump (Miscellaneous) on the advice of Rivertorch:
Even for a seasoned editor like myself, WP's help pages can seem labyrinthine - I hope this is a reasonable place to post this . . . the page at River Usk has become unavailable though the talk page still functions. Last edit was by Verbcatcher a few hours ago. Can it be fixed please. Geopersona (talk) 07:20, 24 October 2013 (UTC)[reply]
When using Internet Explorer the title of the page appears followed by nothing more then I get locked out of Wikipedia and replaced by this error message: res://ieframe.dll/acr_error.htm#wikipedia.org,https://en.wikipedia.org/wiki/River_Usk (with Internet Explorer has stopped trying to restore this website. It appears that the website continues to have a problem. on the screen). Another user has reported the same phenomeneon on the WikiProject Wales talkpage where I'd also made mention of it. A third regular editor reported no problem when using Google Chrome. I've just tried Chrome and had no problem but IE still won't open it. I've never had this issue on WP before in several years of editing and browsing the site. thanks Geopersona (talk) 17:53, 24 October 2013 (UTC)[reply]
And the answer is 'yes' in reply to Rivertorch's recommended actions - it causes the same problems on other computers I've tried when using IE as a browser (it's IE8 on this machine). Hope someone can figure out what's going on and fix it! cheers Geopersona (talk) 16:05, 25 October 2013 (UTC)[reply]
Everytime i come to wikimedia, i run a background sniffer, it detected that this site connects to DigiCert on ip (198.35.26.106) which isn't actually a wikimediaIP range (208.80.154.xxx), is this necessary?. It started recently, about a week or 2 back. I'm asking cause it uses around 252kb of data everytime I refresh the Watchlist or go to any page. I don't care about any of my pages being "safe" but I don't like unnecessary usage of Data as i'm on a Limited data plan and on wikipedia, you have to refresh your watchlist a LOT...Is there a way to disable this? Was this set in place after the recent "passwords compromised" debacle a few weeks back?--Stemoc (talk) 00:12, 26 October 2013 (UTC)[reply]
I'm on Firefox 24. If true, is there a way to disable it? cause when i do a search for the IP, it tells me its owned by Wikimedia Foundation Inc. it does eat a lot of unnecessary data and ignores the cache which means if i go to a page, it will get cached right, so when i close that tab and open it again, it should not eat the same amount of data it used in the first time since its cached but it does, which probably means that now wikipedia is running on https instead of http...--Stemoc (talk) 00:46, 26 October 2013 (UTC)[reply]
Open Tools -> Options or Edit -> Preferences (depending on your build), then go to Advanced -> Certificates -> Validation. If the first option about OCSP is on, then it's connecting to DigiCert to verify the certificate hasn't been revoked since it was issued. You can disable this if you want, but it will probably reduce your security and barely save any data at all. Jackmcbarn (talk) 03:42, 26 October 2013 (UTC)[reply]
Why do you say it is connecting to "DigiCert"? When I check whois it tells me that the IP range 198.35.26.0/23 is registered to Wikimedia, and doing a PTR lookup for the IP tells me the host name is bits-lb.ulsfo.wikimedia.org; in other words, it's the load balancer for bits.wikimedia.org at the new ulsfo caching datacenter. The 252kb is going to be the CSS and JavaScript, although your browser should be caching it so it isn't re-fetched on every page view. Anomie⚔01:38, 26 October 2013 (UTC)[reply]
Not necessarily connecting, but that IP is new to me and yes its registered to wikimedia, i'm only familiar with the (208.80.154.xxx) and the toolserver Ip ranges. Yes my sniffer tells me its for the 'bits-lb.ulsfo.wikimedia.org' caching datacenter but what i haven't noticed until today is that my url now has a https instead of http, It wasn't https a few days ago..the new server not only makes wikipedia run through a secured server but also prevents the browser from caching it....Is that really necessary? How do i opt-put of the "secured" option?...it is set to cache in my browser so that isn't the problem..--Stemoc (talk) 03:12, 26 October 2013 (UTC)[reply]
Hi (and welcome back I was under the impression that you left. Sorry.) Cometstyles. I think you can use http by going to Special:Preferences and deselecting "Always use a secure connection while logged in", but I would generally recommend that you use https. If caching is the issue, you should file a report on bugzilla: (I can do so if you don't want to). πr2 (t • c) 04:12, 26 October 2013 (UTC)[reply]
Why did i not see that option? a slight reduction (from 272 to 140kbs), thanks...I don't know who you are lol, hehe..yeah i didn't completely leave, i created a lot of articles and i had to come around from time to time to update and fix them so i decided to create an account to look after them..thanks for the option..I'm not worried about "security" I only use wikipedia from home from my PC which is hard to hack..no need to file a bug...Thanks though :) ...--Stemoc (talk) 04:53, 26 October 2013 (UTC)[reply]
Your computer being hard to hack isn't relevant. HTTPS prevents people on other computers intercepting the login token your browser sends to the Wikimedia servers, by encrypting everything sent on the connection. The login token can be used to impersonate you on Wikimedia wikis until you log out, so it's a bad thing if someone else can discover it.
The risk is particularly high if you use Wikipedia over Wi-Fi. Even without Wi-Fi, you don't know how secure the other systems used to connect to Wikipedia are. (Using traceroute, I see my requests to Wikipedia pass through four computers before reaching Wikimedia. This will probably be different for you, depending on your ISP.) – PartTimeGnome(talk | contribs)20:39, 26 October 2013 (UTC)[reply]
I ran a trace route on mine, went through 21 computers and yet i still feel secured. We were all on an unsecured method till last week or so when they changed the url to a secured server. WiFi is not common in my country and those that are available are protected. My IP hopped between 2 other countries before reaching USA...I think others need to be notified of the new changes, though most in developed countries wouldn't care but those like me in a very much 3rd world country with limited data would..--Stemoc (talk) 01:38, 27 October 2013 (UTC)[reply]
When I download Main Page over HTTPS, Firefox 24 makes two POST requests to http://ocsp.digicert.com/ and receives two 200 OK answers. Both the requests and the answers have Content-Lenght < 500 bytes. When I tested this, I first cleared all browser data (cookies, history, cache et cetera) and then typed in the full address manually. Are you doing something differently, and if so, what are you doing differently? --Stefan2 (talk) 16:37, 26 October 2013 (UTC)[reply]
WMF notification email marked as spam by Yahoo
An email to me with the subject "You have a new notification at Wikipedia" (without quotation marks) sent October 23 or 24, 2013 (seen today), from wikipedia.org at IP 208.80.152.133 (involving terbium.eqiad.wmnet and mchenry.wikimedia.org) was treated by Yahoo as spam. (I have omitted sender and intermediate email addresses in here per a WP policy or guideline but can supply them if desired.) That may mean that the sender's mailing list administrator is not deleting bounced addresses after a second bounce, if that's still the current standard, or it may mean something else. It's not a problem for me, but it would be for many other recipients, because Yahoo sends such email into a spam folder instead of the regular inbox, and many users may not check the spam folder. Nick Levinson (talk) 20:03, 26 October 2013 (UTC)[reply]
It doesn't on my main monitor. However, if I move it to a smaller monitor maybe 15" I do see the overlap. The reader reports the problem existing on a monitor with dimensions of 17 x 11.
I do not get overlap in Chrome, nor IE even on a laptop. In those cases, when there isn't enough room, it first squeezes the TOC, and eventually slides it down.
However, in Mozilla, the image will overlap the TOC if I view it on a 15" screen or reduce the window size to approximately that size.
I trying throwing in a {{Clear}} but that didn't help.
I was able to force the overlap with some zooming in FireFox and fixed it by re-arranging the elements on the page (moved the TOC up and that long image down a little... I also threw in a {{-}} just above the next section header which I find is a more efficient method than {{Clear}}> Technical 13 (talk) 23:45, 26 October 2013 (UTC)[reply]
Dear tech experts: I would like to find a way to search articles inside a category for specific text. For example, right now there are about 1400 pages that are the category "Pending AfC submisssions". If I want to find out which of them have the word "football" in the text, is there a search tool that will do this? CatScan is good at searching categories for many things, but it doesn't appear to have a field for text in the body of the article. The standard search engine finds text in an article, but doesn't seems to do categories. I would like to create a link to a preprepared search (maybe in a user box) which could be placed on a WikiProject page or on a user page which would allow users to quickly check from time to time to see if there were any pending Afc submissions in their area of interest. I left a message about this at the help desk, HERE but didn't find an answer there so far so I thought that I would try here. —Anne Delong (talk) 00:26, 27 October 2013 (UTC)[reply]
Followup: in the Help:Searching article, there's a function called "incategory", and after reading the instructions I tried this search: football incategory: Pending_AfC_submissions, but it didn't find any results. Can it be because this is a "hidden" category? —Anne Delong (talk) 13:58, 27 October 2013 (UTC)[reply]
I fear you may be right. For example, doing this search:
This is because the search isn't, strictly speaking, for articles in that category - rather, it's for articles with that category link in the wikitext. As a result, "German chemists" will get a hit, but "Wikipedia articles with VIAF identifers" won't, because this is generated from within a template. The same problem applies to AFC categories - it's not that they're hidden, it's that they're template-generated rather than in the article code. Andrew Gray (talk) 14:33, 27 October 2013 (UTC)[reply]
Well, I tried something different, "football -missing -draftnew prefix:Wikipedia talk:Articles for creation", which at least limits mention of the articles in other spaces and eliminates ones with missing templates and unsubmitted drafts, but I haven't found a way to distinguish between current submissions and already declined ones, because they both use the same basic template. If there was some common item that was only on pending submissions... —Anne Delong (talk) 18:16, 27 October 2013 (UTC)[reply]
In the last couple of days I've noticed that I'm no longer getting the orange notification when somebody edits my talkpage. Is this a known bug; or a feature; or am I seeing things? bobrayner (talk) 11:53, 27 October 2013 (UTC)[reply]
The yellow bar has been appearing fine for me today. The usual behaviour, as far as I'm aware, is to get the yellow bar for all posts made to your talk page apart from minor edits made by bots. So the edits on your talk page by SineBot shouldn't have triggered the orange bar, but all the other edits should have. If the yellow bar hasn't been showing up for you at all then it sounds like a bug. I'll test it by making a post on your talk page now - let me know if you see the yellow bar this time. — Mr. Stradivarius on tour♪ talk ♪12:09, 27 October 2013 (UTC)[reply]
You can see the text of the "talk" link is "talk: you have new messages" instead. It should be in a yellowish orange – if it's not, then it muse be overridden by some of your custom styles. Matma Rextalk13:20, 27 October 2013 (UTC)[reply]
Already done, although my understanding was that it was redundant, as it only controlled the behaviour of {{Usermessage}}; and having been like that since January, it's unlikely to have triggered the recent change in notification behaviour.
Just as a followup, the problem persists after removing that line from my monobook.css. Can recreate in IE10. Cannot recreate in the Modern, Vector, and Cologne Blue skins. Good grief, Cologne Blue looks horrendous in IE10. My eyes are burning. bobrayner (talk) 16:27, 27 October 2013 (UTC)[reply]
Yes, yes, it's broken, thanks guys :) I submitted a patch to the bug: https://gerrit.wikimedia.org/r/92181 – it'll hopefully be merged after the weekend. The problem isn't very critical, so deployment to Wikipedia and friends will probably go per the standard timetable (within two weeks of the patch being accepted). Matma Rextalk17:19, 27 October 2013 (UTC)[reply]
What's happened to the little yellow bar that lit up the 'talk' button? It used to come up for talk page messages while thanks and such just had the red square. Peridon (talk) 18:12, 27 October 2013 (UTC)[reply]
Usually do say, but got boggled by a dispute. That could also be why I didn't see that thread - I did really look at the Contents list. Thanks anyway. Peridon (talk) 18:36, 27 October 2013 (UTC)[reply]
wikimedia.org or filed under the "Security" product in Bugzilla.
.png)
Thanks for the reminder. — dainomite 03:07, 26 October 2013 (UTC).svg){kind=small}
). WQUlrich (talk) 19:28, 23 October 2013 (UTC)
). Pretty sure that's all that would be needed to produce all those backlinks, no error required. equazcion → 23:52, 23 Oct 2013 (UTC)
Moved to #Unable to delete file – merged with an existing discussion about the same problem – PartTimeGnome (talk | contribs) 22:21, 25 October 2013 (UTC)
Resolved
![[9]](https://commons.wikimedia.org/wiki/File:US_Open_2009_4th_round_122.jpg){kind=link}
