Policy | Technical | Proposals | Idea lab | WMF | Miscellaneous |
- Table of contents
- First discussion
- End of page
- New post
Newcomers to the technical village pump are encouraged to read these guidelines prior to posting here. Questions about MediaWiki in general should be posted at the MediaWiki support desk.
New toy to play with: Gradient background
Introducing {{Gradient}}. Still a bit a work in progress. Works in Firefox, Safari and Chrome, but not in Opera and IE (yet). It seems Microsoft filters are blocked, and I am still looking for a way around that. — Edokter • Talk • 17:16, 28 September 2010 (UTC)
- OMG I didn't even know there was a -moz gradient property. I am so in love. /ƒETCHCOMMS/ 18:13, 28 September 2010 (UTC)
- That is beyond awesome. Just one thing... is there a way to make it work within tables? sonia♫ 22:51, 28 September 2010 (UTC)
- Within hardcoded HTML tables should be no problem, but wikitables may require some hacking around with {{!}}. — Edokter • Talk • 22:55, 28 September 2010 (UTC)
- How would I do that? I've screwed around with it in my emailnotice, but can't get it to work without just copying in the code. sonia♫ 00:12, 29 September 2010 (UTC)
- Should be working now; I removed all the linebreaks from the template, and now seems to work, even without {{!}}. — Edokter • Talk • 00:27, 29 September 2010 (UTC)
- Damn. Did I manage to miss a linebreak?
Thanks! sonia♫ 00:28, 29 September 2010 (UTC)
- Yep, just before the disabled MS filter. — Edokter • Talk • 00:36, 29 September 2010 (UTC)
- It is beautiful. But from a usability point of view, overuse of gradient is quite distracting for the eye and reduces readability. Too much useless strain on the eye. Dodoïste (talk) 00:40, 29 September 2010 (UTC)
- Only if use with stoopid colors. — Edokter • Talk • 00:43, 29 September 2010 (UTC)
- Not only. I meant overuse in a matter of quantity. Not really in a matter of color choice quality because soccer projects will have their way anyway (as well as many other projects, its just an example so don't take it personally). Do you consider Template:Gradient/testcases as good examples? I absolutely don't. There are too may contrast and color variations, and the eye needs to get adapted at every change of contrast. In the end, it simply makes it waaay longer to read. Dodoïste (talk) 01:12, 29 September 2010 (UTC)
- Only if use with stoopid colors. — Edokter • Talk • 00:43, 29 September 2010 (UTC)
- It is beautiful. But from a usability point of view, overuse of gradient is quite distracting for the eye and reduces readability. Too much useless strain on the eye. Dodoïste (talk) 00:40, 29 September 2010 (UTC)
- Yep, just before the disabled MS filter. — Edokter • Talk • 00:36, 29 September 2010 (UTC)
- Damn. Did I manage to miss a linebreak?
- Should be working now; I removed all the linebreaks from the template, and now seems to work, even without {{!}}. — Edokter • Talk • 00:27, 29 September 2010 (UTC)
- How would I do that? I've screwed around with it in my emailnotice, but can't get it to work without just copying in the code. sonia♫ 00:12, 29 September 2010 (UTC)
- Within hardcoded HTML tables should be no problem, but wikitables may require some hacking around with {{!}}. — Edokter • Talk • 22:55, 28 September 2010 (UTC)
- That is beyond awesome. Just one thing... is there a way to make it work within tables? sonia♫ 22:51, 28 September 2010 (UTC)
- (←) Testcases is only a technical testbed, although the navbox there looks quite stunning now. :) Everything can be overused, but I believe this template can be put to good use. — Edokter • Talk • 01:17, 29 September 2010 (UTC)
- It would be great if it was only used in table headers, for example. Thus, it would not disturb the reading of the main content. Dodoïste (talk) 01:22, 29 September 2010 (UTC)
- It is not up to me to restrict any use. I ammended the documentation and I expect those wanting to use to do so with care. — Edokter • Talk • 01:28, 29 September 2010 (UTC)
- I've checked the testcase and I can only say that there's no way you can fix the navbar color leak. Because in the {{navbox}}, the style defining the {{navbar}} is as follow: {{Navbar|{{{name}}}|fontstyle={{{basestyle|}}};{{{titlestyle|}}};border:none;|mini=1}}. So unless you can justify the editprotected to remove the predefining fontstyle of navbar, the gradient leak will persist. Or you have to give up the gradient on the title row. -- Sameboat - 同舟 (talk) 01:37, 29 September 2010 (UTC)
- It is not up to me to restrict any use. I ammended the documentation and I expect those wanting to use to do so with care. — Edokter • Talk • 01:28, 29 September 2010 (UTC)
- It would be great if it was only used in table headers, for example. Thus, it would not disturb the reading of the main content. Dodoïste (talk) 01:22, 29 September 2010 (UTC)
- I was looking into creating nifty iPad-esque buttons on GeoTemplate the other day, but it uses a color stop 3/4 way through for more depth. Would it be possible to add it to this one or create another template, perhaps {{button}}. — Dispenser 19:47, 29 September 2010 (UTC)
- That is a tough one... -webkit-gradient uses quite a convulted syntax for adding color stops. I don't see it going into {{gradient}}, as it is coded to share common features supported between browsers. So perhaps coding it into the button is the better way to go. — Edokter • Talk • 20:39, 29 September 2010 (UTC)
Two more toys
{{Border-radius}} and {{Box-shadow}}. Have fun. — Edokter • Talk • 16:49, 29 September 2010 (UTC)
- These doesn't cause readability issues. :-) They are great, and it's only the beginning, since CSS3 is already fairly well implemented in every good browser (and I meant good as opposed to major because of IE). Plus, IE9 is finally coming. ^_^ Dodoïste (talk) 17:01, 29 September 2010 (UTC)
- too much fun. I've gone ahead and asked to have these added to {{quote box}} (Template_talk:Quote_box#adding_some_minor_coolness). You might want to weigh in there on technical tweaks before it gets implemented, if you have any ideas. --Ludwigs2 17:54, 29 September 2010 (UTC)
For an exapmle of how much typing these templates can save, see this edit. — Edokter • Talk • 23:00, 1 October 2010 (UTC)
- Please don't turn Wikipedia into a cheap Powerpoint presentation. Use these as little as possible. —Noisalt (talk) 04:41, 7 October 2010 (UTC)
- Yes keep them very limited in project space, but for other stuff, "Woooooooooohoooo!". Rich Farmbrough, 09:30, 7 October 2010 (UTC).
- Yes keep them very limited in project space, but for other stuff, "Woooooooooohoooo!". Rich Farmbrough, 09:30, 7 October 2010 (UTC).
Nice one! I've already been using box-shadow and its variants on certain Signpost templates... I think subtle use looks great with the new Vector skin. — Pretzels Hii! 18:37, 9 October 2010 (UTC)
Adding as improvement to wikipedia boxes
see the discussion at Template_talk:Quote_box#adding_some_minor_coolness and the test cases at Template:Quote_box/testcases. my feeling is that these make for some nice improvements to the interface, and so long as they fail invisibly in browsers where they fail, we might as well start a discussion about using them more broadly in the project (either as separate templates or as additions to the core CSS classes). need some broader opinions on that, though. --Ludwigs2 23:19, 3 October 2010 (UTC)
characters that take up zero space?
Are there any characters that take up zero space?
Control characters maybe?
I am using but it takes up too much space and messes up my layout.
(Some columns refuse to go to zero width when they are supposed to)
Template:Nuclides#Example_charts
The template code demands that I put something there before it will give me a carriage return so I cant just leave it empty.
Just granpa (talk) 04:02, 30 September 2010 (UTC)
- Have you tried using <nowiki /> as a spacer? The parser evaluates that to nothing at all, so if you need less than one space that might work. — Gavia immer (talk) 04:21, 30 September 2010 (UTC)
- Yes! That did it. The layout is now perfect. Thank you. Just granpa (talk) 04:51, 30 September 2010 (UTC)
- For future reference, the character you asked for does exist: ​. Anomie⚔ 11:01, 30 September 2010 (UTC)
- You can also use {{ns:0}}, which also evaluates to a clean empty string. --NYKevin @739, i.e. 16:43, 4 October 2010 (UTC)
- ​ is available as {{zwsp}},
or &zwsp;-- being Zero White Space. —Preceding unsigned comment added by DePiep (talk • contribs) 17:03, 4 October 2010 (UTC)- Note that the entity &zwsp; is not supported by all browsers. Anomie⚔ 17:11, 4 October 2010 (UTC)
- Oops, better not mentioning then. -DePiep (talk) 18:48, 4 October 2010 (UTC)
- For other purposes there is also {{Null}} - which is ... null and {{Void}} .. er.. well you get the picture. I don't think there is a {{Null and void}} yet though. Rich Farmbrough, 19:05, 8 October 2010 (UTC).
- For other purposes there is also {{Null}} - which is ... null and {{Void}} .. er.. well you get the picture. I don't think there is a {{Null and void}} yet though. Rich Farmbrough, 19:05, 8 October 2010 (UTC).
- Oops, better not mentioning then. -DePiep (talk) 18:48, 4 October 2010 (UTC)
- Note that the entity &zwsp; is not supported by all browsers. Anomie⚔ 17:11, 4 October 2010 (UTC)
- ​ is available as {{zwsp}},
- Yes! That did it. The layout is now perfect. Thank you. Just granpa (talk) 04:51, 30 September 2010 (UTC)
External link icons proposal
This is something I've always thought and a brief discussion has pointed me here. Basically all PDFs give a symbol in links (http://www.example.pdf) instead of the diagonal arrow link thing (http://www.example.com/). I think xls should use {{XLSlink}}
type icons instead (and equivalently for Word Documents, .doc) as they a user may be much more reluctant to open these links, may be technically restricted or otherwise. Also, let's face it. The
|format=
parameter in citation templates isn't utilised by everyone. Rambo's Revenge (talk) 15:19, 30 September 2010 (UTC)
- I am fairly sure this has been discussed before. The PDF icon is defined at MediaWiki:Common.css. You can add the XLS icon with:
#content a[href$=".xls"].external,
#content a[href*=".xls?"].external,
#content a[href*=".xls#"].external,
#content a[href$=".XLS"].external,
#content a[href*=".XLS?"].external,
#content a[href*=".XLS#"].external,
#mw_content a[href$=".xls"].external,
#mw_content a[href*=".xls?"].external,
#mw_content a[href*=".xls#"].external,
#mw_content a[href$=".XLS"].external,
#mw_content a[href*=".XLS?"].external,
#mw_content a[href*=".XLS#"].external {
background: url("http://upload.wikimedia.org/wikipedia/commons/b/ba/Page_white_excel.png") center right no-repeat;
padding-right: 16px;
}
- You can test by adding the CSS to Special:MyPage/skin.css. This link should have the icon: http://www.example.org/example.xls ---— Gadget850 (Ed) talk 16:13, 30 September 2010 (UTC)
- I have amended the initial example from http://www.example.pdf/ to http://www.example.pdf because the presence of the trailing slash means that the URL doesn't end with ".pdf", which defeats the object of the example. --Redrose64 (talk) 16:16, 30 September 2010 (UTC)
- Think this is a great idea...many of this types of links are sometimes considered dangerous and i would love to know in advance if this is the type of link i am about to click on.16:49, 30 September 2010 (UTC)
- This seems helpful for DOCs (which are occasionally used as links or sources), but XLS is pretty rare and it would often be clear from context or description. Might as well have both though. Rd232 talk 16:56, 30 September 2010 (UTC)
- That is the perfect solution and works brilliantly Gadget850. I think it's worth making this the default for everyone. You mention this being discussed before. If an elegant solution such as yours was so feasible, I'd be suprised to see anyone oppose it. Is there any disadvantage of having such a function which is already so universally used for PDFs? Rambo's Revenge (talk) 17:40, 30 September 2010 (UTC)
- IIRC, it was opposed because "we should discourage linking to propriety formats" and there was no free icon. And you may have to write a detailed explanation why it's free. — Dispenser 19:27, 30 September 2010 (UTC)
- That is the perfect solution and works brilliantly Gadget850. I think it's worth making this the default for everyone. You mention this being discussed before. If an elegant solution such as yours was so feasible, I'd be suprised to see anyone oppose it. Is there any disadvantage of having such a function which is already so universally used for PDFs? Rambo's Revenge (talk) 17:40, 30 September 2010 (UTC)
- This seems helpful for DOCs (which are occasionally used as links or sources), but XLS is pretty rare and it would often be clear from context or description. Might as well have both though. Rd232 talk 16:56, 30 September 2010 (UTC)
- Think this is a great idea...many of this types of links are sometimes considered dangerous and i would love to know in advance if this is the type of link i am about to click on.16:49, 30 September 2010 (UTC)
- I have amended the initial example from http://www.example.pdf/ to http://www.example.pdf because the presence of the trailing slash means that the URL doesn't end with ".pdf", which defeats the object of the example. --Redrose64 (talk) 16:16, 30 September 2010 (UTC)
- I def support making icons for .doc, .xls, and possibly other non-web formats default for users. Protonk (talk) 17:44, 30 September 2010 (UTC)
- I've imitated what you did to give a code that works with .doc[1] Is it worth integrating docx & xlsx into those. Anything else common enough? Rambo's Revenge (talk) 19:36, 30 September 2010 (UTC)
- While I like that this makes it clear when a document is in one of these formats, it doesn't address the basic problem with using such proprietary-format sources. The vendor can render them unreadable practically overnight. We need to have them rendered into a more stable, more trustworthy, and more widely accessible format. PDF would suffice. If we need to do this then the links to such a stable format should be provided, if not in preference to the proprietary format, then at least in parallel to it. A WebCitation or link may be helpful in archiving the result, and List of PDF software suggests several converters. LeadSongDog come howl! 21:39, 30 September 2010 (UTC)
- I don't think such a solution is feasible or practical for most of these links. Protonk (talk) 22:51, 30 September 2010 (UTC)
- I second this and also don't think it's related to this proposal. Can we try and keep this specific proposal rolling as I don't wish for it to sink into obscurity. What would be the next steps? Rambo's Revenge (talk) 23:44, 30 September 2010 (UTC)
- What is the issue with proprietary document formats? PDF did not become open source until 2008, and was well used here before that. DOCX, XLSX and other Office Open XML formats are an open standard. ---— Gadget850 (Ed) talk 06:33, 1 October 2010 (UTC)
- The second "O" in OOXML is widely treated as a very bad joke in the open software community. While some parts of the format are disclosed, it doesn't come close to a fully open format. The fact that the vendor has a long history of embrace and extend tactics doesn't help. But the real issue is that their active content enables so many malware attacks that the formats are routinely blocked by firewalls. Still, you are correct that visibly marking them with their file formats is helpful to users independently of whether they consider the content to be safe to use. LeadSongDog come howl! 20:03, 1 October 2010 (UTC)
- LeadSong is absolutely right. Providing the icons is less an issue of promoting open formats and more an issue of providing a convenience to users who (esp. if they are on linux or a mobile device) might want to know that the link target will appear as a format which they can't or won't read. I would prefer that external links all land on simple HTML, but that isn't the way of the world. Protonk (talk) 21:46, 3 October 2010 (UTC)
- I don't think such a solution is feasible or practical for most of these links. Protonk (talk) 22:51, 30 September 2010 (UTC)
I have the CSS for XLS, DOC and RTF at User:Gadget850/ExternalLinkIcons.css. I will probably add DOCX, XLSX and others as I find icons. You can copy the CSS or import it by adding this to your Special:MyPage/skin.js:
importStylesheet('User:Gadget850/ExternalLinkIcons.css'); // Linkback: [[User:Gadget850/ExternalLinkIcons.css]]
---— Gadget850 (Ed) talk 16:40, 2 October 2010 (UTC)
- Excellent work Gadget850. Should we make this the default to all users? There seem to be no disadvantages to users so shall we boldly update MediaWiki:Common.css Rambo's Revenge (talk) 18:29, 3 October 2010 (UTC)
- I think we ought to solicit a little more input before we make a wide ranging change like that. Protonk (talk) 21:44, 3 October 2010 (UTC)
- Okay, where? (don't say RfC) Rambo's Revenge (talk) 23:53, 3 October 2010 (UTC)
- Naw, it could be here. Just make a little sub-section to this one saying what you want to be changed, post a notice on the CSS talk page you want to change and put something up on WP:CENT. If we have our heads up our asses, people will tell us. Otherwise we can make the change and be cool. Protonk (talk) 23:55, 3 October 2010 (UTC)
- Okay, where? (don't say RfC) Rambo's Revenge (talk) 23:53, 3 October 2010 (UTC)
- I think we ought to solicit a little more input before we make a wide ranging change like that. Protonk (talk) 21:44, 3 October 2010 (UTC)
Icons
After some fiddling, I find that the icons should be 16px wide and a max of 16px high, and must be bitmap. SVG will not work since the image is being called directly and not through ImageMagic that would convert it to PNG. Icons and are certainly open to more tweaking. ---— Gadget850 (Ed) talk 12:56, 4 October 2010 (UTC)
- The icons are going to have to be smaller than the current size. They clip when the font size is reduced, as with {{reflist}}. ---— Gadget850 (Ed) talk 15:15, 4 October 2010 (UTC)
- The PDF icons are 16x16, is it just a case of doing this? Rambo's Revenge (talk) 15:20, 4 October 2010 (UTC)
Proposal
I propose the contents of User:Gadget850/ExternalLinkIcons.css is added to MediaWiki:Common.css to supplement the existing PDF icon and indentify other formats (namely doc, docx, xls, xlsx, rtf, and txt) that are not html.
- OPPOSE - because using icons of brands in an article is advertising - and that's not allowed in wikipedia. Spamelgoog (talk) 22:24, 8 October 2010 (UTC)
- Support (as proposer) Rambo's Revenge (talk) 11:48, 4 October 2010 (UTC)
NB. This proposal has been listed at Template:Centralized discussion — Rambo's Revenge (talk) 12:01, 4 October 2010 (UTC)
- I don't think it's a big deal, so unless someone brings up a good reason why this should not be done, go for it. /ƒETCHCOMMS/ 12:46, 4 October 2010 (UTC)
- Support, small improvement but useful nonetheless. It adds consistency with the similar PDF icon. Dodoïste (talk) 12:48, 4 October 2010 (UTC)
- Weak oppose as unnecessary CSS bloat. Some browsers will download the additional icons whether they appear in a page or not. All browsers will have to deal with the longer CSS files. We should not be using these formats in the first place, at least when we can avoid it, and certainly not appear to condone them by having special icons for them. Many users do not have the necessary software to open them. Those who do often do not normally use this software. For a user who does not normally use Microsoft Office or Open Office, and consequently has disabled their quickstart function (which bogs down a computer at startup), loading an Office document takes ages.
- However, I would support one common warning icon for all proprietary formats that are prone to viruses and loading problems. Hans Adler 13:15, 4 October 2010 (UTC)
- (edit conflict) Support for easy identification of file formats and for consistency with the PDF icon. It might look small at first, but I'm sure the readers will benefit from this. ~NerdyScienceDude 13:16, 4 October 2010 (UTC)
- Oppose Links to these file are far and few between, and I feel it does not justify bloating common.css for these icons. PDF is by far the most linked document format (after HTML). This is a solution looking for a problem. — Edokter • Talk • 13:31, 4 October 2010 (UTC)
- (@Hans Adler) This isn't about whether we should use these formats (ideally we wouldn't use PDF either, and PDFs were also used long before they became open source) it is about warning users that they may not to click on a link because they might not be able to or take a long time to open. Often
|format
etc. is omitted.
- That's the reason for my alternative proposal of one icon for all of these crappy formats that clueless secretaries use when asked to publish something on the web. Hans Adler 14:15, 4 October 2010 (UTC)
- It might not be a good idea to employ ad-hominem attacks against those who chose to put content online using a file type that you dislike. Vanilla HTML is all very well, but sometimes there are quite good reasons for using proprietary formats. Especially when the intended audience was somebody other than wikipedians, as is often the case with content on the rest of the internet that we might want to cite. bobrayner (talk) 15:14, 4 October 2010 (UTC)
- That's the reason for my alternative proposal of one icon for all of these crappy formats that clueless secretaries use when asked to publish something on the web. Hans Adler 14:15, 4 October 2010 (UTC)
- (@Edokter) I disagree with the "solution looking for a problem". A wiki search shows the existance of many (>4000) with a few false positives of those including "format" and ".xls" and, seemingly, even more for .doc.
- Rambo's Revenge (talk) 13:55, 4 October 2010 (UTC)
- (@Hans Adler) This isn't about whether we should use these formats (ideally we wouldn't use PDF either, and PDFs were also used long before they became open source) it is about warning users that they may not to click on a link because they might not be able to or take a long time to open. Often
- Support I think it would be helpful. I do not consider the possibility of changing standards to be a big problem, since client software to read common proprietary formats (such as Word, Excel) is generally quite good at displaying documents written by a slightly earlier version of the software - the chances of any one cited item failing to be readable by a wikipedia user because of subsequent changes to the standard are, I think, much lower than the chances of the link failing due to vanilla linkrot (my brand-new laptop copes effortlessly with Excel spreadsheets that my team created in the 20th century). And even if it does fail, it's not Wikipedia's job to guarantee that every person must be able to read every cited document even to the extent of removing citations or substituting in less-than-ideal ones. The widespread use of universally-readable formats is a laudable goal, but in the real world there will sometimes be a better .doc or .xls out there, and most people browsing wikipedia will be able to read them; if you want to cater to those who cannot, maybe citing a second source would be better than removing the first choice. bobrayner (talk) 15:10, 4 October 2010 (UTC)
- Support. It's worth noting that those who oppose the use of these formats (for which there are valid reasons) should consider that ensuring that their use is identified might help reduce their usage. Making them more visible may encourage people to replace them with alternatives. Rd232 talk 15:21, 4 October 2010 (UTC)
- Support. About bloody time, too. If the best reference available is in Excel/Word format, the least we could do is to forewarn the readers of that. This has nothing to do with the "promotion" of proprietary formats.—Ëzhiki (Igels Hérissonovich Ïzhakoff-Amursky) • (yo?); October 4, 2010; 15:42 (UTC)
- Weak Support Not strictly necessary, but I don't see much of a downside. --Cybercobra (talk) 15:59, 4 October 2010 (UTC)
- Support Be bold and do it. Lugnuts (talk) 17:26, 4 October 2010 (UTC)
- Support, as should be clear from my comments above. I also think we need to steer clear of "XYZ format is crappy, why support it" discussions. We don't support document formats in external links. We just link to what the source material is. If the source material exists only in .DOC (common for corporate or government sources), our provision of an indication is not a support of that format. It is simply a convenience to the reader. I can't speak to the technical issue of CSS bloat, but I don't feel bloat is a universal argument against changing the status quo. Protonk (talk) 18:49, 4 October 2010 (UTC)
- Support I think it is a good idea. Armbrust Talk Contribs 19:30, 4 October 2010 (UTC)
- Support Not everyone has the capability to read DOC or XLS files. However, those icons should be linked to the article on the respective format. What if someone doesn't know what they mean? — Train2104 (talk • contribs • count) 02:19, 5 October 2010 (UTC)
- Weak Support per Cybercobra. If we can minimize the downsides... let's do it. Jclemens (talk) 04:09, 5 October 2010 (UTC)s
- Support will only help editors understand the abnormal links and it gives them the choice to click on certain types of links before hand.Moxy (talk) 05:59, 5 October 2010 (UTC)
- support seems really quite helpful, and the expressed downsides seem minor. Hobit (talk) 00:38, 6 October 2010 (UTC)
- Question Re: Hans Adler's mention of CSS bloat. How much additional data are we talking about that would need to be loaded, and how much of a potential loading speed decrease for users? Isn't there a new resource loader being written that is supposed to only load the things that each individual user needs to see? Also, are there any legal issues with making icons depicting these formats? Thanks!--Danaman5 (talk) 12:26, 6 October 2010 (UTC)
- Comment For the css bloat, I'd say it wouldn't be significantly more than the audio and video link icon definitions in the monobook stylesheet, which I would guestimate to be around 1-2 kilobytes. As for the images and the formats they represent I would probably suggest using some sort of generic icon set since there are other similar file formats (.odt, .ods, .odp, etc.) out there too. --Dlrohrer2003 19:19, 6 October 2010 (UTC)
- Support, a sensible proposal and sound idea, makes sense. -- Cirt (talk) 05:15, 8 October 2010 (UTC)
- Comment-What about OpenDocument, ppt, image, executable, zip extensions?Smallman12q (talk) 16:54, 9 October 2010 (UTC)
- I don't really buy the CSS bloat argument. In the coming months Wikimedia wikis will be using ResourceLoader, which will make the argument even more spurious (or perhaps dubious, you pick). I can agree that these icons are kind of silly and rare, so I won't outright support this proposal, but I certainly wouldn't oppose it based on the current arguments put forward. Think of this as a mild way of saying don't worry about performance: if you like the icons (aesthetically) and think they should be there, support. If not, oppose. Keep it simple, Sally. --MZMcBride (talk) 22:13, 9 October 2010 (UTC)
Polls are evil, especially when ignoring the technical niceties of how the WWW works
I de-rail this bandwagon by noting, as no-one else has done so far (unexpectedly for the technical Village Pump, and also unexpectedly since we do have an encyclopaedia around here that sort of explains this stuff), that there's no such thing as a file extension in a URL. Internet media type is determined by the Content-Type: header in the HTTP server response. One could publish an image/jpeg object with a URL ending in ".xls" and it wouldn't make it a spreadsheet file. The little pictures are not determined by actual content type. They have no guarantee of being right for readers. You really are better off ensuring that every non-HTML external hyperlink has a |format=[[Portable Document Format|PDF]] or a |format=[[Microsoft Word|DOC]] parameter in the citation template or uses one of the external link file type templates. The right thing cannot be deduced from just the URL. For universally correct results, it has to be specified alongside the external hyperlink, from knowledge of what the thing being hyperlinked-to is. Uncle G (talk) 02:11, 5 October 2010 (UTC)
- How often is a URL used on WP that looks like an XLS or a DOC not one of those? I've never come across that, so I'd guess it's rare. Rd232 talk 07:57, 5 October 2010 (UTC)
- On Wikipedia we'd have to measure. But in the world at large it's not rare. The tag ends of URLs not matching, when misinterpreted as filename extensions, the Content-Type: of the object are why Internet Explorer gained "MIME Handling Enforcement".
"Internet Explorer MIME Handling Enforcement". Microsoft TechNet. Microsoft.
Uncle G (talk) 17:14, 5 October 2010 (UTC)
- It doesn't matter how common it is Out There, it matters how common it is In Here. How often do articles use such URLs? Generally, the type of thing we're talking about will be a source; a source that doesn't even get the extension right is quite likely not a reliable source. So in general, I would call it rare, and calling attention to the cases where such extensions are used would likely make it rarer still. Rd232 talk 10:20, 6 October 2010 (UTC)
- That you still talk of "getting the extension right" indicates that you've missed the point. There is no extension to "get right". We really do have an encyclopaedia around here that sort of explains this stuff. Uncle G (talk) 01:52, 7 October 2010 (UTC)
- Er, what? Of course there's an extension to get right - extensions haven't been abolished (albeit superseded by mimetypes where possible), and it remains true that a document ending in .XLS ought to be an Excel spreadsheet. Mimetypes schmimetypes, that's what people expect, not least because it's usually the case. Rd232 talk 11:58, 7 October 2010 (UTC)
- That you still talk of "getting the extension right" indicates that you've missed the point. There is no extension to "get right". We really do have an encyclopaedia around here that sort of explains this stuff. Uncle G (talk) 01:52, 7 October 2010 (UTC)
- It doesn't matter how common it is Out There, it matters how common it is In Here. How often do articles use such URLs? Generally, the type of thing we're talking about will be a source; a source that doesn't even get the extension right is quite likely not a reliable source. So in general, I would call it rare, and calling attention to the cases where such extensions are used would likely make it rarer still. Rd232 talk 10:20, 6 October 2010 (UTC)
- On Wikipedia we'd have to measure. But in the world at large it's not rare. The tag ends of URLs not matching, when misinterpreted as filename extensions, the Content-Type: of the object are why Internet Explorer gained "MIME Handling Enforcement".
- I think if we were designing a security protocol, you would have exposed a serious hole. There is no requirement that the stated extension be declared in the given URL and no reason that a given URL be a final landing point (and not a redirect page). A more full featured solution would be to write a web crawler which follows external links, classifies the content and adds a tag. But we are merely adding a convenience icon. False negatives may produce confusion (resolved by the format tag), but false positives would be the only really disagreeable outcome for readers. And like RD 232 says, how many false positives have you discovered on the web? Protonk (talk) 15:51, 5 October 2010 (UTC)
- More than you'd like to think. Author of User:PDFbot and WP:Checklinks — Dispenser 16:40, 5 October 2010 (UTC)
- Thats a possibly indeterminate number. :) enough to make URLs a bad first approximation? Protonk (talk) 17:06, 5 October 2010 (UTC)
- ...or remove the existing PDF icon? Rambo's Revenge (talk) 17:08, 5 October 2010 (UTC)
- See above. This problem is why Internet Explorer has MIME Handling Enforcement. And this isn't discovering a security hole. This security hole actually existed, some several years ago. It is CVE-2001-0727 if memory serves correctly.
It's a fairly good idea to learn from experience in these things, and the experience is that "extensions" (which are not in fact extensions at all, and were never intended to be) don't match content types, that this is widespread enough that IE has a specific mechanism to avert the problems that this causes with IE's cache, and that really it isn't a good idea to make these assumptions, lest one repeat an error that was discovered the same year that Wikipedia was launched. Uncle G (talk) 17:14, 5 October 2010 (UTC)
- More than you'd like to think. Author of User:PDFbot and WP:Checklinks — Dispenser 16:40, 5 October 2010 (UTC)
Question: would it be (a) possible and (b) desirable for a bot to check for mismatches of extensions and MIME type? Rd232 talk 10:21, 6 October 2010 (UTC)
- Yes possible - but desirable? It certainly would be limited protection against abuse, since MIME types can change, and presumably we would only check once, and as Uncle G says the hole is fixed, and we would probably be finding mistake rather than attacks. Or if you mean to "icon" the urls better, then I'm not sure I see the advantage of "iconing" them, certainly with what look like Microsoft logos. As to the extent of .xls terminated urls, there seem to be about 8,730 articles (compared with some 307,000 for ".pdf") containing them which is small compared with the total number of articles, but large in absolute terms. However many of these are "references no one will ever check" e.g. a spreadsheet of populations for several hundred towns, ref'd in each of the several hundred articles. Rich Farmbrough, 04:19, 7 October 2010 (UTC).
Caveats
URLs for sites that us a query will not show an icon. For example, this New York Times article is a PDF:
http://query.nytimes.com/mem/archive-free/pdf?_r=2&res=9F00E6DE1E3CE633A25754C1A9659C946396D6CF
There are probably other ways in which an URL extension nay not be recognized or spoofed. ---— Gadget850 (Ed) talk 17:27, 9 October 2010 (UTC)
- a bot could recognise common ones (and/or check mimetypes) and add the format= parameter to cite templates. Rd232 talk 18:54, 9 October 2010 (UTC)
- PDFbot already does that, its just it's a pain running the bot. — Dispenser 20:03, 9 October 2010 (UTC)
Template help needed
![](https://upload.wikimedia.org/wikipedia/en/thumb/f/fb/Yes_check.svg/20px-Yes_check.svg.png)
I am passing a parameter with value "06" and it seems to be received with a value of "00" - no doubt it is either something incredibly subtle or something incredibly stupid - however I'm not likely to spot either it seems. Symptom: page displaying "December" ewhen it should be "June"
- Category:BLP articles lacking sources from June 2006 (see diagnostics on page)
- Passing template Template:Monthly clean up category
- Receiving template Template:Monthly clean up category/outer core
Just possibly this is WM limit on depth of transclusion? (You can see this limit in effect I believe at template {{Weather box}} where the sample box has some empty cells - the doc page however shows them filled in. Also template {{Delink}} which works beautifully in test but in anger breaks inexplicably.)
Rich Farmbrough, 16:58, 1 October 2010 (UTC).
- Bit of an aside here, but does {{delink}} not largely duplicate the function of {{unlink}}? PC78 (talk) 18:11, 1 October 2010 (UTC)
- Deja vu all over again! Yes you are right I wrote that one too, and had forgot it quite. In both cases the result was too good to throw away, but not good enough to use, much. Delink simply fails when used from another transclusion (but works on it's own doc page I think) - but is reasonably smart. Unlink is less smart but still suffers the limitations of our very-clever-but-hacked template string functions. Oh for proper string functions! It might however be useful in certain cases - thanks for reminding me of it. Rich Farmbrough, 23:15, 1 October 2010 (UTC).
- Incidentally you can see {{Delink}} fail here - had it worked it would have been a nice (part) solution to the random encoding of country names in infoboxes. Rich Farmbrough, 23:18, 1 October 2010 (UTC).
- Deja vu all over again! Yes you are right I wrote that one too, and had forgot it quite. In both cases the result was too good to throw away, but not good enough to use, much. Delink simply fails when used from another transclusion (but works on it's own doc page I think) - but is reasonably smart. Unlink is less smart but still suffers the limitations of our very-clever-but-hacked template string functions. Oh for proper string functions! It might however be useful in certain cases - thanks for reminding me of it. Rich Farmbrough, 23:15, 1 October 2010 (UTC).
Is it related to the typo in {{Monthly clean up category}} and also {{Monthly clean up category/outer core}}, where }}{{#if:{{diags|}}}|
should presumably be }}{{#if:{{{diags|}}}|
(with an extra brace before "diags")? — Richardguk (talk) 22:27, 1 October 2010 (UTC)
- Well spotted, (and thanks - that bug fixed) but no, the diags are just to try and resolve this problem - and not display the cruft that is on Category:BLP articles lacking sources from June 2006 on all the cat pages. Rich Farmbrough, 22:58, 1 October 2010 (UTC).
- Any one? I'm reluctant to troll individual talk pages 'cos I know people are busy, especially template gurus (I have thought we should have a WP:Template Clinic).
- I think hitting some template transclusion limit is probably the problem and while I don't know how to properly solve the problem, there is a simple workaround – specify the month number manually. Svick (talk) 11:55, 2 October 2010 (UTC)
- Yes.. that's where the template started. OK at least it's not something obvious I was overlooking, that's good to know. I'll have another think about it. Thanks for looking. Rich Farmbrough, 14:46, 2 October 2010 (UTC).
- The template is more complex than I have the time to work through at the moment, but I'll through out the possibility that it might be a data-typing problem. some magic words and parser functions throw fits if you feed them strings rather than numbers, and they may be interpreting '06' as a string rather than as the integer '6'. what happens if you feed the template a simple 6 without the leading 0? --Ludwigs2
- OP: one template sends a "06" (by parameter), the receiving template receives a "0", if I read it well. Could you provide a demo of the fail at a templates /testcases page? I had this problem elsewhere, and the explanation indeed was calling depth (so not data-type not param-defaults). I remember there is a limit about 20 (or is it 40) deep. As yet, there is no error message when this happens. User talk:Patrick can use & interpret the testing template {{edt}}.
- The template is more complex than I have the time to work through at the moment, but I'll through out the possibility that it might be a data-typing problem. some magic words and parser functions throw fits if you feed them strings rather than numbers, and they may be interpreting '06' as a string rather than as the integer '6'. what happens if you feed the template a simple 6 without the leading 0? --Ludwigs2
- Yes.. that's where the template started. OK at least it's not something obvious I was overlooking, that's good to know. I'll have another think about it. Thanks for looking. Rich Farmbrough, 14:46, 2 October 2010 (UTC).
- I think hitting some template transclusion limit is probably the problem and while I don't know how to properly solve the problem, there is a simple workaround – specify the month number manually. Svick (talk) 11:55, 2 October 2010 (UTC)
- Any one? I'm reluctant to troll individual talk pages 'cos I know people are busy, especially template gurus (I have thought we should have a WP:Template Clinic).
The error has apparently disappeared. Earlier Category:BLP articles lacking sources from June 2006 displayed "This category combines all BLP articles lacking sources from December 2006 ...". Now it correctly says June 2006. I don't know what fixed it. I made some tests during the discussion but didn't find the cause of the error. Previewing
{{Monthly clean up category/monthno|from June 2006}}
correctly displayed 06. Previewing
{{Monthly clean up category|monthno=06}}
correctly displayed "... articles lacking sources from June 2006 ...". But earlier, previewing
{{Monthly clean up category|monthno={{Monthly clean up category/monthno|from June 2006}}}}
incorrectly displayed "... articles lacking sources from December 2006 ...". PrimeHunter (talk) 12:17, 5 October 2010 (UTC)
- RE:
Well, it works, but that's all. No /doc, testcode is still in main template, unused subtemplate (not cheap) still in main code, original bug not found not changed not tested. This is what I found:
- RE:
- One day after RF (OP) posted here, and a few hours after RF's last post here, RF created
{{Monthly clean up category/monthname}}
. This template produces the correct month name now. Earlier subtemplate{{Monthly clean up category/monthno}}
(input param #1= {{PAGENAME}} i.e. to use the part "from August 2009") is still used to pass through monthno, but monthno is not used in subtemplates /outer core and{{Monthly clean up category/core}}
at all. So it's calculated, possibly wrong and unused. - RF left testing code in /core template, behind the switch {{#ifeq:{{{diags|}}}|yes| in
{{Monthly clean up category/outer core}}
. -DePiep (talk) 15:46, 5 October 2010 (UTC) - PrimeHunter, your tests produced right or undecided, because monthno is not used at all (not as input, not as subtemplate), and monthname is determined directly from the mw PAGENAME. I could not test whether /monthno is OK. Also, the /monthno should be tested in a stack (nested) too.
- If you want to sandbox, create a page that has the pattern "/... from March 2008".
- All together, this is sloppy programming in the template. -DePiep (talk) 15:46, 5 October 2010 (UTC)
- I'm going to re-read, but quickly, yes inserting "diags=yes" will show what is going on. The fix is actually an improvement, since the very old version required one to specify as input both the month name and the month number, version 2 (as it were) worked out the name from number, version 3 got the number from the PAGENAME - it was a bit crazy to convert the PAGENAME -> number -> name (although obviously not a speed/cpu cycles issue since there is no loop and deriving them separately calls more string handling functions - which are a clever, but expensive, hack) - so version 4 derives them separately. Rich Farmbrough, 16:23, 5 October 2010 (UTC).
- I'm going to re-read, but quickly, yes inserting "diags=yes" will show what is going on. The fix is actually an improvement, since the very old version required one to specify as input both the month name and the month number, version 2 (as it were) worked out the name from number, version 3 got the number from the PAGENAME - it was a bit crazy to convert the PAGENAME -> number -> name (although obviously not a speed/cpu cycles issue since there is no loop and deriving them separately calls more string handling functions - which are a clever, but expensive, hack) - so version 4 derives them separately. Rich Farmbrough, 16:23, 5 October 2010 (UTC).
I think Ludwigs is probably correct in his diagnosis. The value of monthno (if not passed in as a parameter, as none of the current uses that I am aware of do) is as DaPiep says, correctly calculated and passed down the stack => outer core => core (at one point I thought I was going to need "inner core"!), although /core does use it - successfully - as a category sort key (which was it's entire function originally) see for example Category:BLP_articles_lacking_sources. It also uses it to calculate the age of the template and nominate it for speedy deletion if it is old and empty - again successfully. Thanks for all the input, it is a great help (especially reassuring me that I'm not just "template blind" and missing a piece of punctuation!) if I do not hear otherwise I will remove the diagnostic parameter in the relatively near future. Rich Farmbrough, 16:43, 5 October 2010 (UTC).
- You are right, monthno is well used there. Only the string-datatype now unresolved & unused. -DePiep (talk) 17:31, 6 October 2010 (UTC)
Clicking on signature icon gives erratic results
Sometimes when I click on it it works as normal, other times it adds my sig to the section heading or the search field (the one above the editing field), or elsewhere. I'm using the current version of Chrome. Thanks. Dougweller (talk) 17:40, 1 October 2010 (UTC)
- You know that it places it where the text cursor is. So if you select the search box, and then press the signature icon, it will place a signature in the search box. Is this what the problem is? - Kingpin13 (talk) 17:04, 5 October 2010 (UTC)
Bunching Problem
I tried to fix the section edit link, bunching problem in the United States Congress, but I guess I don't understand the Template:Fix bunching well enough. Can someone please help?--Bbb23 (talk) 18:43, 2 October 2010 (UTC)
Done. mostly done. there's still a little bunching farther down in the article, but that's because the 'congress' section has too many images. move them about a bit. place some on the left, delete some; you can work that out. --Ludwigs2 19:45, 2 October 2010 (UTC)
- Thanks! I did the same thing with the template as you did, but it didn't work. You did one more thing, though, you moved up the image in the Overview section, which fixed it. As for your comment about too many images, that'll be harder because the article was recently revamped, and the revamping editor LOVES images. But that's not your problem.--Bbb23 (talk) 15:20, 3 October 2010 (UTC)
- well, that's kind of everyone's problem.
images are supposed to enhance an article, not get in the way of the reading experience. I may go do it myself - either by pruning or by shifting the images to a gallery section at the bottom. --Ludwigs2 16:20, 3 October 2010 (UTC)
- well, that's kind of everyone's problem.
- I've fixed the bunching problem by playing with the images (shortening captions, moving images, deleting) - with the blessing of the editor who added so many images. I'm not sure whether the article and the images are in the best state at the moment, but I'll try to look at the substantive issue later. The technical problem, though, is gone.--Bbb23 (talk) 18:01, 3 October 2010 (UTC)
- Thanks people for fixing. Yes I do love images (guilty as charged) and I sometimes go overboards (definitely one of my problems) with pictures. So thanks for fixing/deleting/ improving them. That said, I'd LOVE to have a picture of a beautiful celebrity like Angelina Jolie or Katie Holmes or Dana Delany or George Clooney testifying before Congress -- it would be SUCH an improvement over Archibald Cox -- I hope that was one of the pictures you people nixxed. I mean, Cox is important, but hey this is America, and we love pretty people, right?--Tomwsulcer (talk) 00:53, 4 October 2010 (UTC)
- Once again plugging the solution to the entire problem: User:Drilnoth/lefteditlinks.js/doc. OrangeDog (τ • ε) 09:56, 6 October 2010 (UTC)
Help with table - alphabetical order of St
Hi, can anyone help with the table at User:Skinsmoke/Sandbox/Civil parishes/Kernow please? The places starting with "St" should of course sort at "Saint", how do we make the table do this? DuncanHill (talk) 09:37, 5 October 2010 (UTC)
- I made an edit to the page with <span style="display:none">Saint Agnes</span> as an example of how to use a hidden sort key. Cheers. HausTalk 09:55, 5 October 2010 (UTC)
- Thank you! DuncanHill (talk) 10:29, 5 October 2010 (UTC)
- See Help:Sorting. ---— Gadget850 (Ed) talk 15:04, 5 October 2010 (UTC)
Part of Refdesk vanishing
![](https://upload.wikimedia.org/wikipedia/en/thumb/f/fb/Yes_check.svg/20px-Yes_check.svg.png)
This diff [2] makes everything below it disappear from the visible page. Can anyone fix it please? DuncanHill (talk) 14:31, 5 October 2010 (UTC)
Done, it was missing a closing tag.[3] Rambo's Revenge (talk) 14:42, 5 October 2010 (UTC)
- Thanks! DuncanHill (talk) 14:57, 5 October 2010 (UTC)
redirects in search engines
Two questions: 1) Why don't redirect pages and titles appear in search engine results?. Is it deliberate? if so, are we (wikipedia) doing it or the search engines are programmed to ignore redirects?
2)I read some where that article talk pages in en.wiki are not indexed by google. But i found that talk pages in some other wikis (like Tamil Wiki) are indexed. Who decides this? google or us (i mean, can we turn this on or off from wikipedia's end?) --Sodabottle (talk) 17:54, 5 October 2010 (UTC)
- 1) Redirects are served with HTTP 301 status code ("Moved Permanently"). I can imagine that this makes search engines ignore them.
- 2) Us, via robots.txt (whose local part is editable as MediaWiki:robots.txt). I don't see any blanket prohibition of talk pages there, though.—Emil J. 18:11, 5 October 2010 (UTC)
- Thanks Emil. We have a situation in Tamil wikipedia, where we want the redirect pages to be indexed. Tamil script has two slightly different ways of writing some words and currently we use one method for standardisation. Article titles written in the other method are used as redirects. We want the redirect pages to be indexed for enabling those who use the alternate writing method to search. Is there a way to work around the HTTP:301 feature? or get wikimedia to turn it off for ta.wiki alone?.--Sodabottle (talk) 18:27, 5 October 2010 (UTC)
- Actually, no, redirects are served with ordinary 200 OK (at least to my browser, I doubt it serves something else to Googlebot). But it seems Google understands that. For example Ablepsia redirects to Blindness and searching for ablepsia on en.wikipedia.org gives link to http://en.wikipedia.org/wiki/Blindness, although on the last position. Also, why should searching for ablepsia return that page (from the POV of Google)? It doesn't even mention that term. I don't know what could be done to solve that problem.
- Also, talk pages are indexed, Svick (talk) 23:04, 5 October 2010 (UTC)
- Clicking "Cached" on your Google result to en.wikipedia.org/wiki/Blindness correctly gives: "These terms only appear in links pointing to this page: ablepsia".
- http://www.seoconsultants.com/tools/headers also reports "200 OK" and not "301 Moved Permanently" on http://en.wikipedia.org/wiki/Ablepsia. The page is not noindexed so I don't know how Google actually detects Wikipedia redirects to avoid indexing them. Maybe it's a part of their normal duplicate content removal and they choose the version where title and url matches or where there are many incoming links, or maybe they coded specifically for a Wikipedia hint like "(Redirected from ...)", but I'm just speculating. http://www.google.com/support/webmasters/bin/answer.py?hl=en&answer=66359 asks sites to use 301 redirects for duplicate content. Regarding the Tamil Wikipedia, if we could get Google to index both url's (that may not be possible) then a lot of searchers would get two Wikipedia hits to the same page unless perhaps if they search exactly on the redirecting term and that term doesn't occur on the real article. I don't know Tamil but wouldn't it be possible to just write the title in both ways on the real article? Maybe the way that isn't in the page title will be lower in Google's search results but we should be writing an encyclopedia, not making search engine optimization. PrimeHunter (talk) 23:43, 5 October 2010 (UTC)
- In addition, Template:NOINDEX can be used to remove individual pages from some search engines. See Category:Noindexed pages. -- Ϫ 22:45, 5 October 2010 (UTC)
- Thanks for the replies. I will try raising this in the google SEO forums.
- >>don't know Tamil but wouldn't it be possible to just write the title in both ways on the real article? Maybe the way that isn't in the page title will be lower in Google's search results but we should be writing an encyclopedia, not making search engine optimization
- If only it were that simple :-). There is language politics involved and the purists dominate ta.wiki and the consensus there is not to use the second method (it uses a few characters loaned from Sanskrit to represent sounds not in Tamil) at all. Using both versions in the same article is discouraged and only allowed in redirects. But a overwhelming majority of real world Tamil speakers use the second method and their searches don't end up in wikipedia (despite have good content). --Sodabottle (talk) 08:27, 6 October 2010 (UTC)
- Easy peasy. Transclude the pages. Rich Farmbrough, 04:43, 7 October 2010 (UTC).
- Easy peasy. Transclude the pages. Rich Farmbrough, 04:43, 7 October 2010 (UTC).
- I am under the impression that Google are smart enough to have special code written for Wikipedia. And to do other smart things. Incidentally, I rather think that putting spaces (%20s?) in the url instead of underscores does return a 301. Rich Farmbrough, 04:43, 7 October 2010 (UTC).
- Thanks Rich, will try transclusion--Sodabottle (talk) 05:14, 7 October 2010 (UTC)
Total size of pictures actually used in articles?
(A) How many GByte do the images actually used in english wikipedia articles use in total?, (B) what would the size of the thumbnail used be? I tried to find this info in the statistics- and download section(s) but only found just about everything about number of articles, requests, size etc.. Electron9 (talk) 00:32, 7 October 2010 (UTC)
![](https://upload.wikimedia.org/wikipedia/en/thumb/f/fd/Portal-puzzle.svg/220px-Portal-puzzle.svg.png)
- There a re a lot of confounding factors here. While there is a default size for thumbnails in px, compression means that the number of bytes can vary. So you would have trouble getting an exact figure. to confuse matters further pictures are used on multiple pages, so you would have to decide whether to double count these, and at different sizes, and in different formats. Also images are used in templates - Image:Portal-puzzle.svg is transcluded on at least 94,765 pages - plus this one. One large picture, 2000x2000, is equivalent (in pixels) to 100 chunky thumbnails of 200x200, or 10,000 icons of 20x20. Moreover if you have your preferences set that way all MathML is downloaded as images too. Rich Farmbrough, 04:37, 7 October 2010 (UTC).
- Case B) I certainly consider a thumbnail (or ..|200px) of the same size used in different articles as one image. Same image in two sizes would in that case count as the image size of both added together. MathML could be as-is. Case A) One source image would be counted as one. Indifferent to the number of articles using it in various sizes. As these are a plain recomputation of the source. Ie if you have "aa.html bb.html cc.html dd.html" and "ee.jpg". The image would only count once despite being used from all pages. The joy of hyperlinking.. Electron9 (talk) 11:00, 7 October 2010 (UTC)
- I don't think this is tracked, due to the way the images are stored and used by the various projects. Anecdotally, I believe that all of the full-sized images used on wikipedia would require less than 1TB of space, and the thumbnails would take few hundred GB. -Steve Sanbeg (talk)
IRC Recent changes seems to be down
Just a note, my huggle isn't connecting to IRC and is using slow api queries, making vandalism reversion a bit slow. am I the only one having this problem? Access Denied [FATAL ERROR] 02:23, 7 October 2010 (UTC)
- I don't know, how do you tell? Mine seemed jumpy and crashy a few days ago.Rich Farmbrough, 04:22, 7 October 2010 (UTC).
- Igloo works fine right now, but Igloo runs through an external feed AFAIK. Access Denied [FATAL ERROR] 04:33, 7 October 2010 (UTC)
- Igloo is alright, but Huggle only reached API list. Sir Stupidity (talk) 10:27, 8 October 2010 (UTC)
- Igloo works fine right now, but Igloo runs through an external feed AFAIK. Access Denied [FATAL ERROR] 04:33, 7 October 2010 (UTC)
- There is similar conversation at Wikipedia talk:Huggle that gives more information--Wolfnix • Talk • 16:51, 8 October 2010 (UTC)
(edit conflict)WT:Huggle#Unable to stay connected to IRC gives a bit more explanation into this matter. Killiondude (talk) 16:48, 8 October 2010 (UTC)
SmackBot
Please be aware of Wikipedia:Bots/Requests for approval/SmackBot 35 Rich Farmbrough, 04:21, 7 October 2010 (UTC).
Which pages in my contributions list are also on my watchlist
I'd like to know, when browsing through my contributions list, which of those pages I currently have watchlisted, is there any easy way to do this? -- Ϫ 10:30, 7 October 2010 (UTC)
- I don't think there's a straightforward way, especially if you have a very large watchlist. It's probably possible to write a user script to do this, though. --ais523 10:47, 7 October 2010 (UTC)
- You could use AWB to generate a list of all the pages you contributed to and compare it to your watchlist. –xenotalk 15:17, 7 October 2010 (UTC)
More than one watchlist?
Regarding watchlist.. is it possible to have two of them ..?, one for important. And one for less important ones. Without resorting to two users or script-fu? Electron9 (talk) 11:03, 7 October 2010 (UTC)
- Not exactly. However, you can create a page with a list of links (a user subpage is a good place) and then use the "related changes" link from it to see changes from articles on the list, which works almost exactly the same way as a watchlist; the major difference is that other people can see what the list is. --ais523 11:17, 7 October 2010 (UTC)
Edit count discrepancy
I noticed that my total edit count clicked up to 20,000 last night, according toX!'s edit counter and SoxBot's admin stats (15,554 edits plus 4,446 deleted edits). A rough count of my contributions and deleted contributions indicates that these are correct. However, the number of edits according to my preferences is only 18,427. Can anyone explain the discrepancy? — Tivedshambo (t/c) 11:45, 7 October 2010 (UTC)
- See WP:EC. For example, page moves are counted differently by different counters.—Emil J. 11:59, 7 October 2010 (UTC)
- That would explain it - thanks. — Tivedshambo (t/c) 12:04, 7 October 2010 (UTC)
Vandalism I can't find?
Georgetown University (edit | talk | history | protect | delete | links | watch | logs | views)
Still showing for me (blatantly, and it's ugly), and I can't figure out why. It's not in the transcluded template. Is this a fluke in my browser? If not, can somebody fix it? I don't want to publicize this at ANI or a more visible forum, and you guys seemed likely to know. --Moonriddengirl (talk) 16:10, 7 October 2010 (UTC)
- Purge the page, that should fix it. — Edokter • Talk • 16:13, 7 October 2010 (UTC)
- I tried that. It's still there. :/ Even more oddly, it should never have been there for me to begin with, since the first time I visited the page was just after 16:00. The vandalism was reverted hours ago. --Moonriddengirl (talk) 16:14, 7 October 2010 (UTC)
- It looks to be gone from my perspective, so if this is still occurring it must be something on your end. have you tried clearing the cache on your browser? --Ludwigs2 17:50, 7 October 2010 (UTC)
- Yes, multiple times. It did finally disappear a bit over an hour ago. Again, the really weird thing here is that I had never even visited the article until hours after the vandalism was removed, so theoretically it should not have been in my browser to begin with. I mentioned it to another edit who said, "Could be an issue with the proxy in your network..." whatever that means. :) My main concern was that if it was showing on my computer due to some strange glitch, it would probably show on others. --Moonriddengirl (talk) 19:22, 7 October 2010 (UTC)
- I believe a network proxy cache serves the same purpose as an individual's web cache, but for all the users on the network. So, if you're looking at the site through a network proxy, and another user looked at the same site while the offensive material was there, the network proxy server might show you the cached (offensive version). You can read a little about this here.--Bbb23 (talk) 19:31, 7 October 2010 (UTC)
- Interesting! Thanks. :) --Moonriddengirl (talk) 19:00, 8 October 2010 (UTC)
- I believe a network proxy cache serves the same purpose as an individual's web cache, but for all the users on the network. So, if you're looking at the site through a network proxy, and another user looked at the same site while the offensive material was there, the network proxy server might show you the cached (offensive version). You can read a little about this here.--Bbb23 (talk) 19:31, 7 October 2010 (UTC)
- Yes, multiple times. It did finally disappear a bit over an hour ago. Again, the really weird thing here is that I had never even visited the article until hours after the vandalism was removed, so theoretically it should not have been in my browser to begin with. I mentioned it to another edit who said, "Could be an issue with the proxy in your network..." whatever that means. :) My main concern was that if it was showing on my computer due to some strange glitch, it would probably show on others. --Moonriddengirl (talk) 19:22, 7 October 2010 (UTC)
- It looks to be gone from my perspective, so if this is still occurring it must be something on your end. have you tried clearing the cache on your browser? --Ludwigs2 17:50, 7 October 2010 (UTC)
- I tried that. It's still there. :/ Even more oddly, it should never have been there for me to begin with, since the first time I visited the page was just after 16:00. The vandalism was reverted hours ago. --Moonriddengirl (talk) 16:14, 7 October 2010 (UTC)
URLs in diffs not clickable within a cite template
![](https://upload.wikimedia.org/wikipedia/en/thumb/f/fb/Yes_check.svg/20px-Yes_check.svg.png)
Say you're looking at a diff of a deletion edit where someone has left the edit summary, "Not supported by source cited". So you want to look at the url for the source to verify that. If the ref was given as just a bare url between two "ref" tags then all you have to do is click on the URL to open it.
But if the URL is embedded within a "cite" template then the URL won't be "clickable". If you try to click on it you'll get directed to Template:cite_news or Template:cite_web or whatever. And if you try to "highlight" the URL with your mouse so you can copy-paste it to your browser's "destination address" field, you'll find that you can't easily do so: The highlighting behavior is all wonky, and you'll invariably end up grabbing more characters than you want. You'll have to paste the whole mess into the "destination address" field, and then trim off the extra characters there before trying to access the web site. This makes verifying sources from diffs much harder than it needs to be.
I'm guessing it should be fairly easy to correct this in the MediaWiki software, to just "turn off" the feature whereby hovering over a cite template in diffs targets Template:cite_web or whatever. Would other editors also be in favor of such a feature change request? – OhioStandard (talk) 20:39, 7 October 2010 (UTC)
- The on-hover or on-click behavior in the two-column diff display sounds like a browser or third-party-extension feature. But as to the issue of the amount of text picked in the cite-templates, does it recognize whitespace as the limits of the link? I usually write template parameters as |url= http://example.com |title= example link |etc (among other reasons) to make it easy to keep "the URL" as a separate word easily selectable. Could you give more specific examples (links to diffs) where it does vs doesn't work, and details about your OS, browser, other options, etc.? DMacks (talk) 20:55, 7 October 2010 (UTC)
- Thanks for your reply. Here's a diff where I encounter the problematic behavior. The first ref to The Guardian occurs without a "cite" template being used, and I can click on the target URL to reach the corresponding web page. The second ref to The Guardian occurs within a "cite news" template, and clicking on the url there just takes me to Template:cite_news. As to the second part of your question, in the example the URL and other parameters used inside the cite template were set off with spaces, I find, and this doesn't seem to make any difference for me. In other words, wherever I "hover" over the cite template I always have the "hand" icon for my cursor, never the "vertical line" ("select text"?) cursor.
- The only way I can grab part of the contents (e.g. a url) embedded within the cite template is to start with the cursor placed outside the darker green "diff column" on the right, press the mouse button, and then move the cursor into the darker green diff column. In the example I've given here the resulting behavior is predictable and unsurprising: As I move the cursor horizontally across the darker green diff column, text is progressively highlighted in the direction corresponding to the direction of travel of my mouse/cursor. In other instances, though, the behavior as to what gets highlighted as I do so is not predictable; sometimes a dozen or more lines will be highlighted. I haven't been able to figure out what differentiates the circumstances in which the "predictable" versus the "wonky" behavior in this regard occurs, btw.
- Sorry to have neglected to include O/S, browser details, and such. I'm using Firefox 3.6.10 on Ubuntu Linux 8.04 ("Hardy Heron") with all updates installed. I have the usual Ubuntu Firefox Modifications Pack version 0.9rc2 installed; I believe this is the default for Ubuntu users. I also have Adblock Plus 1.2.1 installed, and two other add-ons, Ghostery 2.2.1, and "TACO with Abine" ver 3.10. These latter two are privacy and cookie opt-out browser add ons.
- So does this appear to be browser or configuration specific? That is, do you find you get a different on-click behavior re a url embedded within a cite template? – OhioStandard (talk) 21:55, 7 October 2010 (UTC)
- Yes, I am: good call! Thank you. It didn't even occur to me that this might be due to one of the "gadgets" I have installed, but that was it. Turns out wikEd is responsible both for making a url (that's not embedded within a cite template) "clickable" in the first place, and also for the "wonky" behavior I described above in trying to access urls that are so embedded. Weird. So it's not a feature/bug in MediaWiki software at all, but has to do with wikEd. I'll miss that "undo" feature in wikEd, especially, but I've uninstalled it for now, and will file a bug-report/feature-request with wikEd's developer. Thank you very much, Fences&Windows, and thanks to DMacks, as well. I really appreciate your help on this. Best, – OhioStandard (talk) 01:00, 8 October 2010 (UTC)
Place or tool to get WP stats
Is there a place or tool where one can get particular statistics? If I wanted to know (as I do, in fact) what percentage of articles created between August 1, 2009, and July 31, 2010, by newcomer editors who had less than 20 edits at the time they created the article were still in existence 60 days after the day they created the article, is there a place or a way to get that information? Best regards, TRANSPORTERMAN (TALK) 21:18, 8 October 2010 (UTC)
- Have you checked Wikipedia:Statistics? Many links there that I'm sure should help. -- Ϫ 23:56, 8 October 2010 (UTC)
- No, didn't know it existed. Thanks for the pointer. Best regards, TRANSPORTERMAN (TALK) 00:03, 9 October 2010 (UTC)
Adding vertical bars to version comparisons
How difficult would it be to add vertical bars next to the line(s) that have been changed in a comparison of different versions of an article? I sometimes go crazy trying to find a single character that has changed. Sometimes, as here, I can't find the change at all. Some publications that have recurring updates (like computer manuals and legal treatises) use the convention of inserting vertical bars next to the changes. Would it be hard for Wikipedia to implement such a system? Wine may improve with age, but my eyes don't.--Bbb23 (talk) 00:27, 9 October 2010 (UTC)
- Each paragraph is a single line. Changed paragraphs are highlighted with colour. Changed text is highlighted in red. Unfortunately, changes in whitespace aren't highlighted properly, even if, like me, you have changed your user CSS to highlight changes with a red background instead of with red text. It frustrates me to no end that whitespace changes aren't highlighted properly, because they're the hardest to find—but the diff algorithms are complex enough that improving them to catch this sort of thing is no trivial change. Actually, I just had a little brainwave on how I might solve the problem for myself using JavaScript… I'll go play around a bit and see if I can solve the problem as an end-user. {{Nihiltres|talk|edits|⚡}} 02:13, 9 October 2010 (UTC)
- I think whitespace changes get removed from diffs by Wikipedia's use of HTML Tidy when rendering pages, in which case it would not be possible to identify the relevant characters without a change to MediaWiki. — Richardguk (talk) 03:33, 9 October 2010 (UTC)
- Ah, that would explain much. Then HTML Tidy would have to be configured to ignore those particular span elements … I wonder if that's feasible, or for that matter desirable—for example, would it cause us to output invalid HTML? I'm supposing HTML Tidy is removing it for a good reason… {{Nihiltres|talk|edits|⚡}} 04:41, 9 October 2010 (UTC)
- HTML Tidy is applied by the MediaWiki software on enwiki to fix or make consistent various errors in wikitext. Tidy is an optional post-wikiparsing configuration and it would be technically simple to switch it off altogether, but that would cause a large number of pages to have unpredictable rendering problems. In principle, MediaWiki could be amended to apply HTML Tidy only to the preview part of a page (excluding the diff). Alternatively, it might be possible for significant spaces in diffs to be replaced by non-breaking spaces or other characters that would not be moved or removed by Tidy. If desirable, that change would need requesting at Wikimedia's Bugzilla. — Richardguk (talk) 14:14, 9 October 2010 (UTC)
- Ah, that would explain much. Then HTML Tidy would have to be configured to ignore those particular span elements … I wonder if that's feasible, or for that matter desirable—for example, would it cause us to output invalid HTML? I'm supposing HTML Tidy is removing it for a good reason… {{Nihiltres|talk|edits|⚡}} 04:41, 9 October 2010 (UTC)
- I use the default vector skin. What would I do to show changed text as a red background?--Bbb23 (talk) 13:38, 9 October 2010 (UTC)
- Try adding the following line to "/vector.css". (At your user page which would be "User:Bbb23/vector.css".)
- I think whitespace changes get removed from diffs by Wikipedia's use of HTML Tidy when rendering pages, in which case it would not be possible to identify the relevant characters without a change to MediaWiki. — Richardguk (talk) 03:33, 9 October 2010 (UTC)
.diff-addedline * .diffchange { background-color: #FF7458 !important; color: black !important; }
Citing google cache copy of Bloomberg
I wanted to cite this bloomberg article, but it gives a 404 now. Google cache has a copy though. How can I preserve the google cache copy? (Webcitation doesn't allow citations of google cache).Smallman12q (talk) 16:34, 9 October 2010 (UTC)
Add -webkit-text-size-adjust: none; to MediaWiki:Handheld.css
Proposal here: http://en.wikipedia.org/w/index.php?title=MediaWiki_talk:Common.css&oldid=389737066#Add_-webkit-text-size-adjust:_none.3B_to_MediaWiki:Handheld.css --MZMcBride (talk) 17:44, 9 October 2010 (UTC)
What is the CSS class for external links?
I'm trying to figure out the css for external links, but none of the ones I use are working? Is there any place that lists all CSS classes used by Wikipedia? Access Denied [FATAL ERROR] 18:19, 9 October 2010 (UTC)
- I changed your monobook.css to use
#content a.external, #bodyContent a.external
, which should just about always work. Go bypass your cache and see if it's fixed. {{Nihiltres|talk|edits|⚡}} 18:28, 9 October 2010 (UTC)
- As for your second question: Wikipedia:Catalogue of CSS classes. — Edokter • Talk • 18:35, 9 October 2010 (UTC)