Template talk:Citation/core/Archive 2

Removal of dateformat parameter
While I am inclined to agree with Fullstop, that all dateish parameters should be left alone by all cit(e)(ation) templates, the dateformat parameter should not have been removed, which removes what little control editors had over the matter. --Gerry Ashton (talk) 04:55, 10 February 2009 (UTC)


 * The change in question has been undone. --Gerry Ashton (talk) 05:26, 10 February 2009 (UTC)


 * If you follow what I wrote above, including the provided link to the fix (alt: citation/fixdate), you might also observe that the dateformat parameter is superfluous.
 * The "little control editors had over the matter" is not necessary when editors retain full control. JUST FOLLOW THE LINKS if you can't understand what I'm talking about. -- Fullstop (talk) 06:35, 10 February 2009 (UTC)


 * The Citation/fixdate approach seems like a good solution. I edited the documentation to indicate the valid range, which I determined through experiments at User:Gerry Ashton/sandbox3. --Gerry Ashton (talk) 17:13, 10 February 2009 (UTC)
 * Making it so YYYY-MM-DD converts to DD-MM-YYYY is such a stupid idea and is a major inconvenience, unless there is a bot that can go and fix articles that don't use that formatting. I don't recall a discussion, so why not go back to the better format? YYYY-MM-DD should display as YYYY-MM-DD. Or at least go back to letting editors pick the format, Template:Cite web had it right. Gerry, how has it been undone? I just tried using that parameter and it isn't working since articles are still displaying as DD-MM-YYYY in the ref section of Jeff Hardy (which I used the dateformat for in refs).  TJ   Spyke   03:08, 11 February 2009 (UTC)


 * Dates in the format YYYY-MM-DD are not accepted in any citation style because they are very unfamiliar to readers. You should not be using them. —Remember the dot (talk) 03:16, 11 February 2009 (UTC)
 * Had they not been the default in many cite templates, I wouldn't. This wasn't an issue either when we could just link the date (which would then display as either MM-DD-YYYY or DD-MM-YYYY depending on what the user has in their preferences), but then some users started complaining about how we shouldn't link dates in refs.  TJ   Spyke   03:24, 11 February 2009 (UTC)

Spyke, an example of how an access dates in the Jeff Hardy article displays is "8 October 2007". It is confusing to call this a DD-MM-YYYY format, because some might think you mean it dispays as "08-10-2007". Also, your statement "This wasn't an issue either when we could just link the date (which would then display as either MM-DD-YYYY or DD-MM-YYYY ..." is 99% false, because around 99% of our readers do not have accounts.

Finally, you asked how has it been undone? I didn't change anything (I am not an administrator). I surmise that the date processing function has been changed so it will convert dates in the YYYY-MM-DD format to the day-month-year format, and leave other formats alone. The dateformat parameter seems to be ignored. To fix the "Jeff Hardy" article, you could go in and convert all the dates like "2007-10-08" to "October 10, 2007". --Gerry Ashton (talk) 04:22, 11 February 2009 (UTC)


 * This is a terrible change to the template. So if the dates need to be in MM-DD-YYYY, we now have to change them all by hand?  Can someone at least give us a tool to do so?  Something like citation/fixdate, but that putputs in American format, or a script that would permit international dates to be changed into American format in the thousands of pages that need that format? -- 2008 Olym pian chit chat 04:52, 11 February 2009 (UTC)


 * I wrote a tool a while back that automatically converts ISO dates to either DMY or MDY format throughout the article: User:Remember the dot/ISO date format unifier.js. You might find that helpful. —Remember the dot (talk) 04:38, 11 February 2009 (UTC)


 * Personally, I'd rather use the "1 February 2000" format in all citations so that you don't have to worry about date formats at all when copying references between articles. —Remember the dot (talk) 04:43, 11 February 2009 (UTC)


 * Thanks! You wouldn't happen to know about one that changes from International to American or vice versa?-- 2008 Olym pian chit chat 04:54, 11 February 2009 (UTC)


 * User:Lightmouse/monobook.js/script.js doesn't currently handle ISO dates in references or treat references differently from prose, but it can convert dates to and from US and international format. —Remember the dot (talk) 05:12, 11 February 2009 (UTC)
 * Lightmouse's script, unfortunately, does not reformat dates in references.-- 2008 Olym pian chit chat 08:42, 11 February 2009 (UTC)


 * A tool that converts YYYY-MM-DD to a format with month names is fairly safe, because few quotations will have the YYYY-MM-DD format. A tool that changes American to International or vice versa is tricky; the operator will have to make sure it isn't allowed to change dates within quotations.
 * Date format described as YYYY-MM-DD is ISO based style.--Namazu-tron (talk) 05:22, 11 February 2009 (UTC)

Given that is designed to be completely transparent when its dateformat is set to "none", surely the least disruptive way to stop the citation templates from reformatting dates undesirably is simply to default the dateformat parameter to "none" on all templates? This is already done on many of the templates, with complete success:

Spot the difference. This 'fix' was, in my opinion, rather rushed. Happy‑melon 08:55, 11 February 2009 (UTC)

Instead, it was the changes that caused the breakage that were "rather rushed"; it was the update to citation/core that was "rather rushed"; it was the supposition that was at fault for the breakage that was "rather rushed"; and it was the carry-over of cite xyz legacy crapola into citation/core that is "rather rushed". All these things happened because someone did not think things through, and did not invite discussion which might have caused the error to be noted. A template used by tens of thousands of articles demands a conservative approach, and a shoot-first approach is inexcusable. Moreover, because mistakes happen even when the greatest of care is taken, the first sign of breakage should automatically provoke a revert. Immediately. Excuse-finding, dilly-dallying and oh-hum-the-breakage-is-elsewhere is completely unacceptable. The obsolete notion that 's date= fields need to be in yyyy-mm-dd (NOTE) format was to facilitate date linkage (by  ). Regardless of how dumb it was to begin with, that practice is history now, and there is no need for that yyyy-mm-dd silliness anymore. Besides, it was only ever (not  ) that did such gratuitous date linking, and if  has this legacy to live with, then  needs to wipe its own backside before calling citation. -- Fullstop (talk) 18:39, 11 February 2009 (UTC)
 * "Spot the difference" between doing nothing and doing nothing? Citation does not have to call another template to do nothing. Citation is quite capable of doing nothing all by itself. Its rather bizarre to suggest that one template should call another template with an explicit command to do nothing.
 * Since breakage to ./core was HappyMelon's responsibility (as committing admin), and it was HappyMelon's opinion that a revert need not be done, HappyMelon is not in an ideal position to determine what constitutes "rather rushed". It was his own actions that precipitated the need to find another solution that doesn't break citations. That (or any other date-formatting routine) can't be called for anything but access-date= has been known for at least a year. No question of "rather rushed" there.
 * Citation/fixdate does what does not do, and was not designed to do.  (or any other date-manipulation routine) cannot be called for data that is not assuredly a calendar date. That is why (previously) citation never called  for anything but access-date=. This should be patently obvious to any template-coder who also actually writes citations on a regular basis (as distinguished from someone just abstractly thinking about them).
 * TJ Spyke's and 2008Olympian's comments above indicate that they haven't understood what is going on. Relax. Nothing is broken anymore, which is not what can be said for the state of affairs before the so-called "rather rushed" fix, the documentation for which can be read at citation/fixdate. There was no such thing as a 'DD-MM-YYYY' date format, and it is not meaningful to suddenly begin writing dates that way. There is no need to change anything (which too cannot what can be said for the state of affairs before the so-called "rather rushed" fix).
 * Fullstop, regardless of it being unnecessary for the YYYY-MM-DD format any longer, it is used in thousands of articles, so unless someone fires up a bot to correctly identify and correct the YYYY-MM-DD dates so they are formatted in the correct date format (US or International), there is going to be problems with articles having some citations displaying YYYY-MM-DD and others displaying DMY. Citation/fixdate also seems to be causing issues with calling #time too many times on articles with a large number of sources. As an example, Barack Obama has 215 sources at the moment and source 181 displays "Error: too many #time calls" instead of the publication date. I don't recall seeing this problem prior to citation/fixdate being implemented. --Bobblehead (rants) 23:07, 11 February 2009 (UTC)
 * Not only is fixdate slow, it displays dates incorrectly when editors want ISO format dates. Let's get rid of fixdate. Please see below. Eubulides (talk) 23:11, 11 February 2009 (UTC)


 * @Bobblehead: citation/fixdate replaced citation's previous calling of for everthing, so citation would in fact have previously caused the error at ref 181.
 * @Eubulides: There is no such thing as a "slow" template on Wikipedia. It is however wasteful to do the same 'X' multiple times in citation/core. Thats what citation is there for, i.e. to do 'X' once, and then pass the results to core.
 * @Eubulides: As I said elsewhere, citation should not ever be second-guessing editors. But I'm curious: when would an editor want ISO format dates in a citation?
 * -- Fullstop (talk) 09:20, 12 February 2009 (UTC)
 * YYYY-mm-dd dates are compact, most-significant figure first, internationally unambigious (translatable), numerically machine parsable (date extraction/date conversion), sortable and can be abbreviated just YYYY-mm or just YYYY. I do want to use YYYY-mm-dd dates in every non-prose location I can.  —Sladen (talk) 12:07, 12 February 2009 (UTC)
 * I also prefer YYYY-mm-dd in citations, for compactness there. Citations are not English text and normally use short notations that wouldn't be acceptable in main text (e.g., "3" rather than "volume 3", "JAMA" rather than "Journal of the American Medical Association") and dates are no exception to this rule. Eubulides (talk) 21:30, 12 February 2009 (UTC)


 * Any consistent date format is machine-parsable; dates don't have to be numeric for that. Sortability is not an issue in citations either -- it's a citation, not a list. Translation is an issue, but not a very large one because English citations should be avoided in non-English projects as much as possible. And however much they abbreviate to save ink, no manual of style recommends using YYYY-mm-dd dates in citations. —Remember the dot (talk) 23:07, 12 February 2009 (UTC)
 * Umm, this is not really the place to get into date-format wars, but ISO 690 ("Information and documentation—Guidelines for bibliographic references and citations to information resources"), perhaps not surprisingly, does say it's OK to use YYYY-mm-dd dates in citations, and recommends it over all other numeric styles. If you take a look at this set of examples of the use of ISO 690 format for citations, it contains only one example that has year, month, and day, and that example uses YYYY-mm-dd format. Eubulides (talk) 02:50, 13 February 2009 (UTC)

Section break
@Fullstop: quite simply, no. There is no responsibility on me as the committing admin to ensure that the changes made are technically correct, that they do not contain bugs, or to resolve any such issues if they arise. Nothing in the protection policy even requires that the committing admin have any knowledge of template code; literally any admin can copy a sandbox to a live page. All that is required is that there be "consensus" for the edit, either explicit or implicit, and in the case of changes to templates, there is a precedent that 'technical updates' - that is, edits that change the internal structure of a template but are not intended to affect its input or output - are uncontroversial if they really do function as intended. Of course, since there is no explicit consensus for such an edit, it is not considered wheel-warring for another admin to revert it for any reason. This is what I agree should have happened here. You are almost correct when you say that "the first sign of breakage should automatically provoke a revert"; the only word I disagree with is "automatically": if the procedures for admin action were so clear-cut, they could be written into software routines and admins would be unnecessary. If I had been online at 6am on 10 Feb, I would probably have reverted it myself; any admin who was online would have been completely correct to do so. That is the responsibility that the committing admin has, to revert their changes if it is clear that the apparent consensus was not in fact present. However, once the 'fix' was implemented, the whole situation changes. What would you consider my position to be if I had also been the admin to commit your equally unsuccessful code? It is rather unkind to conclude that I would thereby be responsible both for breaking the template in the first place, and then in failing to fix it, when all I would have done is implement two changes suggested by other editors. It is very widely accepted, indeed fundamental policy, that administrators are otherwise equal members of the community with all other editors; only when taking administrative actions do we wear our 'admin hats'. As such, you completely misinterpret this comment if you think I was in any way making an 'executive decision' that no reversion was necessary; as the edit summary suggests, I was making a comment in the capacity of an editor familiar with the structure of the templates involved. On the contrary, as more details of the scale of the problem surfaced, I would (had I still been online) have supported its reversion. In almost no situation is it the wrong approach to simply revert a page to the previous stable version, particularly with templates. Then we could have taken the time to construct a superior solution that avoids throwing the baby out with the bathwater (or at least allows us to decide properly whether we want to keep the baby before chucking it). That is what I mean when I say the 'fix' was rushed. In general, of course, if you feel that my actions constitute a misuse or abuse of administrative tools, you're looking for WP:AN or WP:RFC. Generally: It seems that there is a growing consensus that the citation templates should not attempt to be 'clever' enough to anticipate what output format is desired. This is an entirely separate question to whether the citation templates should be 'clever' enough to be able to reformat dates if asked to do so. By removing all date formatting entirely, we return to square one; why did we ever take that circular road if square one is where we want to be? This whole issue is about effort-minimisation: we seek to reduce the amount of time we require editors to put in to produce professional, consistent citations. Editors will enter dates in a myriad of different input formats - generally, whatever is most natural for them - and yet we want them to be consistently output. We have come to realise that attempting to guess which output format that should be is a Bad Idea, so we accept the need to make some form of edit to each citation to aid in standardisation. In this context, we lose nothing by continuing to provide the option to specify a dateformat parameter; IMO it is by far the easier way to standardise citations across a page; simply go to an article, decide which date format it should use, then add xxx to all the citation templates there, error-checking as necessary. The alternative is, as noted, to manually change all the input dates to the correct format. This method is, of course, always available, and is in no way in conflict with a dateformat method. So you ask what the difference is between "doing nothing and doing nothing"? It's the fact that, if asked to do something by an editor who just wants to get on with writing an article, one of them will actually do it, while the other will simply say "no, do it yourself". Which is the more constructive approach? We seem to be in agreement that having the template do things without being asked is Bad. That is not the same as saying that the template should not be able to do them at all. Happy‑melon 16:25, 12 February 2009 (UTC)


 * I infer that Happy-melon thinks that it is inevitable that editors "enter dates in a myriad of different input formats - generally, whatever is most natural for them". I disagree. I think that editors who take the trouble to use a citation template understand the custom of making citations consistent within an article, and will comply with whatever the usage is in the artcie. However, we have made it very difficult for editors to do this, because
 * Articles often have inconsistent date usage, so the editor can't decide which one to use
 * Editors of existing articles have to deal with whatever templates are already in use, so must be able to use the gazillion different Citation/Cite xyz templates. These templates have conflicting instructions and behaviour with respect to dates
 * There is no style manual that guides the development of these gazillion different templates, so neither article editors nor developers can decide which templates are right and which ones are wrong.
 * In short, editors enter a myriad if different date formats because the Citation/cite xyz editors have made it nearly impossible for them to do otherwise. --Gerry Ashton (talk) 16:41, 12 February 2009 (UTC)


 * I agree fully with that conclusion, with the exception of the first statement: I don't think it is inevitable, but I do think it is customary. I can count myself as a prime example here: I find it much easier to enter dates in yyy-mm-dd format into citation templates, and substantially quicker too.  I am usually careful to follow the style of the article, but I find it "inconvenient" to do so by changing the input format: for me, the dateformat parameter is much easier, and I'm sure it is for many other editors too; equally there is a large group of people for whom that parameter makes no sense whatsoever, and they'd rather just change the input format.  For every template, style and format, there is a large block of editors who are accustomed to using them, and they are the people we're trying to help here.  As long as the options we provide are not conflicting, there is no harm in providing more than one; by providing a variety of 'options' (and I say that with a pinch of salt because the "change the input format" 'option' is not really something we can not support :D) we can make the process as easy as possible for the greatest number of editors. Happy‑melon 08:53, 13 February 2009 (UTC)


 * (or any other kind of format conversion) is in fact "really something we can not support". It only works when a date is assuredly a calendar date, and A) "dates" in citations are not necessarily calendar dates, B) anything based on #time requires the input to be a calendar date. Moreover, as we have seen, there are editors who actually prefer seeing yyyy-mm-dd dates, so there isn't even a legacy-mode certainty. A "change the input format" option did not exist before, and the notion that "we can make the process as easy as possible for the greatest number of editors" is a fiction when it adds a new layer of complexity. The process is simple when things are kept simple. And if anyone "likes" to write "X", then "X" is exactly what they should see. Citation is a citation template, and not a date formatter. And date is a date template, not a citation formatter. One tool, one job. Citations don't benefit from a change of date/date-style, so enough already. -- Fullstop (talk) 14:01, 13 February 2009 (UTC) ps:There are umpteen never/rarely used date-manipulation templates that could benefit from consolidation with.


 * I don't consider a method of entering dates in citations to be supported until the following occurs:
 * Every Cite xxx and Citation template accepts the same syntax for that subset of dates that apply to it.
 * Any Cite xxx templates that are not used enough to bother supporting them are deleted, and a bot converts usages of the deleted template to a more popular template.
 * History suggests we are not able to provide this level of support. --Gerry Ashton (talk) 17:54, 13 February 2009 (UTC)


 * Fullstop, how can you suggest that "A 'change the input format' option did not exist before"? When has entering a date into a citation template in a different format (with the exception of accessdates in the datelinking era) not been a way to alter the output format?
 * The "layers of complexity" argument is a strawman when editors who are not familiar with the extra levels are not required to use them; as I said above, including support for a dateformat parameter in no way conflicts with editors' ability to 'fix' citation templates simply by changing the input format, nor should it.  is a method for producing inline citations in a consistent and professional manner, part of which is to ensure that dates are appropriately and consistently rendered.  The template is indeed not a date formatter, that's why it calls a separate template to handle that functionality.  There is no logic in saying that because its primary purpose is not date formatting, it should not in any circumstance even delegate that task.
 * Gerry, once again I fully agree. The number of underused citation forks is obscene. Happy‑melon 20:34, 13 February 2009 (UTC)

Please don't mess with editor's date format
The recent changes to the citation templates have caused problems with the pages I help edit. For example, this pair of citations, taken from the featured article Daylight saving time: currently generates this: The formatting is currently inconsistent. The first line has the behavior that I want, except that it wikilinks the date, a longstanding bogosity which I assume will be fixed eventually. The second line reformats the date to use a format like "20 March 2007", which is even worse: I want the date to appear the way that I wrote it. This behavior of the second line is new, and is a step backwards.

I notice above that several people have requested that the citation templates stop reformatting dates, and just use the dates that editors prefer. Currently this has been done, except when editors prefer ISO format dates. There should be no exception: if an editor prefers ISO format dates, it's not this template's job to override the editor's preference.

There's a simple fix for the problem: remove all occurrences of Citation/fixdate from Citation/core. I have tested this with an edit to the sandbox. This simple change will make the templates easier to use and to describe. There is a backward-compatibility concern with accessdate=, but that is a relatively minor issue compared to the continuing and longrunning hassles of messing with what editors write. I also realize that this fix will not fix the Cite paper problem with wikilinking dates until that template is adjusted to use Citation/core, but that's OK: we can make that change later. Eubulides (talk) 22:30, 11 February 2009 (UTC)


 * Notice to Eubulides: You are put on notice that the ISO 8601 standard requires the use of the Gregorian calendar. Henceforth, if you write a Julian calendar date in the format YYYY-MM-DD and describe it as an ISO date, you will have lied. --Gerry Ashton (talk) 22:37, 11 February 2009 (UTC)
 * That's fine, and it doesn't affect my request. Wikipedia article citations invariably use ISO format to specify Gregorian dates, so the point about Julian dates has no practical relevance to the main subject of this thread. Eubulides (talk) 22:45, 11 February 2009 (UTC)


 * No, books and monuments from before the conversion to Gregorian can be cited. Of course, if you only consider the access date of on-line resources, you would be right.
 * Yes, in theory date= could mention a Julian date using ISO format. However, in practice, articles on older subjects (for example, John Dee, Philitas of Cos) simply don't do that. So in practice the proposed change won't affect any formatting of Julian dates. Any problem in future use of this template can be handled simply by documenting the new behavior (something that should be done anyway, of course). Eubulides (talk) 23:24, 11 February 2009 (UTC)


 * So, are we in agreement that dates in citations should simply be displayed as entered? —Remember the dot (talk) 23:26, 11 February 2009 (UTC)
 * Considering citation/fixdate breaks cites in articles with a large number of sources, dear god, yes.;) --Bobblehead (rants) 23:28, 11 February 2009 (UTC)
 * Yes. The breakage to Barack Obama described above in sounds relatively urgent, so I just now added editprotected to this request. Eubulides (talk) 23:49, 11 February 2009 (UTC)


 * Okay, dates should now be displayed as entered: what you see is what you get. This is a lot simpler and more straightforward, but it also means that we will have to actively convert dates like "2009-02-11" to the format "11 February 2009". I created a tool, User:Remember the dot/ISO date format unifier.js, that can help do this automatically. —Remember the dot (talk) 01:45, 12 February 2009 (UTC)

Remember the dot, could you explain where to look to get Javascript functions to work in Wikipedia, and what to look for on the screen if your function has been successfully installed? --Gerry Ashton (talk) 03:15, 12 February 2009 (UTC)


 * Add the line
 * to your monobook.js page and clear your browser cache to install my tool. "Format ISO dates in American style" and "Format ISO dates in international style" will appear in the toolbox on the left when editing pages. The procedure for Lightmouse's script is similar, only the code is
 * —Remember the dot (talk) 06:37, 12 February 2009 (UTC)
 * —Remember the dot (talk) 06:37, 12 February 2009 (UTC)
 * —Remember the dot (talk) 06:37, 12 February 2009 (UTC)


 * good going. Thanks for the care you've taken with everything. -- Fullstop (talk) 09:23, 12 February 2009 (UTC)


 * Thanks, I got it working. A word of caution: there are more Cite xyz templates than I can count, and there is no telling what some of them might do with a date not in the form YYYY-MM-DD. Editor's using the unifier script must examine the name of the template and determine whether the template can support the change. --Gerry Ashton (talk) 16:10, 12 February 2009 (UTC)