Template talk:Citation/core/Archive 17

DOI --> doi
I don't know why this has changed, but as a URI, the correct format is doi:10.1234/567890, not DOI:10.1234/567890. Please change it back to the correct behavior. Headbomb {talk / contribs / physics / books} 20:54, 29 July 2012 (UTC)

✅ —David Eppstein (talk) 16:05, 30 July 2012 (UTC)
 * support. It was requested and done above, at . It's annoying and incorrect, so I support switching it back. Br&#39;er Rabbit (talk) 21:11, 29 July 2012 (UTC)
 * support. There wasn't any discussion on the previous change. ---— Gadget850 (Ed)  talk 15:11, 30 July 2012 (UTC)

Adding a parameter
I proposed this here first, but I was told I should come here.

I propose to add a parameter to Template:Cite web that accepts two values, say yes and no to indicate, whether subscription is required for a source or not. The default would be subscription=no, which would not change the appearance of the template. If set to subscription=yes, will be transcluded before the source. Currently I have to add the template separately inside the tags. I think it should be inside the template. -- Toshio Yamaguchi (tlk−ctb) 20:46, 23 August 2012 (UTC)
 * Something like this would be useful for more than just the cite web templates; I have been abusing subscription required in citation, for instance, and it would be better to be able to handle this properly. But why do you think it should go before the source. Wouldn't after be a better placement? —David Eppstein (talk) 21:41, 23 August 2012 (UTC)
 * Actually I have no strong opinion on whether it is placed before or after. I see now that the documentation at Template:Subscription required says the template should be added directly after the link or citation template and I am fine with that. -- Toshio Yamaguchi (tlk−ctb) 21:51, 23 August 2012 (UTC)
 * Can I just point out that this will increase processing time, and give more fuel to those who actively seek to destroy all / templates on the grounds of slow processing. They wish to replace them either with plain text, or failing that, with so-called "fast" versions with much reduced capabilities, and will seize on any excuse to further their ends. -- Red rose64 (talk) 16:23, 24 August 2012 (UTC)
 * So tell them to help bring Scribunto here. I've just added   and   parameters to Module:citation on the test2 wiki.  You can see the results at .  You can also see that far from eliminating parameters for "speed", I'm aiming to retain the existing parameter set.  (I've even added one:  .  Commonwealth English speaking editors know where to send their thank-you gifts.  &#9786;)  Uncle G (talk) 15:27, 27 August 2012 (UTC)

treatment of origyear parameter
I'd like to re-raise a problem from three or four years ago which was never resolved satisfactorily.

Cite book has always had a parameter "origyear" where one can specify the original year of publication of a book, as distinct from the publication date of the later edition or reprint which is actually being used by the editor. The origyear parameter appears in square brackets instead of round brackets. Until about four years ago if used it appeared unconditionally, whether or not "year" was also given, but in the restructuring in 2008 this was changed so that if year was absent, origyear was ignored. The documentation was also changed correspondingly in January 2009.

I raised the matter at the time, some of which you can see at Template talk:Citation/core/Archive 3, but the answer I got was that you shouldn't do that, whereas that didn't solve the backward compatibility point that people (including me) had in good faith done that and it was hardly possible to find all those cases and change them, unless a bot searched the entire database to do so. At least some of these cases remain to this day, where origyear is specified but no date at all is displayed.

I asked advice of User:Gadget850 and he clarified that
 * origyear in cite book is passed by YearNote to citation/core.

This probably only affects a few articles, but it would be nice if some proper solution could be found. Thanks. SamuelTheGhost (talk) 10:55, 6 September 2012 (UTC)


 * YearNote is dependent upon Date. Options:
 * Leave as is. Currently documented as a child parameter.
 * Show, regardless of Date.
 * Display an error if YearNote but not Date.
 * And/or place in a tracking category if YearNote but not Date.
 * ---— Gadget850 (Ed)  talk 11:13, 6 September 2012 (UTC)


 * I'd be content with any of those options except the first. SamuelTheGhost (talk) 22:42, 7 September 2012 (UTC)

In view of the lack of further comment, could I request that action is taken? I'd suggest the use of the tracking category, certainly, since this would provide information about the size of the problem, and if it proved to be small would allow a motivated editor to remove it altogether. I'd also prefer the "show regardless" option in the mean time, as being harmless and possibly useful. SamuelTheGhost (talk) 11:47, 21 September 2012 (UTC)
 * Let's set up the tracking category then see how widespread the issue is. Rjwilmsi  11:53, 21 September 2012 (UTC)

This should do it: {{pre2| {{#if:{{{YearNote|}}}|{{#if:{{{Date|}}}||}} }} ---— Gadget850 (Ed)  talk 13:02, 21 September 2012 (UTC)


 * This is something that should be made to work properly. In bibliography sections of author or researcher biography articles, the date of original publication is more significant than the date of the issue seen by a WP editor. Failing silently (Leave as is) should definitely be a non-option. LeadSongDog come howl!  14:13, 21 September 2012 (UTC)
 * Yes. The best thing to do right now is to get the tracking category going. Then if necessary we can debate other options. SamuelTheGhost (talk) 14:35, 21 September 2012 (UTC)


 * Still hoping that this will be implemented soon ... SamuelTheGhost (talk) 22:14, 5 November 2012 (UTC) SamuelTheGhost (talk) 21:27, 26 January 2013 (UTC)

Citation microformat
I propose that we add microformat-style HTML class names to this template; please see Proposal: citation microformat and discuss there. Andy Mabbett ( Pigsonthewing ); Talk to Andy; Andy's edits 17:44, 5 November 2012 (UTC)

Suffixes
Apropos, we should consider tweaking the citation parameters to properly delineate suffixes, which right now are just stuffed into forenames. When the templates were originally created, writing "suffix=III" seemed like a lot of extra typing for no gain in formatting, but now that we're thinking about metadata, WikiData is coming live, and so on, it makes much more sense to separate forenames and suffixes. Choess (talk) 15:24, 16 November 2012 (UTC)
 * I took the liberty of splitting this into a separate subject. Technically this would be an easy fix with low overhead. The issue would be in updating the gazillion existing uses. ---— Gadget850 (Ed)  talk 16:10, 16 November 2012 (UTC)
 * Well, sure, but that's the case with things like "coauthors=", too. It's certainly possible to do large-scale gathering of citation statistics like this, so it could become a regular cleanup project. Choess (talk) 14:11, 17 November 2012 (UTC)

Help copying Cite web-and-friends to my private media wiki installation
My organization has an internal installation of mediawiki software. I've been trying to copy Cite web to our installation, but I'm running into dependency hell and other problems. First, I copied Cite web and anything it depends on, transitively, to my wiki (full list at end of message).

Now, whenever I try to insert a Cite web, my installation gives an error:
 * Warning: This page contains at least one template argument which has a too large expansion size. These arguments have been omitted.
 * Warning: Template include size is too large. Some templates will not be included.

I would very much appreciate any help that anyone can give.

This is the full list of what I copied: Cite web, Citation, Citation/core, Citation error, Citation/identifier, Citation/make link, Loop, Only in print, Namespace detect showall, Hide in print, =, Str endswith, Str left, Str trim, Tl, Str len, Str len/core. Does this seem right? Why does citation need a looping template!?

Thank you, 128.112.139.195 (talk) 23:46, 7 February 2013 (UTC)

loop is used in Citation/core to fill in the authormask parameter field. it inserts the em dash. 65.88.88.127 (talk) 20:58, 10 February 2013 (UTC)
 * i'm sorry, i realised i did not answer your ?: i've no idea why loop is needed. as for the dependency hell, i think this is probably because development in the particular area happens haphazardly. it reminds me of the spaghetti code of my youth. the syntax and args were different, but the real-world effects sure seem similar. 65.88.88.127 (talk) 21:30, 10 February 2013 (UTC)
 * thinking further on this, i believe loop is used so that editors can insert any number of em dashes as mask. i'm all for giving editor options; this however is misguided, and an example of misappplied flexibility, as it can certainly confuse readers, the target audience. editors should have the option to include author masking at their discretion; however as a matter of both wikipedia style consistency, and semantic disambiguity, the presentation of the authormask should be fixed. if i recall correctly, in the past it was, at 2 em dashes. it should be so again. 70.19.122.39 (talk) 16:10, 22 February 2013 (UTC)
 * to add, this will (infinitesimally i'm sure) help the template's performance, as it will mean doing away with yet 1 more nested template or #if expression. it's a very small step in the right direction. 70.19.122.39 (talk) 16:14, 22 February 2013 (UTC)

Display of work date when authorless
Works without authorial (editorial, etc) attribution display date of work publication in a manner outside of style. Style has for authored, edited, etc.'d works the date displayed after the producing authority in brackets. The producing authority for authorless works would either be the whole title, or the title and place / manner published.
 * Display: "The 25 Best Album Covers of the Decade (2000–2009)". Paste. November 16, 2009. Retrieved April 5, 2010.
 * Cite:
 * Cite code:
 * Expected Display (OR):
 * "The 25 Best Album Covers of the Decade (2000–2009)". Paste. (November 16, 2009). Retrieved April 5, 2010.
 * "The 25 Best Album Covers of the Decade (2000–2009)". (November 16, 2009). Paste. Retrieved April 5, 2010.


 * That is the style. Per the documentation:

"date: Full date of source being referenced in the same format as other publication dates in the citations. Do not wikilink. Displays after the authors and enclosed in parentheses. If there is no author, then displays after publisher."
 * This is an online magazine, so cite journal is more appropriate. And Paste is not the publisher, but the name of the magazine— which is why you were having to add the italics.


 * Without author:
 * With author:
 * --  Gadget850 (Ed) talk 14:08, 4 April 2013 (UTC)
 * --  Gadget850 (Ed) talk 14:08, 4 April 2013 (UTC)
 * --  Gadget850 (Ed) talk 14:08, 4 April 2013 (UTC)

Removing COINS metadata
There have been many complaints (e.g. https://bugzilla.wikimedia.org/show_bug.cgi?id=19262) that articles take too long to render. For articles with many citations, the obvious low-hanging fruit is COINS metadata. For example, Muammar Gaddafi takes 28.3 seconds to parse, but with COINS removed, it takes 22.2 seconds.

Nobody ever held a straw poll asking the community "can we please make article parsing 27% slower in order to support a rarely-used metadata feature?" I'm sure that data can be provided in some other way. So I would like to remove it. -- Tim Starling (talk) 06:10, 12 November 2012 (UTC)
 * Could it be switchable. For most pages its not a problem it only when citation count gets into the hundreds that thing become a problem.--Salix (talk): 08:36, 12 November 2012 (UTC)
 * Given that lua and major performance improvements to template parser performance is not far off, it's probably OK to temporarily drop metadata until that point. Rjwilmsi  10:46, 12 November 2012 (UTC)
 * It would be one way to see if anyone is actually using COinS. ---— Gadget850 (Ed)  talk 11:04, 12 November 2012 (UTC)
 * Salix: I would have thought that it would not be useful to anyone to provide an unreliable source of metadata, available only on small articles. Usually metadata consumers want perfect regularity so that they don't have to make special cases. I was more thinking of a separate page where references with metadata would be displayed, with no article content. -- Tim Starling (talk) 22:58, 12 November 2012 (UTC)

I've made the change. -- Tim Starling (talk) 22:58, 12 November 2012 (UTC)
 * A positive side effect of this change is that now &lt;math&gt; in titles of papers doesn't lead to broken formatting — e.g. see the Benjamin and Stein references in Squared triangular number. —David Eppstein (talk) 23:34, 12 November 2012 (UTC)


 * Yes! As to what might be done with COinS in the future: could it be an add-on option? ~ J. Johnson (JJ) (talk) 01:47, 13 November 2012 (UTC)
 * I'm not certain that there is a meaningful consensus to do this removal, but as an interim measure it may be acceptable. Do we know what else is broken by the removal of the COinS data, and when the more performant framework is expected to be in place? LeadSongDog come howl!  15:05, 13 November 2012 (UTC)
 * I suggest to also look at Andy's proposal one section up. A coordinated approach with the developers of Zotero, might have been smart. Now a very popular citation tool is probably broken. —Th e DJ (talk • contribs) 16:00, 13 November 2012 (UTC)
 * Oh and of course the idea of cite quick warrants re-examination. —Th e DJ (talk • contribs) 16:06, 13 November 2012 (UTC)
 * I disagree with their removal, at least in the short term, but see the section immediately above, for an alternative, with no such overhead. Andy Mabbett ( Pigsonthewing ); Talk to Andy; Andy's edits 16:25, 13 November 2012 (UTC)

As someone who installed Zotero with the notion of using it to help author Wikipedia articles, I'm afraid I doubt that it's "very popular" here. The COinS implementation was rather badly broken, because it assumed that the content of the parameters in Citation/core was plaintext and passed it on to COinS accordingly. Of course, it's not; the content of the parameters is wikitext, which (for our sins) is markup. The unparsed wikitext is presumably what was breaking mathematical formatting as well, but it extends considerably beyond that: italics in article titles and wikilinks around journal titles, etc. were being passed as ... and ..., respectively, into the metadata. Even worse, the template inserts a space between last and firstname, using an HTML entity, &amp;#32;. Since this is not parsed, it is, instead, escaped, and winds up tacked onto the front of the author name in the metadata. (So "Jones" becomes "&amp;#32;Jones"). This is evident in the blog post Andy mentioned elsewhere. I love the idea of being able to store my references in Zotero so I can use them to write multiple articles, but it's better to leave it off then to emit metadata as broken as what I've just described. Choess (talk) 04:48, 14 November 2012 (UTC)

Disappointed
Hi - we are deeply disappointed that you removed COinS. LibX (libx.org) is a COinS processor used by over 200,000 users, with configurations for over 1,000 libraries. LibX processes COinS, linking them to a library's OpenURL resolver. We are right in the middle of a project to vastly improve the processing of COinS - with Wikipedia as its primary target/beneficiary. Our goal is to use the emerging discovery services such as Summon to not only *link* users to their library, but to show them directly holdings and availability, as well as direct links, inside a tooltip emerging from the rendered COinS. We have identified tremendous potential for that; often, LibX users are affiliated with academic libraries that have subscriptions to news paper archives for articles referenced in Wikipedia where the Wikipedia link leads to a paysite.

Please restore COinS or provide the necessary metadata using an alternative way! Please do not only think of tools like Zotero that extract metadata, but of tools like LibX who use the metadata to alert their users on how to obtain the item(s), in a way that is tied to the location of the reference. Thank you! — Preceding unsigned comment added by 2001:468:C80:2129:ED30:BB99:133F:FCD5 (talk) 19:48, 15 November 2012 (UTC)

Extracting metadata is one purpose, but *simply accessing an article referenced by Wikipedia* is another. Practically no journal or news paper article to which WikiPedia links is accessible without subscription. Subscriptions that library users have, and that the knowledge base in their OpenURL resolver knows about. FWIW, users really click on COinS. In the last 2.5 months, libx.org's Google analytics shows that 5,461 users clicked, and that's just for 10 libraries who currently have this feature enabled. A realistic estimate, once we open this feature to all libraries, is around 10,000 click per month. I see real use, and it's a shame that this decision was made without consulting with COinS users. Perhaps should have provided a way for COinS processors to signal when a user is making use of the COinS, then base your decision on actual use. — Preceding unsigned comment added by 2001:468:C80:2129:ED30:BB99:133F:FCD5 (talk) 22:07, 15 November 2012 (UTC)
 * As the above comments illustrate, removing such a facility without consulting with, or at least informing (with adequate notice), those who use it, is a really bad thing to do. Andy Mabbett ( Pigsonthewing ); Talk to Andy; Andy's edits 22:19, 15 November 2012 (UTC)

The previous method for delivering COinS metadata was ridiculously expensive, in terms of server time and user experience. 5500 clicks per month is a very small number compared to what it was costing us. You can reconstruct the relevant metadata by parsing the template invocations. The wikitext is available via the API and the XML dumps. The template parameters are documented.

I'm sorry for the lack of notification, but the nature of the metadata delivery (embedded in article HTML) meant that there was no way to gather meaningful access logs. As far as I can tell, WMF was not involved in the deployment of this COinS metadata system, so we had no opportunity to establish a user notification system. -- Tim Starling (talk) 23:07, 15 November 2012 (UTC)
 * Tim the COiNS code was not ridiculously expensive, the code was quite compact, with about 50 if statements, didn't use an sub-templates or anything expensive. The number of if's is largely due to the need to process upto 9 author. The only problem was when you had 200+ invocations. Less than this you don't see any performance hit. --Salix (talk): 23:41, 15 November 2012 (UTC)
 * Performance was not the only issue with this code. In some cases it was generating malformed html with visible bad effects, because of a combination of (1) passing wikicode directly into coins data, and (2) unrelated parts of wikimedia re-interpreting the wikicode within the coins data. —David Eppstein (talk) 23:56, 15 November 2012 (UTC)
 * Unfortunately it doesn't seem to be shown in the various cite template documentation pages just which parameters are were passed into COinS, just cryptic notes like the one seen at Template:Citation: "This template embeds COinS metadata in the HTML output, allowing reference management software to retrieve bibliographic metadata; see Wikipedia:COinS. Be careful using templates within the citation template, as many will add a lot of extraneous HTML or CSS that will be rendered in the meta-data. Known templates that should not be used: smallcaps." But there is no warning about any other wiki markup; some parameters do state "Do not wikilink", without explaining why; and some, such as publisher state "May be wikilinked if relevant", even though that is  was passed into COinS (as.
 * I have often amended a parameter like  to John DoeDoeJohn; and sometimes this get reverted. Upon reversion, I've tried to explain about COinS - to which I have occasionally received responses like: "I never use COinS, so it doesn't matter"; "Never heard of it - seems useless"; "Is there really something that uses it?"; "why complicate things?" etc. or the most telling: "why should we care what outside agencies do with our data? Our priority should be to get the Wikipedia page looking consistent and professional, so if the journal paper's title includes a mathematical formula, we need to display it exactly how it appears". I did successfully advise one user to use the HTML entity   instead of some  markup, but that was a one-off.
 * We really need something that verifies that the data is clean, either by showing an error message, or by passing the data through a cleanup function. Both of these will probably have to wait for Lua. -- Red rose64 (talk) 09:38, 16 November 2012 (UTC)
 * There was a workaround for the math problems in titles archive but it never made it out of the sandbox.--Salix (talk): 10:06, 16 November 2012 (UTC)
 * The metaparameters that generated COinS are still documented at Citation/core. I am going let this settle a bit before doing any documentation updates. ---—  Gadget850 (Ed)  talk 11:05, 16 November 2012 (UTC)


 * Please elaborate! Are you saying you have an API to extract citation metadata?  API doesn't describe it.  It needs to be a REST API to be of use.  Can you provide an example, say the REST URL for reference 70 on this page? — Preceding unsigned comment added by  user name or IP  (talk • contribs)  time, day month year (UTC)  — Preceding unsigned comment added by 71.206.142.54 (talk • contribs • WHOIS ) 17 Nov 2012
 * The API does not know anything about citations. The suggestions Tim gave are slightly preposterous for what LibX and Zotero had been using COinS for.  The XML dumps seem absolutely useless for a client.  One would have to use them on another server that would do the heavy lifting.  The API is also not needed.  Tim's proposal is to grab the raw wikitext from pages.  This should not be done by the API, but by using action=raw, a'la .  One would then have to parse out all of the calls to the many cite templates.  This is a site-specific hack that offers no insurance of stability.  --Karnesky (talk) 13:59, 18 November 2012 (UTC)

The LibX click-throughs are definitely a lower bound for actual use of COinS. There are plenty of click-throughs that aren't reflected in their limits & other popular consumers of COinS metadata. Removing COinS seems to be a case of premature optimization. We should put a higher value on WP:V than on performance. Heavily-referenced pages are the minority. On the vast majority of pages where this was used, there will be no measurable performance hit and it will usually render OK. It also provides astute readers users easy ways of collecting and referring to sources used in Wikipedia. Yes, there are edge-cases we should address better. But no, we don't need to have a perfect implementation for it to be useful. There seem to be several comments here that would support reverting this change (though improving how it operates and providing other machine-readable standards-based markup that might accomplish this in the future). --Karnesky (talk) 18:38, 16 November 2012 (UTC)
 * Removing COinS was certainly not "premature optimization". Editing articles was broken by COinS. We're not just talking about slowness here, we're talking about the parser timing out on longer articles. The COinS code probably should have been removed back in April when we got the first reports of timeouts. This most recent round of discussion was prompted by timeout reports for Muammar Gaddafi on en.wiki and Islam on ar.wiki (which also uses COinS in Citation/core). The ability to edit articles is a much higher priority for Wikipedia than any metadata services. Even if COinS had thousands of users per day (rather than per month), I would still support disabling it given the present circumstances. Kaldari (talk) 19:54, 16 November 2012 (UTC)
 * Muammar Gaddafi is a single page. Arguably, it still suffers significant performance issues.  This template is used on 820,917 pages.  Even if there were complaints about dozens of pages (instead of a few), we shouldn't make the template design based around them, alone.  My suggestion would be to NOT USE Citation/core on these few edge cases.  You can optimize other template issues then, as well.  But why degrade the metadata in the hundreds of thousands of articles that had been using COinS without issue? --Karnesky (talk) 22:46, 16 November 2012 (UTC)
 * That's what we've been doing since April, but it's like playing whack-a-mole. The real danger is that it would happen to an article like Hurricane Sandy (during the hurricane). I was actually surprised that the Barrack Obama article survived the election without going over the edge. Kaldari (talk) 23:29, 16 November 2012 (UTC)
 * And aren't we just going to have to play whack-a-mole when this work around proves insufficient? As noted, the page load times are still quite long on some of those articles.  How many citations would push them over again and what is your solution for when that happens? --Karnesky (talk) 13:59, 18 November 2012 (UTC)
 * You're right that this isn't a good long-term solution. But it should buy us enough time to get Scribunto in place. Once that is done, we'll be able to easily support COinS (and with correct formatting). Kaldari (talk) 22:01, 18 November 2012 (UTC)

Wouldn't it be possible for tools like Zotero to read the source code of the article and get the metadata from that? I've written code for parsing cite/citation templates; it's not particularly difficult, and should work a lot more reliably than trying to decode it out of the COinS data with much less impact for people who don't use these tools. —David Eppstein (talk) 20:02, 16 November 2012 (UTC)
 * The nice thing about standards is that there are so many of them?  WP content is mirrored at the HTML-level to many sites and is saved and distributed.  Tools like Zotero and LibX should not have to use site-specific methods to get this metadata. --Karnesky (talk) 22:46, 16 November 2012 (UTC)
 * I just noticed that Zotero was not behaving as it used to and came here to complain - oh dear, researchers and their tools be damned! This is really a bad decision, it really seems to imply that well referenced articles are the problem. Shyamal (talk) 16:38, 29 January 2013 (UTC)
 * Hmm, I just took a look at Muammar Gaddafi for the first time, and sure enough, it is a poster-child for the excessive use of navigation templates. It remains unusable, despite the removal of COinS, because the nav templates are the real problem. LeadSongDog come howl!  21:40, 29 January 2013 (UTC)

I had no idea until just now that my inability to use Zotero to pick up references from Wikipedia pages was. It seems to me to be a very bad decision and I hope it can be reversed. Peter coxhead (talk) 22:30, 4 February 2013 (UTC)
 * in my experience, you should not nurture any such hopes. bad decisions are not reversed here. basically a dispute among less than a millionth (that's 0.000001%) of unique daily readers according to wikipedia's own statistics, about one article's performance, which may or may not be partially or wholly related to the existence of the coins facility in the. template, led to the removal of functionality for everyone. but (supposedly) this is temporary while we are waiting for the greatest and latest. to be delivered by some of the same people who took that very same decision. whether that should inspire any confidence is imo, a fair question. 65.88.88.127 (talk) 21:14, 10 February 2013 (UTC)

LUA deployed
I hear that the new Scribunto extension (deploying today) and LUA templates - like this test http://test2.wikipedia.org/wiki/Module:Citation are expected to bring back this much needed feature that has really become one of the key attractions of MediaWiki in scholarly communities. Shyamal (talk) 08:41, 18 February 2013 (UTC)
 * the point is, coins functionality should not have been removed without wider participation. it worked, and was only marginally related to the dispute that eventually led to its removal. there was a direct connection between the existence of coins facility and the disputed article's performance (this was never demonstrated), on balance coins should have remained and the article fixed instead, not the other way around. waiting for lua templates to be tested and weathered does nothing to ameliorate this. i for one will be eagerly waiting to test the performance of the new citation implementation vs the old one, using real live articles and real end-user configurations. something's wrong when high-impact decisions like the removal of coins are taken in nonchalant manner. 70.19.122.39 (talk) 15:07, 20 February 2013 (UTC)
 * Cite web/lua, Cite news/lua, Cite journal/lua and Cite encyclopedia/lua} are now available for testing. They use the same parameters as the standard templates but do support CoINs. Please don't use them in live articles yet. --— Gadget850 (Ed) talk 15:27, 20 February 2013 (UTC)
 * Suspect you ment Cite journal/lua for the last two.--Salix (talk): 17:29, 20 February 2013 (UTC)
 * I hate this laptop keyboard. Fixed my links. --— Gadget850 (Ed) talk 17:52, 20 February 2013 (UTC)
 * I've now imported and created a test case Template:Citation/India so we can see how they cope with a tricky example.--Salix (talk): 18:05, 20 February 2013 (UTC)

First difference I've spotted is has a  tag for urls but this is not included in. Compare


 * Result:
 * Result:
 * Result:
 * Result:

no normal visible difference but the html sources are considerable different

"Ministry of Home Affairs (Department of Border Management)" (DOC). http://mha.nic.in/docs/BM_Intro(E).doc. Retrieved 1 September 2008. .

"Ministry of Home Affairs (Department of Border Management)" (DOC). Retrieved 1 September 2008. &#160;.

The latter has the CoiNS but no printonly.--Salix (talk): 19:11, 20 February 2013 (UTC)
 * Please report issues to Module talk:Citation/CS1. I just added a problem with anchors. --— Gadget850 (Ed) talk 20:01, 20 February 2013 (UTC)


 * Looks like this might actually be a bug in . When printing the class="external text" causes the url to be added after the element. So you actually get the url appearing twice when you try and print.--Salix (talk): 23:00, 20 February 2013 (UTC)
 * Template talk:Citation/core/Archive 9 --— Gadget850 (Ed) talk 00:40, 21 February 2013 (UTC)

HTML classes
Now that Lua has deployed, we should add (as well as, or instead of, COinS) HTML classes to describe the various parameters. For example, instead of emitting, say,



we could emit:



The visual rendering would not change.

By agreeing (and sharing with the wider web community) a standard set of such class names, others can write tools to parse our citations, and allow them to be inserted into other documents or web services (or, indeed, into other Wikipedia articles). The makers of Zotero, for example, have already expressed an interest in parsing citations that use such classes.

As some of you may have realised, what I am talking about, a standard, shared, set of class names, is a microformat. I have written more about how we could use a citation microformat, at Wikipedia talk:WikiProject Microformats. Andy Mabbett ( Pigsonthewing ); Talk to Andy; Andy's edits 23:27, 11 March 2013 (UTC)

COinS restored
With the new Lua version, COinS has been restored for the six updated templates. There are still issues with polluted output to work out, such as math exposing the strip markers in the COinS. See Module talk:Citation/CS1. --  Gadget850 (Ed) talk 14:43, 4 April 2013 (UTC)