[[:Category_talk:Cities_and_towns_in_Kolkata_district]] is not accessible. Please fix. --[[User:VasuVR|<font color = "#F60">VasuVR</font>]] ([[User talk:VasuVR|<font color="blue">talk</font>]], [[Special:Contributions/VasuVR|<font color="green">contribs</font>]]) 05:15, 28 March 2012 (UTC)
[[:Category_talk:Cities_and_towns_in_Kolkata_district]] is not accessible. Please fix. --[[User:VasuVR|<font color = "#F60">VasuVR</font>]] ([[User talk:VasuVR|<font color="blue">talk</font>]], [[Special:Contributions/VasuVR|<font color="green">contribs</font>]]) 05:15, 28 March 2012 (UTC)
:Yes, causing problems :( And when I view the history it shows nothing, strange or maybe there's a sysop involved. By the way, the category is enough to be filed for deletion. [[User:Extra999|<span style="color: green">'''e'''<sup>x</sup>tra<small>999</small></span>]] ([[User talk:Extra999|<span style="color: black">talk</span>]]) 08:58, 28 March 2012 (UTC)
:Yes, causing problems :( And when I view the history it shows nothing, strange or maybe there's a sysop involved. By the way, the category is enough to be filed for deletion. [[User:Extra999|<span style="color: green">'''e'''<sup>x</sup>tra<small>999</small></span>]] ([[User talk:Extra999|<span style="color: black">talk</span>]]) 08:58, 28 March 2012 (UTC)
== Does everyone have the span/dir/ auto message now on top of almost every Wikipedia page they consult? ==
Does everyone now have the span/dir/auto message on top of almost every Wikipedia article they consult?
The technical section of the village pump is used to discuss technical issues aboutWikipedia. Bugs and feature requests should be made at Bugzilla (How to report a bug). Bugs with security implications should be reported to securitywikimedia.org.
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.
Problem displaying wikipedia content in iPad application
I'm seeing Wikipedia mobile content not sized correctly when displayed within iPad apps using UIWebView. It is as though the mobile version has the page width fixed to the size of the iPad device, not the actual view it is displayed in.
This is what I'm seeing: [1]
This is what I want to see: [2]
This used to work correctly. I suspect Wikipedia has become smarter in detecting it is displaying on an iPad and adjusting the frame size assuming it is being displayed with the iPad Safari browser.
Is there any way to specify the content width along the lines of http://en.m.wikipedia.org/w/index.php?title=Jupiter&device-width=320.0 (which doesn't work by the way)? — Preceding unsigned comment added by 24.87.55.100 (talk) 02:03, 22 February 2012
I'm begging you, please, please, please in the name of the sun and the moon and the stars, stop using javascript to inject the banner into the DOM after the page renders. I'm so fed up with the page layout changing just as I'm about to click something because jQuery fired a callback and a banner appeared. Thanks. For now, I solved the problem myself with the below adblock plus filter
*.wikipedia.org/*BannerListLoader*
PS It looks like text after a pre close tag gets the pre treatment. Put this here to save my sig.
I have asked this before at least once, however nobody really seems that interested in the problem. As intelligent as a lot of the technical people here are, I am very surprised that such a poor design choice would be allowed to stay around for so long. —danhash (talk) 15:07, 13 March 2012 (UTC)[reply]
Thanks for the info on the pre tag. What about the banner? Couldn't the server side rendering include a blank div of fixed height during banner season, then let the JS inject whatever content at wants? At least then the page layout wouldn't change after the useragent renders. --76.18.43.253 (talk) 20:27, 13 March 2012 (UTC)[reply]
Why is any banner being loaded with JavaScript anyway? It would be better from a browsing standpoint to have any banner loaded statically as part of the page (and better to have no banners at all). An additional reason not to use JavaScript is that if users update their personal css files, Adblock Plus settings, etc. to block banners, the JavaScript will still load even though it is not shown on the page. —danhash (talk) 20:33, 13 March 2012 (UTC)[reply]
It was historically done that way when the classic sitenotice started being used (abused) full-time for various messages/requests/notices. The search engines would grab the text for a page's summary in the order it appeared in the page source, which would then give results full of the sitenotice. The other workarounds were insufficient (such as excluding sections of a page from indexing, or floating the sitenotice), so we ended up with the JS-loaded sitenotices we know and love today. --Splarka (rant) 07:25, 14 March 2012 (UTC)[reply]
Thanks for the explanation; it makes a bit more sense now. However, surely it wouldn't be that big of a deal to change how sitenotices are coded, and it is a long overdue and dreadfully needed change. Any devs or admins with the ability care to take a look? —danhash (talk) 13:48, 14 March 2012 (UTC)[reply]
There is an open bug related to this:
Bug 26234 - CentralNotice moves content down when a link to a heading is followed
Regarding the question of why CentralNotice loads the banner via JavaScript, the only reason is so that CentralNotice banners can be geotargeted, i.e. displayed only to certain users based on their physical location, while not destroying Wikipedia's caching. Kaldari (talk) 00:04, 16 March 2012 (UTC)[reply]
Thanks for the explanation and back story. FWIW an empty div would at least fix the geometry of the page, then let you inject geotargeted content and avoid the search engine issue. --76.18.43.253 (talk) 00:23, 19 March 2012 (UTC)[reply]
An empty <div>...</div> of nonzero height would mean that should the geotargeted message not be relevant, there would be a big blank space at the top of the page. This in turn would mean that people would complain either about the wasted space, or that the page was misbehaving in either of two ways: that the page content was being shown too far down, or that something that should have been shown wasn't being shown. Putting in a placeholder like "This space intentionally blank" would solve the apparent misbehaviour issue, but wouldn't stop complaints about wasted space. --Redrose64 (talk) 11:43, 19 March 2012 (UTC)[reply]
It can hardly be worse than it already is; this should be fixed as soon as possible. Users should have a site-wide, if not global way to opt-out of all banners and notices. No banner or notice is so important that we must mutilate usability and annoy the heck out of users to force them to see it. I have never seen a sitenotice as long as I have used Wikipedia that has been at all relevant to me. I have donated to WMF once, but am very hesitant to donate again until and unless I am given the option for all such annoyances to be removed once and for all. I would happily click on a relevant banner or donate notice on the main page (provided it was loaded statically), and I would happily watchlist a page specifically for such notices or support the addition of a link to such a page in a relevant part of the interface. —danhash (talk) 16:13, 19 March 2012 (UTC)[reply]
It applies even to other elements loaded by js. This is a known bug, the solution (do I dare say?) is to have the reserved space loaded as part of the page, and the content by JS if it's really necessary. RichFarmbrough, 20:22, 24 March 2012 (UTC).[reply]
Why can't the whole page be loaded statically? There is no good reason to have extra JavaScript executing on every page loading crap nobody wants to see in a way that breaks every major usability guideline and user expectation. It is almost as bad to have a section of the page load blank and then fill with content as it is to have the whole page render and then move down after the content loads. —danhash (talk) 15:17, 26 March 2012 (UTC)[reply]
User talk pages not NOINDEXed for mobile site
It seems that as of very recently (the last day or two) user talk pages on the http://en.m.wikipedia.org version of the site are not NOINDEXed (i.e. they show up in Google results), even when NOINDEX is explicitly set on the page. —danhash (talk) 14:18, 13 March 2012 (UTC)[reply]
This is still an issue, and it includes NOINDEXed user pages as well. Has there been a change to NOINDEX or the mobile site recently that could cause this? —danhash (talk) 16:04, 19 March 2012 (UTC)[reply]
See bugzilla:35233: the mobile site became indexed on about March 8. If it really doesn't support NOINDEX keyword then someone needs to file another bug request. — AlexSm 20:51, 22 March 2012 (UTC)[reply]
Is there an easy way to get a notice out to a number of Wikiprojects?
It may be necessary to do this by hand but I thought I'd ask here first. At Wikipedia:WikiProject Religion there is a plan to draw up a manual of style for religious articles, and (at least in my opinion) a notice needs to be placed on the talk pages of the associated Wikiprojects. Is there any easy way to do this? Thanks. Dougweller (talk) 12:25, 17 March 2012 (UTC)[reply]
Thanks, but doesn't that just deliver to user talk pages? Dougweller (talk) 17:32, 17 March 2012 (UTC)[reply]
If it's well defined and has consensus I can have one of my bots take care of this - would need a BRFA of course, or it could be done manually with AWB if it's small enough and a one off. RichFarmbrough, 20:25, 24 March 2012 (UTC).[reply]
JPEG size
Another user believes that jpeg files should have a width which is a multiple of 8. For example a width of 296px is good but 300px is not. I Suppose this has something to do with Jpeg#Block splitting but I'm not sure. Is there any good reason to prefer his practice in Wikipedia? –droll[chat] 22:39, 17 March 2012 (UTC)[reply]
I am still hopeful that someone will give me advice on this issue. If this is the wrong place to ask, I will gladly move the discussion elsewhere. –droll[chat] 02:46, 21 March 2012 (UTC)[reply]
It is hard to answer. When I was doing stuff with rendering graphics I always went for powers of 2 when creating the original artwork, so scaling by powers of 2 in either x or y were clean. But to look at an example, in this domain, things are not so simple. The reason is that we have, if you will, three numbers to cope with (per dimension), the original size, the size we upload and the display size. If we introduce artifacts in the first scaling then it doesn't matter how clean the server's scaling to display size is. So my take on this would be to upload at the original size for any externally sourced images, and to create new artwork at suitable multiples of powers of two (since we don't have any transforms other than scaling available) and let the WP servers and algorithms worry about everything else. RichFarmbrough, 20:34, 24 March 2012 (UTC).[reply]
"My sandbox" link is broken, and the help link isn't helpful.
At least with secure login to Wikipedia, the "My sandbox" link at the top of my page is broken. When I clicked "Send feedback on editing", I was limited to maybe 100 characters, so I was forced to be terse. The pound signs below really were pound signs; originally I had "1.WP secure login", etc., but digit + period was costing me five precious characters out of my meagre allotment.
header: How did you feel about editing Wikipedia Incubator?" [OWTTE]
[HAPPY FACE] [SAD FACE] [CONFUSED FACE] (I.e., I checked sad)
header: Tell us why!
Didn't WANT to! (Originally, "I wasn't TRYING to edit Wikipedia Incubator!")
#WP secure login
#click "My sandbox"
#Error: This page is unprefixed!
#Click unprefixed
#Help:contents.
Need specific help
I don't recall ever having that trouble before. And a link in an error message, of all things, should lead to specific, helpful information, not "Welcome to our huge help space. You're on your own. Good luck, you'll need it."
If you have registered an account you can find or create your own user sandbox here. For future easy access, you can put {{My sandbox}} on your userpage.
And that, or something like it, is what should be implemented in the user's page header, rather than a link that
looks right
but leads to an incomprehensible (to a non-editor) error message
with a link that seems to point to a specific help page for the problem
but which actually points to the whole forest of help pages.
(Analogy for that link: You're trying to find a certain kind of service in a strange town; you're pretty sure it's available, but you have no idea where such a shop might be. You stop at a convenience store for directions, and they sell you a map of the world.)
Thank you! I think I see my problem now: I had the Wikimedia login page bookmarked as the secure login page for WP. I see that WP's own secure login page now logs the user in to other WM sites. Was that ever not the case? b/c that would explain why I did it that way. -- Thank you also for putting my post in the appropriate place. I will respond there. --Thnidu (talk) 04:56, 18 March 2012 (UTC)[reply]
Thanks, switching my bookmark to point to https://en.wikipedia.org seems to have done the trick. Can I assume that your "Solution for Wikipedia to prevent this happening again" will reach the right eyes? --Thnidu (talk) 05:23, 18 March 2012 (UTC)[reply]
Note if anyone wants to know why the old secure server link goes to Incubator: as the /w/index.php?... link is not converted into the secure link correctly, the missing wikis script catches is and redirects you to Incubator. — Preceding unsigned comment added by MF-Warburg (talk • contribs) 05:40, 18 March 2012 (UTC)[reply]
You can't use interwiki links in a gadget URL... Anyway I wanted to say that I changed the missing.php script so it'll go to a descriptive page on Meta instead of Incubator (still needs deployment). The "unprefixed" link is also changed in the Incubator extension so it can be adapted to a more helpful page. SPQRobin (talk) 21:32, 18 March 2012 (UTC)[reply]
I made a fix [3] with wgServer+wgScript which worked everywhere, but it was reverted [4] with summary "secure.wikimedia.org is deprecated, as are the wgServer and wgScript variables". Is there are recommended way to make working url's in scripts at secure.wikimedia.org, or should we just accept that some scripts break in confusing ways there? Many users probably still have the old secure server bookmarked and it doesn't inform visitors that it's deprecated. PrimeHunter (talk) 19:35, 21 March 2012 (UTC)[reply]
Does anyone know a good way to conditionally pass a parameter to a template, such that if the condition is not met, the parameter is left undefined?
Let's say I have a template tempA that calls template tempB. tempA takes parameters foo and bar, which are normally passed to tempB. However, if the parameter bar is not given to tempA, it should not be passed to tempB. {{{bar}}} in tempB must be left undefined in this case, so tempB will use its own default for it that can vary depending on the other parameter, foo.
I'm currently using this trick in tempA to achieve this:
{{tempB | foo={{{foo|}}} | {{#ifeq:{{{bar|X}}}|{{{bar|Y}}}|bar|dummy}}={{{bar}}} }}.
With this code, if the bar parameter is not passed to tempA, dummy={{{bar}}} gets passed to tempB, which tempB will ignore as it doesn't use a dummy parameter. (If you want a non-hypothetical example, see Template:Service awards, which uses this trick to pass the caption2 parameter to Template:Service awards/core.)
This code is not particularly elegant and is hard to understand, so I was wondering if anyone knew of a better way of doing this? I feel there's something blindingly obvious that I haven't considered. – PartTimeGnome(talk | contribs) 16:29, 18 March 2012 (UTC)[reply]
OK; I hadn't seen that help page before. I guess there isn't a better solution with MediaWiki's current parser. Thanks for your help. – PartTimeGnome(talk | contribs) 21:33, 18 March 2012 (UTC)[reply]
Maybe nothing better in general, but for the condition you give {{#if:{{{bar|}}}|bar={{{bar}}}}} (does not pass empty parameters) or {{#ifeq:{{{bar|Y}}}|{{{bar|N}}}||bar={{{bar}}}}} (does not pass undef. paramters) does the job more simply and safely, you might even want to consider bar={{{bar|undefined}}} and testing for "undefined" in the target template. E&OE as ever. RichFarmbrough, 15:38, 24 March 2012 (UTC).[reply]
Additional note the examples 1 & 2 will pass an empty numbered parameter as written if they "fail" since the template call will end |}}. To deal with this futz around with including {{!}} in the conditional. RichFarmbrough, 15:40, 24 March 2012 (UTC).[reply]
Right, I see. The first three entries in the references list each have one superscripted letter (the last one of three or four) for which the backlink does not work. These are:
but I don't see what might cause these extra links to appear. --Redrose64 (talk) 21:27, 18 March 2012 (UTC)[reply]
thanx for your reply. this is what i did –
single instance of Template:refn in the reflist, in the order it appears on the text = same problem persisted.
single instance of refn, moved it to top of reflist = problem disappeared, everything ok.
first instance of refn moved to top of reflist, added 2nd instance of refn in the order it appears on text = no ghost ref links in the first 3 refs, but 1 extra ref link appeared in the original refn, as in Notes item 54, link c:abc [this is the current page]
first instance of refn moved to top of reflist. 2nd instance of refn anywhere in reflist but with its nested ref above it = 1 extra ref link appeared in the original refn, as above.
first instance of refn moved to top of reflist. 2nd instance of refn anywhere in reflist but with its nested ref below it = 1 extra ref link appeared in the original refn as above, plus cite error re: the nested ref of 2nd refn instance.
it seems the bug is still there but does not generate the same visible cite errors? thank you. 65.88.88.127 (talk) 15:53, 19 March 2012 (UTC)[reply]
I don't know if it's an error with my settings, but on Chrome I cannot edit templates. Instead they appear in a grey box that I can't edit. I do not have this problem on Firefox. JDDJS (talk) 02:08, 19 March 2012 (UTC)[reply]
I cannot reproduce this in Chrome. Here is an example unprotected template: Template:Seed. Do you see an "Edit" tab? What happens when you click it? It should contain the link http://en.wikipedia.org/w/index.php?title=Template:Seed&action=edit (if you are at http://en.wikipedia.org) and give a normal edit box. Have you tested whether you can currently edit non-template pages in Chrome? Have you tried whether you can edit when you are logged out? PrimeHunter (talk) 02:22, 19 March 2012 (UTC)[reply]
It's the templates that are in articles that I can't edit. It works when I log out. JDDJS (talk) 21:48, 19 March 2012 (UTC)[reply]
If you can edit Template:Seed then please name a specific template you can edit in Firefox but not Chrome, say exactly what you do to edit it in Firefox, and at which point it fails in Chrome. PrimeHunter (talk) 20:00, 20 March 2012 (UTC)[reply]
When I want to edit a reference template in article to be 30em instead of 22em, the template is in a gray box. I found out that if I click on the gray box and then press delete while my cursor is before the box, I can see the normal code and edit as I would normally. When I am logged out or editing from Firefox I do not see the gray box and instead just the normal code. JDDJS (talk) 17:11, 21 March 2012 (UTC)[reply]
I propose that we enable suffix searching. (We already have the ability to search for prefixes.)Curb Chain (talk) 21:28, 19 March 2012 (UTC)[reply]
Wikipedia uses Lucene for searching. It's a Open Source, anyone who wants to can add suffix support. The fact that it doesn't have it already probably indicates that it's not as easy as it sounds... :D "The reason why Lucene's QueryParser doesn't allow suffix queries ("*keyword") is the inefficient nature of such query types" [5] —TheDJ (talk • contribs) 15:26, 24 March 2012 (UTC)[reply]
Toolserver replication lag
Any toolserver experts about? The replication lag graphs suggest that copying of data from en.wp to toolserver s1 has stopped. Data is still being copied from commons to s1. --Redrose64 (talk) 17:19, 20 March 2012 (UTC) My estimate (to nearest 5 min) of the stop time is 19 March 2012 21:15 (UTC) --Redrose64 (talk) 23:08, 22 March 2012 (UTC)[reply]
Yeah, Asher Feldman is tinkering with some of the boxes; this typically involves taking some capacity out of circulations and then rotating. The replication lag is actually a MAX() type query, hence the high figures. I believe. - Jarry1250[Deliberationneeded] 18:02, 20 March 2012 (UTC)[reply]
A little more in this e-mail. Dpmuk (talk) 18:06, 20 March 2012 (UTC)[reply]
Here jira:MNT-1225 are some details on the problem, which extends well beyond tools and bots. I haven't been able to find an estimated time for completion. — cardiff | chestnut — 01:00, 23 March 2012 (UTC)[reply]
...I'm going to juggle which db watchlist queries go to during the migration, so nothing should be noticeable on the site.
Ha! I assume that this was all tested on a large non-production database first? I also hope that the thumbnail calculation that someone made about this taking 48 days to resolve is not correct. Fingers crossed for the Wikipedia brand.
Why dosen't MySQL allow for this kind of thing to be run in the background? Tim1357talk 02:51, 23 March 2012 (UTC)[reply]
Coz MySQL is a toy playing at being a big kid's database. There's ways this could have been done without taking out toolserver, but they would have had the effect of an inconsistent database for a while. And the database schema wouldn't have been identical to production. That's all in the past. Josh Parris 09:45, 23 March 2012 (UTC)[reply]
Current estimate is somewhere between 3 and 18 days (total). The lower end of the estimate has already been exceeded. Josh Parris 09:57, 23 March 2012 (UTC)[reply]
It is 4 days 6 hours now. Looks like this will be a record. --SupernovaExplosionTalk 04:15, 24 March 2012 (UTC)[reply]
6 days, 5 hours. @_@ We're approaching the replication singularity! SilverserenC 02:24, 26 March 2012 (UTC)[reply]
Yes, the replag graphs can't handle a delay of over 500,000 seconds (5 days 18 hours 53 min 20 sec). --Redrose64 (talk) 08:56, 26 March 2012 (UTC)[reply]
A site like Wikipedia should care for technology and invest more resources in this field. It is unfortunate this happens in Wikipedia. --SupernovaExplosionTalk 09:44, 26 March 2012 (UTC)[reply]
I don't know how this is handled, but if it's a case of being one volunteer and that we should not ask much because it is just one volunteer, maybe that is not enough. North8000 (talk) 10:52, 26 March 2012 (UTC)[reply]
It is investing - Wikimedia Labs is set to "internalise" key features of the Toolservers over the next year or so. - Jarry1250[Deliberationneeded] 11:51, 26 March 2012 (UTC)[reply]
Is there any sort of workaround for this? I believe a large number of projects are set to generate monthly reports on March 31st (I know WP:DPL does so), and the lack of such reports will, for the short term at least, gum up a lot of regular maintenance. bd2412T 19:32, 26 March 2012 (UTC)[reply]
More than 1 week (+1 day+ 12hours) is disastrous. My editcount is suck at 7992 for a good lot of time. extra999 (talk) 08:49, 28 March 2012 (UTC)[reply]
Same here; 7,567 for as long back as I can remember. Gosh, I don't even know how many edits I've done in these 8 days! But my Preferences still seems to be working. ~*~AnkitBhatt~*~ 14:36, 28 March 2012 (UTC)[reply]
I am hardly well-versed with any of the complex technicalities of running Wikipedia, but a week-long replication lag is disastrous for the continued running of the project. Edit counters have all become useless as edits for over a week have not been counted. In addition, some events such as WP:INDIA's Tag and Assess event have also been thrown out of gear due to wrong statistics. There seems to be no end in sight, and in my years on Wikipedia I have never had to encounter a replication lag as severe as this. Is there any fixed deadline allotted to rectify this problem? Because we need the replication lag to disappear fast, otherwise too many problems are going to build up. Perhaps setting a proper deadline and strictly sticking to it will speed up the rectification (if it hasn't already been done). ~*~AnkitBhatt~*~ 10:18, 27 March 2012 (UTC)[reply]
According to the latest Signpost, the toolserver issue is set to be resolved by Friday. bd2412T 14:28, 27 March 2012 (UTC)[reply]
March 30? Let's hope so. ~*~AnkitBhatt~*~ 14:41, 27 March 2012 (UTC)[reply]
MySQL rollback after the HDD filled up thus requiring a full dump from the WMF, I'm revising my estimates to early April. This isn't the worst:
Not enough disk space to hold enwiki
WMF discarded database when splitting clusters
Deleted replication file (took too much space)
Replication was ever-so-barely catching up
IIRC, the record hold is two months on dewiki. — Dispenser 18:52, 27 March 2012 (UTC)[reply]
For the 99% of us who aren't insiders on this, could you explain? Are you the person who is handling this? Thanks. North8000 (talk) 00:38, 28 March 2012 (UTC)[reply]
I'm sorry, early April? How early? Do you realize how much of a problem this is going to become if this issue isn't fixed by March 31? How can the issue be so severe? It hardly matters what the record is on some other language Wikipedia, enwiki is the biggest by far and needs continuous running. Cracks are appearing all over the enwiki now, for God's sake the lag is 8 days! ~*~AnkitBhatt~*~ 14:34, 28 March 2012 (UTC)[reply]
Edit count
The edit counter for user contributions seems to have stopped. The current replication lag is more than four days. Is anyone working on it? HandsomeFella (talk) 09:13, 24 March 2012 (UTC)[reply]
Catscan 2 has been showing the same error message for a while now: "MYSQL error : The MySQL server is running with the --read-only option so it cannot execute this statement [INSERT IGNORE INTO cat2 ( catname ) VALUES ( "Suburbs_of_Johannesburg" )]"--eh bien mon prince (talk) 08:31, 27 March 2012 (UTC)[reply]
I've removed the page name from the above link so it appears as a same-page link. This probably means that it will still work when this discussion is archived. Graham87 02:53, 28 March 2012 (UTC)[reply]
New functions and the old functions
Besides the fact that importscripturl will soon be deprecated, what are the benefits of mw.loader.load? Also, I've seen some scripts use mw.config.get('wgPagename'); (or something like that). What is the benefit of that compared to wgPagename? Finally, why is addOnloadHook deprectated and apart from $(func), with func being the function, what is the new function replacing it?
Thanks, --Kangaroopowah 03:13, 21 March 2012 (UTC)[reply]
You can use $(document).ready(func) instead of addOnloadHook. Also, if you pass a URI to mw.loader.load, it will load it for you, just importScriptURI did. — This, that, and the other (talk) 06:40, 24 March 2012 (UTC)[reply]
Infinite loop?
It struck me that one could mistakenly construct a template that transcludes itself, and that that could pose a risk to wikipedia.
Assume that a template named A is created. Assume that it contains the code:
{{A}}
Templates A and B transcluding each other would pose the same risk, as would any other infinite loop of template transclusions.
What could be the consequences? Are there barriers to prevent this from happening?
On the other hand, what would the server do if you made template "A" transclude "B" two times, "B" transclude "C" two times, "C" transclude "D" two times, and so on forty levels deep? The final template would simply be empty to avoid reaching the 2,048,000b page size limit, but would this take up an enormous amount of server resources? Reaper Eternal (talk) 20:19, 21 March 2012 (UTC)[reply]
It will actually say "Template loop detected". There are good reasons to want to do this, and it is hard to work around this barrier. RichFarmbrough, 20:39, 24 March 2012 (UTC).[reply]
See also bugzilla:25644 (limited recursions should be allowed). Helder 21:56, 24 March 2012 (UTC)[reply]
is it me or is wiki very slow?
Wiki seems very slow. Sometimes I can't load Colorado River at all - it never finishes loading. What's up? MathewTownsend (talk) 19:45, 21 March 2012 (UTC)[reply]
Don't abbreviate "Wikipedia" as "Wiki"! There are other wikis out there – Wikipedia is just one of them.My god yes. It keeps getting gradually slower and slower, and has been doing so for a while. PS. I'm an IT guy and I've gone all the regular routes to solve things on my end -- it's not my end that's the problem. I hate to complain about something free, but, this is becoming a problem for people who want to contribute -- and even those who just want to read. Equazcion(talk) 20:03, 21 Mar 2012 (UTC)
Ditto. Days. I thought it was my computer, except it's limited to Wikipedia. Today, my first login was maybe 8 - 9 hours ago. Man...today is the worst it's ever been. It takes forever to for a page to fully load. Take a nap while you wait for a Save or Preview to take effect in Edit. Right now, it looks like I'm on Server mw2. Don't know what it was earlier in the day, or in days past. This is seriously bad. Maile66 (talk) 20:35, 21 March 2012 (UTC)[reply]
Shortly after I posted the above, it magically cleared. For about two minutes. Then it reverted to worse than ever. It literally took minutes for this page to load itself. We'll see how long for it to Save Page.Maile66 (talk) 21:38, 21 March 2012 (UTC)[reply]
At that time, about an hour ago, yes and I did some E-mail instead. Now, here in New York, no. Jim.henderson (talk) 21:14, 21 March 2012 (UTC)[reply]
Wish I could get it to load that quickly it took 31 seconds for me. Keith D (talk) 23:48, 21 March 2012 (UTC)[reply]
This is an intermittent in-and-out problem. I can get it really fast at times. And then it will take a minute or two to load any page. Colorado river loaded in about 15 seconds for me. I clicked on My Watchlist and got "could not load twinkle options" up in the banner area. I've cleared the server cache on this Village Pump page, because it's slowest of all. I've cleared my browser's cache. But there are still problems. It comes and goes. Maybe those who are having no problems at all are just on a different server. Maile66 (talk) 23:54, 21 March 2012 (UTC)[reply]
I recall something about a database migration or something; maybe it has something to do with that? Or maybe I have no idea what I'm talking about. Actually the latter is incredibly likely. — Isarra (talk) 00:16, 22 March 2012 (UTC)[reply]
The intermittent slowness is horrendous for me (in Australia). It's been happening for over two days now and it renders WP unusable for much of the time. I have no problems with any other website. If I go to a toolserver edit counter page (say, http://toolserver.org/~tparis/pcount/index.php?name=Jimbo_Wales&lang=en&wiki=wikipedia) I see a message (that would seem to be related):
Caution: Replication lag is high, changes newer than 3 days, 2 hours, 11 minutes, 6 seconds may not be shown.
My experience mirrors GFHandel's. I'm also in Australia. DH85868993 (talk) 01:47, 23 March 2012 (UTC)[reply]
I've been experiencing the same. I'm in Japan, an APNIC IP like Australian users. It seems to me the cause of the slowness is transferring data from bits.wikimedia.org. It's really frustrating. Oda Mari (talk) 06:55, 23 March 2012 (UTC)[reply]
The replication lag problem (see #Toolserver replication lag above) affects everybody's edit counter, but it does not affect how quickly Wikipedia itself behaves. Basically, to avoid large reports and bots placing too great a strain on the normal Wikipedia databases, these processes obtain their data from a set of duplicate data. This duplicate data is held on toolserver, and so it is on toolserver that these reports and bots run. The toolserver data is supposed to be updated from the live Wikipedia almost in real time; there is often a very small delay in updating - a fraction of a second is normal. The replication lag is the time difference between the actual clock time and the timestamp of the most recent edit that has been copied to toolserver. The current replag is certainly abnormal with a delay of approximately 300,000 seconds; compare the replag for German Wikipedia which is showing single figures as I type this - the sort of replag that English Wikipedia normally experiences.
I have to admit that over the two days that this thread has been up, I have not experienced any problems with Wikipedia itself. The most likely cause is your local servers: I believe that I go through one in the Netherlands, and I'm pretty certain that users in Australia or Japan will not go through the Netherlands. --Redrose64 (talk) 09:06, 23 March 2012 (UTC)[reply]
Florida here, and I've been getting intermittent lag - oddly, pages usually half or mostly load, then sit there "trying to finish", as it were; usually a refresh makes them work, but then going on to the next page, slow again. It's quite annoying. - The BushrangerOne ping only 16:16, 23 March 2012 (UTC)[reply]
That's what I'm getting too now, intermittent half-loaded pages just hanging, with a refresh fixing it. Then it happens again. NYC here. Equazcion(talk) 16:17, 23 Mar 2012 (UTC)
(edit conflict) I've been experiencing the same loading issues for the last few days. Sometimes it works okay (not great, but okay). More of the time, it does what Bushranger says, it loads only partway and then stops. It does this in a variety of different contexts, including editing where it doesn't finish the edit window, for example, or just loading a watchlist, and it only loads part of it. Worst is sometimes I don't realize it hasn't finished loading when I edit and then it jumps throwing me off. FWIW, I'm in California and use the latest version of Firefox and a very fast cable connection. (Just got stuck on Show preview "transferring" and never finishing - had to hit the back arrow.)--Bbb23 (talk) 16:22, 23 March 2012 (UTC)[reply]
I'm from India and I'm also facing the problem sometimes. --SupernovaExplosionTalk 16:45, 23 March 2012 (UTC)[reply]
If it has nothing to do with IP, is it something related to my preference? I'm a TW user and use mono skin. And I select the old style view, meaning the one with the search box on the left on the page. Oda Mari (talk) 17:02, 23 March 2012 (UTC)[reply]
The problem is finding the right person to take note of the problem and fix it. I recollect having a performance problem like this before, and it took quite a long time to be fixed. Unfortunately, I don't remember who took the laboring oar and saw it through. Perhaps someone else does.--Bbb23 (talk) 17:04, 23 March 2012 (UTC)[reply]
I came here wondering whether it was just me. I've been having issues since 14:30 hours on 21 March. Pages are very slow to load, or are half loading. It's particularly noticeable with preview. Only a portion of the page loads, then slowly the rest of the article appears -- sometimes all of it, sometimes not. SlimVirgin(talk) 17:19, 23 March 2012 (UTC)[reply]
Later small pages load before earlier big articles: Going through Atlanta (Georgia) today, I have noticed with several long delays, there is not even a local partial cache copy (setting browser offline & redisplay shows no page content at all). Typically, in prior months, when a page began to display, there would be a cache copy after the initial "mini-cache" copy (View Source), before downloading images, but even that does not come after a long delay. I have always suspected that short articles transfer at faster priority than long articles, so I am not surprised that a long article might arrive after a more-recent short article view. However, the post-HTML-fixup format has also been bizarre in recent days. -Wikid77 (talk) 00:33, 24 March 2012 (UTC)[reply]
I'm editing from the U.S. and things seem slower than normal, I'm seeing similar issues to what SlimVirgin describes. Not enough that I can't edit, but just enough to be irritating if I'm trying to keep up with changes in the TFA. Mark Arsten (talk) 01:45, 24 March 2012 (UTC)[reply]
(From US) Recently, there have been periods of slowness interspersed with normal speeds. No discernible pattern that I can see. Chris857 (talk) 01:48, 24 March 2012 (UTC)[reply]
Also editing from US, and it seems to occasionally stall when loading a page. Hitting the refresh page button seems to fix it right up. StuRat (talk) 01:49, 24 March 2012 (UTC)[reply]
Yes. I (USA) noticed the change starting about 2 days ago. Including half loaded pages. Not a maybe. Took a dramatic turn for the worse. North8000 (talk) 02:30, 24 March 2012 (UTC)[reply]
Thanks MZ, I'd been wondering who we should be notifying. Equazcion(talk) 11:33, 24 Mar 2012 (UTC)
Thanks from me too, MZ. It's worse for me today than it has been since this started. SlimVirgin(talk) 16:37, 24 March 2012 (UTC)[reply]
It has been intermittently slow for me in New Zealand for at least the past 20 hours. Nurg (talk) 22:09, 24 March 2012 (UTC)[reply]
Additional issues
At the top of any page, both user menu items and page menu items are missing. Since they're missing, I can't recall what should be there, but it seems the Purge, Speedy Delete, those kind of things are missing.
Preferences - all tabs are gone under Preferences. Now it just rolls out into one long page, and I don't know if it's all there or not.
Edit mode does not totally load, either.
When typing in the Search bar, it no longer brings up any matched selections. Like that option is dead.
Overall speed issue, pages not loading entirely - much worse than yesterday.Maile66 (talk) 21:32, 24 March 2012 (UTC)[reply]
I'm not experiencing any of those issues. It sounds like that might be a problem with your javascript in your browser? Equazcion(talk) 22:15, 24 Mar 2012 (UTC)
As with everything else on the issues the last few days, not everybody gets the same case of measles, so to speak. If I experienced it elsewhere, I might agree with you. But I do not have these issues with Wikipedia's sister sites, such as the German Wikipedia. or the French Wikipedia. The Preferences options are there on the other Wikipedias. No problems with Commons or Wikisource. Just the English Wikipedia. Maile66 (talk) 22:53, 24 March 2012 (UTC)[reply]
Well, I'm just saying the general lag shouldn't cause those kinds of problems, at least not permanently. A simple hard refresh (usually Ctrl-F5 or Ctrl-Shift-R) might fix your issues, if you say other sites don't show the same errors. Equazcion(talk) 23:26, 24 Mar 2012 (UTC)
Well, thanks there for trying to help. But these additional issues have been going on at my end since yesterday. As for that post below you speedily reversed out, I was going to add something to the feeling of everything being fixed. By the end of the day yesterday, Wikipedia looked fixed at my end. I was so happy. I booted up this morning, and the pages downloaded in a flash. And then...it all came back. Right now, it's better. Some of the above issues still exist, but not all. We'll tune in tomorrow and see . Thanks for your input.Maile66 (talk) 23:37, 24 March 2012 (UTC)[reply]
I'm guessing that the points you raise are because of the unpredictable effects of pages that fail to load completely. At bugzilla:35448 is the report "Over a dozen users seem to be reporting a similar general slowness issue on the English Wikipedia...", however please note that those dozen are the ones who both know about the Village Pump, and took the time to post here—so I shudder to think how many people this issue is affecting (perhaps all over the world)? Has it been definitely ruled out that the current db migrations are related? The timing and duration of them and the current issues is one big coincidence if they aren't. Is it possible to pause those processes so that we can test responsiveness? The issues I've been experiencing for four to five days continue this morning (in Sydney, Australia), so I'm pretty well resigned to the fear that someone has broken our wiki. :-( GFHandel♬ 22:22, 24 March 2012 (UTC)[reply]
There is a post here confirming that there has been a problem since March 21, and offering a possible explanation, and a graph here that apparently illustrates it. SlimVirgin(talk) 22:59, 24 March 2012 (UTC)[reply]
I'm in the UK. Some time after 16:00 (UTC) today I returned to my computer after an hour or so away to find that I could not access several WMF sites through their "normal" URLs: English Wikipedia; Commons; Simple English Wikipedia; or Meta. Quite simply, nothing loaded, not even a completely unstyled page. I didn't try any others, so I guessed it was a general problem. I did find that the old secure serverdid bring pages back, but was noticeably slower than normal, but not unusably slow. This page showed the red "Service disruption" icon on the API row; and as shown here there was a downward spike on the "Availability over the last 24 hours" graph. I went out for a couple of hours: when I came back after 18:00 I found that the number of edits in my watchlist had taken a significant dip between 16:00 and 17:00, so I guess that many others were affected at this time. --Redrose64 (talk) 23:20, 24 March 2012 (UTC)[reply]
Just now I was waiting ages for an edit preview to finish displaying. Eventually pressed Escape and tried again and it displayed after a relatively short wait. It is as if something got lost in transmission and I just had to start the request again. Nurg (talk) 00:23, 25 March 2012 (UTC)[reply]
As of this very moment, all my Additional Issues - including one other I forgot to mention - cleared up at once. Downloading is still slow. But for this nano-second in time, the Additional Issues are cleared. Maile66 (talk) 00:28, 25 March 2012 (UTC)[reply]
fixed
The issue reported here, with pages stalling before being fully loaded, appears to have been a networking issue causing packet loss and timeouts, which should be resolved. Please reopen bug bugzilla:35448 and provide details if you can reproduce the issue at this time.--Eloquence* 01:28, 25 March 2012 (UTC)[reply]
Issue seems to be resolved for me since this comment was posted. Thanks! Equazcion(talk) 01:58, 25 Mar 2012 (UTC)
Firefox update causing svg > png problems.
In the update to firefox 11, they changed the way that image files are displayed so that when an image is opened in full size, it is centered in the screen and has a grey background. This is good for jpegs, but many of our .svg files have a see-through background, meaning that images like File:Beta-aminobutyric_acid.svg appear like this when they are rendered by mediawiki as pngs. Is there anything that we can be do to solve this? If not us, then is there someway to change firefox's settings? SmartSE (talk) 20:33, 21 March 2012 (UTC)[reply]
Can you give us a screenshot? I don't feel like installing Firefox. —Designate (talk) 23:10, 21 March 2012 (UTC)[reply]
Apparently this works. SD5 23:38, 21 March 2012 (UTC)[reply]
(edit conflict) It's like any other browser except really dark. Most of them do a light greyish, but Firefox is... dark (this colour). Honestly sounds more like a Firefox is dumb problem than something that we could do much about here, but for now you could do what SD5 suggests. — Isarra (talk) 23:41, 21 March 2012 (UTC)[reply]
Two solutions found from googling:
Technical: Expand omni.ja, found in Firefox's root program folder, using an archive program (7zip etc). Edit TopLevelImageDocument.css, search for "background-color:", and replace "#222" with "#FFFFFF".
Just realized the addon was already mentioned above by SD5, my bad :) Equazcion(talk) 00:17, 22 Mar 2012 (UTC)
Thanks for the replies - I agree that it is a bit stupid of firefox. I'll try using that add on later. SmartSE (talk) 13:43, 22 March 2012 (UTC)[reply]
At the top of my watchlist I have this - note the visible HTML markup <strong>...</strong>. I don't have StumbleUpon, but I am using Firefox (3.6.28) under Windows XP. Which is the javascript file that generates that banner - alternatively, what's causing it? --Redrose64 (talk) 19:35, 22 March 2012 (UTC)[reply]
The banner comes from MediaWiki:Geonotice.js. I'll ping User:HJ Mitchell, who was the last to edit that particular notice and has several edits adjusting the formatting of it. DMacks (talk) 19:44, 22 March 2012 (UTC)[reply]
I believe we're using a partial wikitext parser implemented in JavaScript somewhere to parse those messages; it is undoubtedly fragile, and seems to have broken here. Ucucha (talk) 19:59, 22 March 2012 (UTC)[reply]
OK, it's now showing visible Wiki-markup (three apostrophes instead of bolded text), i.e. '''Liverpool''', 24 March; etc. --Redrose64 (talk) 20:22, 22 March 2012 (UTC)[reply]
Now showing without markup. --Redrose64 (talk) 12:01, 23 March 2012 (UTC)[reply]
Robots.txt and wrong casing
Hi. After reading above I decided to do some googling to see how well MediaWiki:Robots.txt is being handled. There seems to be a few dozen pages (perhaps more) that don't use the standard sentence case in page titles and are at "Wikipedia:Articles for Deletion" rather than "Articles for deletion". [7] This probably happens for other areas around Wikipedia. Should all of those pages be moved or should something be added to fix this on Robots.txt? Killiondude (talk) 22:42, 22 March 2012 (UTC)[reply]
There also seems to be an issue where the trailing slash can sometimes be encoded. Right now MediaWiki:Robots.txt has entries such as "/wiki/Wikipedia%3AArticles_for_deletion/" and "/wiki/Wikipedia:Articles_for_deletion/" (accounting for an encoded semicolon). But robots.txt is prefix-based, so if spiders comes across "/wiki/Wikipedia%3AArticles_for_deletion%3F" (where the trailing slash is encoded), there isn't a match and it's indexed by search engines. Silly, silly. --MZMcBride (talk) 22:47, 22 March 2012 (UTC)[reply]
They should be moved, for various reasons. RichFarmbrough, 20:44, 24 March 2012 (UTC).[reply]
Can someone please tell me how I permanently disable these messages? I'm not interested in any of them. Ever. It seems that every day, a new one pops up, which I have to "dismiss". Frankly I'm sick and tired of it. Parrotof Doom 14:51, 23 March 2012 (UTC)[reply]
The first will disable the centralnotice and the second will disable the messages about the current RFCs, etc. Good luck! Reaper Eternal (talk) 15:18, 23 March 2012 (UTC)[reply]
I had to create that page, but I've done that, pasted the above text and saved, and the messages are still there. Parrotof Doom 18:56, 23 March 2012 (UTC)[reply]
Bypass your browser's cache. Usually ctrl+F5 will do this. Killiondude (talk) 19:15, 23 March 2012 (UTC)[reply]
I'm on Chrome. No combination of keys gets rid of them, and even clearing my cache in the tools menu doesn't get rid. They're still there, stubbornly telling me about things I'm not interested in. :( Parrotof Doom 19:34, 23 March 2012 (UTC)[reply]
That common.css page also says this "The accompanying .js page for this skin can be added at User:Parrot of Doom/common.js." - is that important? Parrotof Doom 19:36, 23 March 2012 (UTC)[reply]
Both codes work for me. :/ (And by the way, the .js page is not needed.) Reaper Eternal (talk) 19:47, 23 March 2012 (UTC)[reply]
I've just tried the both suggested forms (with periods, and with hashes) and can confirm that in both Chrome and Opera the sitenotice is still present but the watchlist messages are removed. --Redrose64 (talk) 20:42, 23 March 2012 (UTC)[reply]
Hmm, I've clicked nothing but they seem to have gone for now. I'll keep an eye out, but thanks very much for all your help chaps. Parrotof Doom 21:37, 23 March 2012 (UTC)[reply]
Nope, they're back. It seems Wikipedia insists that I read its spammy messages. Parrotof Doom 10:52, 24 March 2012 (UTC)[reply]
I've just tried that, the page is now blocked but the spam still appears. Who exactly is responsible for these messages? I may just go over there and insist they remove their unwanted crap from my watchlist. Parrotof Doom 23:06, 24 March 2012 (UTC)[reply]
I couldn't find anything that looked like what was being described, and I don't have the knowledge to interpret these questions, but this, this and this are all questions on the Help Desk that have gone unanswered. I suggested coming here, but they might not come back.— Vchimpanzee· talk·contributions· 21:48, 23 March 2012 (UTC)[reply]
Ref tool bar
I don't have the "cite" tool bar, I have no idea where is this coming from. Worked fine a couple of days ago. --Tachfin (talk) 01:30, 24 March 2012 (UTC)[reply]
I'm referring to the "cite tab" which you can see in in this image: File:RefToolbar_2.0b.png. The "enhaced editing toolbar" is enabled, as well as Java. I use Firefox & the default skin (Vector). I'll try logging out and clearing the cache and tell you if it comes back. Tachfin (talk) 06:46, 24 March 2012 (UTC)[reply]
I logged out and it was there. I re-logged-in and the "Cite tab" is back. Thanks for the tips --Tachfin (talk) 06:51, 24 March 2012 (UTC)[reply]
List of Recent Edits for WikiProjects Quit Working
I have been using the following link to watch recent edits for specific WikiProjects. It broke in the past months and I was looking for an alternate solution. Also, the link for a userbox is broken too. Hopefully someone else has a toolserver account to continue this support and/or hopefully there is another simple solution. Please suggest solutions! If this is the wrong place to ask this question, then please point me to the correct page instead of throwing rocks at me :-) • Sbmeirow • Talk • 04:28, 24 March 2012 (UTC)[reply]
[http://toolserver.org/~tim1357/cgi-bin/wikiproject_watchlist.py?template=WikiProject+Electronics&limit=100 Last 100 edits of all WikiProject Electronics articles]
User:Femto bot maintains on-wiki pages where all project pages are listed, for selected projects, these are for related changes to projects. Example for Connecticut. (Just enabled me to rvv..) RichFarmbrough, 15:47, 24 March 2012 (UTC).[reply]
Categories transclusion lag
I created the Category:Johannesburg Region A through Category:Johannesburg Region G and edited the Template:Infobox South African subplace to have it transclude these categories when appropriate. That was a week ago, but the categories remain empty or populated by only a handful of articles. Even though the right category does appear at the bottom of the article, they might not show up in the category page itself (see Aeroton for an example of this). What is causing this issue, and how can I solve it?--eh bien mon prince (talk) 12:09, 24 March 2012 (UTC)[reply]
There's laggy browsing that's presumably caused by the 5-day server replication lag. I'm not sure if replication lag = category listing lag too, but with all the various lags we're experiencing, it's probably all the same issue. Equazcion(talk) 12:44, 24 Mar 2012 (UTC)
A null edit (or any other edit) of an article will force an update of link tables for that article. A purge will not. I made a null edit of Noordwyk so it now appears in Category:Johannesburg Region A. It should eventually happen by itself but it sometimes takes a long time. Don't make a bunch of null edits unless it's needed to fix a more significant problem. PrimeHunter (talk) 13:13, 24 March 2012 (UTC)[reply]
The #Toolserver replication lag should not affect the update of category pages. Cat page updates are part of the job queue, where the backlog is hovering between 2000 and 2500 jobs. This is peanuts compared to how it was when seriously stuck just over a year ago. --Redrose64 (talk) 19:14, 24 March 2012 (UTC)[reply]
Hiding a particular user from watchlist
I swear this has been asked before, but I couldn't find an answer on the FAQ or the watchlist helppage. Anyway, is there a way to hide a user from your watchlist. Specifically for me Helpful Pixie Bot (talk·contribs) which does a lot of mundane edits. I know I can hide bots or minor edits, but I'd like to watch other edits of those types. Hot StopUTC 18:09, 24 March 2012 (UTC)[reply]
For that, the rows in Special:Watchlist would need either a CSS class= or id= which related to the user who carried out the action, against which you could apply the display:none; style; unfortunately I can't find a suitable class= or id= --Redrose64 (talk) 19:33, 24 March 2012 (UTC)[reply]
Ahh. Thanks anyway for the explanation. Hot StopUTC 20:34, 24 March 2012 (UTC)[reply]
It might be possible with Javascript. I don't know how (since I don't know javascript), but maybe someone else knows? Rchard2scout (talk) 23:00, 24 March 2012 (UTC)[reply]
Well, there is a button to hide all bot edits. Would that be sufficient? Chris857 (talk) 23:22, 24 March 2012 (UTC)[reply]
I prefer not to do that, since there are bots I'd like to see. Hot StopUTC 05:47, 25 March 2012 (UTC)[reply]
Wow that was odd. I purged the page and it seemed to have fixed it. For posterity's sake: It was the "The requested page or revision cannot be found" note generally found when reaching a page using a diffid after it has been deleted (or by a variety of other methods). Killiondude (talk) 02:35, 25 March 2012 (UTC)[reply]
Since the last two days I'm having the following problem in wikipedia and in wikipedia only. When I want to edit an article, the edit page opens and then, after a couple of seconds, suddenly the text disappears. I can add new text, but when saving all previous text is lost.
Changing from the vector skin to monobook doesn't change the problem. I'm using Windows 7, browser: Firefox 11.0
Using on the other hand IE9, I can make an average of two edits before the edit mode stalls and gives for this article Leucorhynchia (as an example) the following result ( a printscreen): Commons:File:Printscreen of the problem with editing in wikipedia.png. The layout is gone and wikipedia doesn't react anymore to mouse commands.
I don't have any problems with other websites, nor with any other programs on my computer. A thorough check for viruses, anti rootkits or malware gives negative results.
Does any one have any idea what is going on and how to help me out of this mess ? JoJan (talk) 17:13, 25 March 2012 (UTC)[reply]
Not exactly, since the pages load very quickly in Firefox or in IE (I have a wireless vdsl-cable connection). It's just when editing them causes the problems to begin. This morning I could upload several images to the Commons without any problem, it took just 2 or 3 seconds per photo. The problem is inside the en.wikipedia. And as to a possible connection to any problems with the servers, I reside in Europe. I really would appreciate if this could be fixed in a short while because I have a lot of contributions to make in our project. JoJan (talk) 21:52, 25 March 2012 (UTC)[reply]
Are you using the monobook, or the vector skin ? The Javascript for your vector skin was broken (now fixed) and the javascript for your mono book (link) was also totally broken. I blanked the latter for you (doubt it was doing anything at all for you other than thoroughly breaking your browsing experience), and if you want to have some of those elements enabled, you can ask for help here and I'm sure someone will be able to assist you to properly enable those. —TheDJ (talk • contribs) 22:33, 25 March 2012 (UTC)[reply]
I'm using the vector skin. But, even after the changes mentioned here above, the problem still persists. Today I only could make a few edits (see my history). After adding text to Turbonilla ignacia and saving it, I could not add the image I just has uploaded to the Commons. The edit mode showed the text of the article for a second or so and then removed it. If it is a problem with javascript, as suggested, I have the two following plug-ins installed in Firefox: Java Deployment Toolkit 6.0.310.5 (NPRuntime script Plug-in Library for Java(TM) Deploy) and Java (TM) Platform SE 6 U31 6.0.310.5 (Next generation Java plug-in 1.6.0_31 for Mozilla browsers). Do I need them ? Should I disable them ? Or is the solution to be found elsewhere ? JoJan (talk) 15:03, 26 March 2012 (UTC)[reply]
I've installed Google Chrome and here everything is working fine. But I can't used Firefox or Internet Explorer anymore to edit wikipedia. JoJan (talk) 14:11, 28 March 2012 (UTC)[reply]
It seems the existing articles using it are working, but the ID numbers used in the template don't match the IDs from the site. Are the docs for the template lacking information? This is the link to the details Wikipedia:Help_desk#Finding_the_ID.23_for_Template:British_pathe. I also mentioned it in the template talk section.--Canoe1967 (talk) 19:19, 25 March 2012 (UTC)[reply]
importScript from another language project?
Is there a way to import a script hosted on another Wikipedia language? Equazcion(talk) 19:39, 25 Mar 2012 (UTC)
Pretty sure those work, though you do have to get the entire file. But given mediawiki's new ability to use subpages for scripts withough one having to import them all, well... this could become THE FUTURE! — Isarra (talk) 19:45, 25 March 2012 (UTC)[reply]
Hehe. Well, the JS one doesn't seem to work for me (I tried with and without the '//' in case you added that to comment out the line on this page). I'm testing with a script I know to be working here on en, see User:Equazcion/vector.js. Thanks for the help though, it's good to know this is possible even if I'm not getting the code right yet... Equazcion(talk) 20:04, 25 Mar 2012 (UTC)
You need to use //en.wikipedia.org/w/index.php?title=User:Equazcion/ContribsTabVector.js&action=raw&ctype=text/javascript rather than //en.wikipedia.org/wiki/User:Equazcion/ContribsTabVector.js&action=raw&ctype=text/javascript -- WOSlinker (talk) 20:11, 25 March 2012 (UTC)[reply]
Ah I see, gorgeous. Thank you both. Equazcion(talk) 20:16, 25 Mar 2012 (UTC)
Excellent. For random reference you can also replace the first & in what you had with a ?, but what WOS linker said is definitely the less messy in the long run. Point is you need a ? before you can use &s, though. — Isarra (talk) 20:17, 25 March 2012 (UTC)[reply]
On Talk:IPv4 address exhaustion, the latest section does not insert the normal "/*section*/" into the edit summary when I edit it. Apparently, something is messed up with sections with spaces in them.Jasper Deng(talk) 05:11, 26 March 2012 (UTC)[reply]
Searches of Wikipedia with ISBN-10s do not find the corresponding ISBN-13s and vice-versa.
When hyphenated ISBNs wrap at the hyphens (which only occurs in some browsers, e.g., Safari), it makes them more difficult to read and more difficult to select and copy.
Hyphens make it more difficult to search for the ISBNs on the Wikipedia with web search engines, e.g. Google or Yahoo, since quoted hyphenated search strings often must be used to find the target ISBNs.
Proposed enhancements
Modify Wikisoftware so that:
Hidden ISBNs are added in the formats which are not displayed: ISBN-10s and ISBN-13s, hyphenated and unhyphenated, which will serve as targets of web search engines regardless of the format used as the search string. (A hidden alternative hyphenated target will only be added when the ISBN in the edit window text contains hyphens.)
ISBNs are displayed using "nowrap".
Example 1
ISBN 0-8288-2586-6
current code: <a href="/wiki/Special:BookSources/0828825866" class="internal mw-magiclink-isbn">ISBN 0-8288-2586-6</a>
The solution will be automatic and invisible to editors. The hidden numbers will not appear in the edit window as a source of confusion.
Editors can add ISBNs with or without hyphens.
Bots can still be used to add hyphens to ISBNs that lack them, thus bringing Wikipedia into conformance with the ISBN standard.
Either ISBN-10s or ISBN-13s can be added by editors, since the hidden search targets can be derived from either (except for ISBNs which do not have an ISBN-10).
Disadvantages
The hidden ISBNs are not typically found by the local browser page search function.
Questions
Will the interconversion of ISBNs and/or formatting have a significant adverse impact on Wikipedia performance?
What does an ipv6 userpage look like? This change is coming in just 2 months so I'm surprised it isn't being talked about more. I'd think those long addresses with all those colons would break a bunch of regex's and so forth. I read that about 1% of all traffic to ipv6-enabled sites (MediaWiki is) will occur over ipv6 starting that day. That seems significant. Equazcion(talk) 04:23, 27 Mar 2012 (UTC)
IPv6 address contribution pages are without any contractions using ::, but still omitting leading zeros in each group of 4 digits. All letters are capitalized. This is all done by MediaWiki when dealing with other possible formats. Thus, 2001:db8::1 is read by MediaWiki as 2001:DB8:0:0:0:0:0:1.Jasper Deng(talk) 04:26, 27 March 2012 (UTC)[reply]
I just bought a 2005 Aztec. It has very low miles & I like the truck alot. My dash is lit up with 3 things,service vehichle soon-code po420 on the tester. This means co2 sensor or catalatic converter,the awd disable light is on, and the airbag light is on!
Have you tried Wikipedia's Reference Desk? They specialize in knowledge questions and will try to answer just about any question in the universe (except technical issues about using Wikipedia, since that is what this page is for). Just follow the link, select the relevant section, and ask away. I hope this helps. --Redrose64 (talk) 22:47, 26 March 2012 (UTC)[reply]
The current tablesorter algorith is "unstable": it does not preserve the relative order of rows where the cell content is equal.
For example, clicking the "Col 1" sort button twice in the table below should preserve the order of rows 2 to 8 (or invert them then restore them). But instead, the tie-break rows sort pseudorandomly (returning to the orginal order after 6 clicks instead of 2):
Col 1
Col 2
Col 3
aa
aba
1
ab
abb
2
ab
abc
3
ab
abd
4
ab
abe
5
ab
abf
6
ab
abg
7
ab
abh
8
ac
abi
9
It is confusing for tables to change the order of rows with identical sort values. Since Wikipedia articles are not expected to contain extremely large tables, computational optimisation should be less important than intuitive operation.
How to make includeonly category appear only in mainspace
This should be an easy one, though I'm not sure how to do it myself. I want to make the includeonly category in {{Studio Pierrot magical girl series}} appear only when the template is transcluded in mainspace. I've seen it done before, but I can't on which templates I saw it. Thanks for your help! ···日本穣? · 投稿 · Talk to Nihonjoe · Join WP Japan! 06:03, 27 March 2012 (UTC)
[reply]
A question about Special:Random was raised at Wikipedia:Reference_desk/Mathematics#Special:Random and I suggested a simple change could make the chances for article much more even. Currently some articles are much less likely to be chosen than others if Wikipedia:FAQ/Technical#random is what really happens. The suggestion is to associate the golden ratio multiplied by the page index and take the fractiona part as Page Random rather than assigning them so randomly. As a person says there "This suggestion is actually a very neat, simple replacement approach, and has the minor drawback that it uses additional state: the latest index number. It also does not deal with page deletions perfectly, but ignoring that, the distribution of page probabilities will, I guess, be quite tightly bounded, not extending to zero as with the exponential distribution" The good property of the golden ratio here is related to it having the slowest convergence of any irrational number as a continued fraction. Dmcq (talk) 23:07, 27 March 2012 (UTC)[reply]
You should divide by the square root of two and adjust for the precession of the equinoxes. But seriously, it might not be a big deal to tweak the value computed for page_random that is stuck into the page table originally. At least, on your personal installation of MediaWiki. Information about a page will reside in a certain index in the page table. Unclear whether it is best to use the page index or the page_id to compute your improved value for page_random. (Does MediaWiki compact the page table when it gets holes in it? A certain page_id would then move to a new index). If you were doing statistics you might want some reproducibility. Here is an example of someone tweaking their MediaWiki installation by using a different value for page_random. EdJohnston (talk) 04:56, 28 March 2012 (UTC)[reply]
In reviewing and motivating a change to the random page selection algorithm, other issues will have to be considered.
The current algorithm is robust and simple (as is the suggested alternative).
Because the pool of articles is large, that some articles have very low probability of being selected will not significantly affect the user experience. This is possibly what underlies the apparent cynicism in the comments above...
Most importantly, a policy for relative probabilities of articles should be determined as a first step in any change: the experience of randomly selected articles may be significantly improved if some non-uniform probability weighting is chosen, e.g. weighted by article quality or page hits.
Yes the policy is the thing. If you want to slant to the better, or worse!, articles measured some way then you'd need to do something like the one adjusting the size of the gaps periodically. Getting the gaps much more even as using the golden ratio does may not change an individual experience, but it would mean that overall the various pages are all looked at by somebody every so often rather than some being practically totally invisible. Dmcq (talk) 10:58, 28 March 2012 (UTC)[reply]
Revdel issue
I deleted a revision at Talk:Pedophilia earlier today, and a misguided editor was able to successfully revert it (more precisely, they were able to revert a reversion after I'd revdel'd the original edit). When the matter was called to my attention, I was able to see the successful reversion, noting that for some reason the initial redaction didn't deal with the succeeding Sinebot diff: perhaps the redaction was interrupted? I followed up and revdel'd all of the revisions, and this time it appears to have worked. Is this a bug, or an interruption in execution, or ... ? Acroterion(talk) 01:25, 28 March 2012 (UTC)[reply]
You have to hide Sinebot's edit as well (and any other edits made after the offending revision but before the offending text was removed), because each version of the page will contain the offending text until the text is removed. Obviously, this tends to make a mess of the history (and the more revisions, the bigger the mess), which is a known issue but is at least better than the old-fashioned method. HJ Mitchell | Penny for your thoughts? 01:52, 28 March 2012 (UTC)[reply]
Yes, on further thought I normally kill the Sinebot revs too: for some reason I thought this morning that the Sinebot edit was picked up automatically. I was in a hurry to leave and didn't check. Acroterion(talk) 03:03, 28 March 2012 (UTC)[reply]
But the situation was like this:
[1] User1 added a troll message to talk page.
[2] Sinebot added signature to [1].
[3] User2 (me) reverted to revision before [1].
[4] Acroterion did a revision delete of [1].
[5] User3 undid [3].
I understand how it would be possible to view revision [2] and see the troll message. But how was [5] possible? Also, after the above, I tried clicking "undo" next to [3] and I was able to see the troll message in the resulting window. Since then, [2] and [5] have been revision deleted, and there is now no "undo" button. That makes sense, but why was there an undo button after step [4]? Johnuniq (talk) 06:47, 28 March 2012 (UTC)[reply]
Because Acroterion needed to get rid of [2] as well. The text was still available in Sinebot's revision, which must be how he was able to revert your edit. RevDel is a bit fiddly like that, and you have to make sure you hide every revision that contains the text you're trying to get rid of. HJ Mitchell | Penny for your thoughts? 12:33, 28 March 2012 (UTC)[reply]
Yes, causing problems :( And when I view the history it shows nothing, strange or maybe there's a sysop involved. By the way, the category is enough to be filed for deletion. extra999 (talk) 08:58, 28 March 2012 (UTC)[reply]
Does everyone have the span/dir/ auto message now on top of almost every Wikipedia page they consult?
Does everyone now have the span/dir/auto message on top of almost every Wikipedia article they consult?
wikimedia.org.
.png)
Have you tried Wikipedia's Reference Desk? They specialize in knowledge questions and will try to answer just about any question in the universe (except technical issues about using Wikipedia, since that is what this page is for). Just follow the link, select the relevant section, and ask away. I hope this helps. --Redrose64 (talk) 22:47, 26 March 2012 (UTC)
Done [8] Incnis Mrsi (talk) 07:02, 27 March 2012 (UTC)
···日本穣? · 投稿 · Talk to Nihonjoe · Join WP Japan! 07:09, 27 March 2012 (UTC)
![[1]](http://www.gosoftworks.com/download/wiki/What_I_see.png){kind=link}
![[2]](http://www.gosoftworks.com/download/wiki/What_I_want.png){kind=link}
