Help talk:Citation Style 1

From Wikipedia, the free encyclopedia
  (Redirected from Template talk:Cite speech)
Jump to: navigation, search
the Wikipedia Help Project (Rated B-class, High-importance)
WikiProject icon This page is within the scope of the Wikipedia Help Project, a collaborative effort to improve Wikipedia's help documentation for readers and contributors. If you would like to participate, please visit the project page, where you can join the discussion and see a list of open tasks. To browse help related resources see the help menu or help directory. Or ask for help on your talk page and a volunteer will visit you there.
B-Class article B  This page does not require a rating on the project's quality scale.
 High  This page has been rated as High-importance on the project's importance scale.

Time to show date error messages?[edit]

Currently, errors categorized in Category:CS1 errors: dates are hidden, pending fixing of as many as reasonably possible by a bot. The hiding decision was a result of this RFC.

BattyBot task 25 has been processing Category:CS1 errors: dates periodically for about eight months now. When BattyBot is not running, date errors are added to the category at a rate of hundreds per week, maybe more. The category population was about 100,000 when BattyBot started running; it is about 60,000 now, and would be tens of thousands higher than 100,000 if not for the bot's work.

I have been working with GoingBatty, the bot's operator, to add more patterns to the list of bot-fixable errors. You can see the latest round of proposed fixes on GoingBatty's talk page, and there are many previous rounds of this exercise in that page's archives.

I believe that we have reached a point of diminishing returns with the bot, and that the bot has "run to sufficient completion" (quoting the RFC closure decision). I believe that the date error messages should be exposed by default to editors when the live version of the citation module is next updated. Thoughts? – Jonesey95 (talk) 02:30, 26 August 2014 (UTC)

I agree that it's unlikely we'll find many more patterns that will significantly reduce the number of errors, although suggestions will always be appreciated. I think our next task is to prepare ourselves (and others) for the onslaught of questions and concerns that will be raised when these errors are visible to all.
  • People will see a "(help)" link in the article, which will take them to Help:CS1 errors#bad date. Are there any improvements needed here, such as:
    • Maybe mentioning why other text in the date field is bad, and providing a link to a simplified explanation of WP:COinS with examples?)
    • A link to Wikiblame to help research accessdates?
  • From there, people will see a link to Wikipedia:Help desk. We should announce there (and where else? WP:VP/T?) when the errors are scheduled to be turned on, and ensure we have extra people checking there.
  • People may also follow the link to Help:Citation Style 1#Dates. Are there any improvements needed here?
  • What guidance will we give people who have valid dates that are flagged as false errors?
Thanks! GoingBatty (talk) 03:56, 26 August 2014 (UTC)
Do you know of any valid dates that are being flagged as errors?
Trappist the monk (talk) 13:00, 26 August 2014 (UTC)
Seasons (e.g., winter 2013) are being flagged as errors if they are capitalized, but seasons should not be capitalized. This is actually a subtle issue. If the separator between citation elements is a period, and each clause separated by a period is punctuated as a sentence, and if the season is the first word in one of these elements, then it should be capitalized. But if the separator between elements is a comma, or if the season is not the first word in the element, it should not be capitalized. Finally, neither our dates nor the date element in the APA style, is properly punctuated as a sentence, so it isn't clear that seasons should ever be capitalized. I'll poke around to see what other styles do. Jc3s5h (talk) 13:13, 26 August 2014 (UTC)
To clarify: seasons that are not capitalized are flagged as errors.
In the CS1 citations, the separator between elements is a period. In CS1 citations, season dates begin with the season. So clearly in CS1 citations, the season portion of a season date shall be capitalized.
So is your question: What should we do in the {{citation}} template citations where the separator between elements is a comma? And as a corollary: what should we do when CS1 citations override the period separator with some other character?
My answer to those is: nothing different. The element should take precedence over punctuation because the element is the important part. The element is not like the element type indicator words 'retrieved', 'archived', etc.
In the magazine and journal world, do those periodicals that use season dating capitalize the season on the cover of each issue? My experience is that they do; excepting those hipster periodicals that don't capitalize stuff for stylistic reasons and just because they're cool.
Seasons in dates in CS1 citations are roughly synonymous with months in dates. As such they should be treated in the same fashion.
Trappist the monk (talk) 14:37, 26 August 2014 (UTC)
I checked the APA Style Blog. Their advice is

If the periodical uses a season with the year, put the year, a comma, and the season in parentheses (2008, Early Spring).

But I think this requires some consensus; we don't automatically follow other style guides. The distinction between the period and comma separator still applies. Jc3s5h (talk) 13:20, 26 August 2014 (UTC)
APA style dates are not supported by MOS:DATE and in the example you've posted, the season is capitalized.
Trappist the monk (talk) 14:37, 26 August 2014 (UTC)
The last time I asked about unhiding hidden error messages, I got no real support. I remain in favor of showing error messages so I support showing the date error messages at the next CS1 update.
Trappist the monk (talk) 13:00, 26 August 2014 (UTC)
BattyBot's latest run, which incorporated the new fixes linked above, fixed only 500 articles, out of the 60,000 in the category, and many of those articles had errors introduced between the bot's last run (which finished a few days ago) and this one. I think the bot has reached a point of substantial completion and we have met the conditions of the RFC.
GoingBatty makes some good suggestions above. Let's make sure our documentation is in order before the next code update. I have tweaked the help text a bit already. – Jonesey95 (talk) 13:42, 27 August 2014 (UTC)
I think it is a bad idea to enable display of 60,000+ error messages at a stroke (I assume we're talking about just Category:CS1 errors: dates not Category:Pages using citations with accessdate and no URL‎ too). The intent can only be that they should all be fixed, but the sheer size of the task is demoralising and the payoff is uninspiring. Preparing thoroughly would be essential but not necessarily sufficient. I think it is also important to be seen to have prepared (What is the effort required? How long will it take? Why is that a favourable cost/benefit ratio?), and to have considered other options (Are |date= and |accessdate= equally important? Could we focus on probable typos over incorrect style? Do we really want to enforce MOS:BADDATEFORMAT regarding dd-mm-yyyy and mm-dd-yyyy? What about trading Category:CS1 errors: dates for Category:Pages with citations having bare URLs‎ and Category:Pages with citations lacking titles‎?). To try to understand some of this better I started my own analysis of the remaining errors here which may be of interest (eg help should address the most common errors first). Is that analysis worth taking further (the sample size is currently far too small), or is there a better analysis already available? I realise some of this has been covered in the past (possibly multiple times) but I find it hard to get a handle on what is agreed, what is still controversial, and what has not been discussed.TuxLibNit (talk) 20:33, 1 September 2014 (UTC)
@TuxLibNit: Thanks for putting together this analysis. Some comments:
  • I've updated my bot's code to fix mm/dd/yy where "dd" > 12 and "yy" = 13 or 14.
  • The bot can be updated to change undated to n. d., but I was hoping that CS1 would be changed so that undated would be acceptable.
  • While my bot normally fixes July/August 2010, it did not in this case because of the (unnecessary) brackets within the citation template.
  • 1997-98 needed to be changed to 1997–98.
  • Ministry of Defense (Kuwait) contained |date=May–June 2004|year=84|volume=3, which needed to be changed to |date=May–June 2004|volume=84|issue=3
  • My bot will now fix Octrober, and fixed all five existing instances
  • I had already updated the bot's code to not create ranges such as Winter 2003–1904, and fixed all existing instances.
The rest are misuses and/or typos that can be fixed manually, and the red error will help editors to notice the issues so they can be fixed. GoingBatty (talk) 21:43, 1 September 2014 (UTC)
I've manually fixed as many of the articles that TuxLibNit identified as I could. I would appreciate if others could provide input on the validity of using undated (see above) and work on fixing the following articles:
Thanks! GoingBatty (talk) 15:54, 4 September 2014 (UTC)
Re "undated", here is a previous discussion. Style guides (at least Chicago and MLA) say that "n.d." should be used, but I think that abbreviation is needlessly obscure. I think that "undated" or "no date" should be allowed. I would suggest taking this to MOS or a wider forum, but I have been told recently that MOS is not the right place for citation-specific discussions. A few editors are fond of reminding us that nearly any consistent citation style may be used, so can we just be bold and decide that "undated" is part of CS1's consistent citation style? That would be my preference. – Jonesey95 (talk) 16:44, 4 September 2014 (UTC)
As far as undated goes, if we're going to have a style, we should follow it. If you want to let folks do their own thing, just remove all the requirements from this page and scrap all the error messages and bots. Jc3s5h (talk) 16:42, 4 September 2014 (UTC)
For the record, I did eventually update my stats to correct errors in my analysis identified from GoingBatty's comments, then added another 20 pages and used those results to guide some updates to the help. TuxLibNit (talk) 21:41, 10 October 2014 (UTC)

Is there a decision here? If not, what about some of the other hidden error messages?

Category:Pages using citations with format and no URL – 2649 pages
Category:Pages using citations with old-style implicit et al. – 248 pages
Category:Pages using citations with old-style implicit et al. in editors – 1167 pages

or, dare I even suggest it:

Category:Pages containing cite templates with deprecated parameters – 27,815 pages (down from 163,762 pages on 4 January 2014)

Trappist the monk (talk) 12:50, 15 September 2014 (UTC)

Nobody has disagreed with the statement that BattyBot has run to substantial completion, so I believe we should show the date error messages.
I believe that we should also show the deprecated parameter errors. Monkbot and BattyBot have removed the vast majority of errors from this category (its population has decreased from 164,000 articles to 28,000!). A little tweaking of Monkbot's code to add a few more template aliases may remove another 1,000 or so, but that is "substantial completion" by any measure.
The "et al." messages should be shown. I have worked with Citation Bot and manual edits to get the two categories down from about 12,000 articles to less than 1,500. This one has also run to substantial completion.
I haven't examined the "format and no URL" category in detail, but I don't think it is tractable with a bot, so those errors should probably be exposed as well.
We'll have to be ready for some communication about these error messages from people new to the conversation. Let's remember to treat them gently. – Jonesey95 (talk) 13:19, 15 September 2014 (UTC)
For the record, I agree that the bot fixes have run to substantial completion but I'm opposed to displaying the error. While I appreciate the responses that I got to an earlier post here, those responses did nothing to address my reasons for objecting. If the error is to be enabled I'd appreciate at least a few days advance notice because I intend to modify the help text so that it more directly addresses the most common errors.TuxLibNit (talk) 23:42, 17 September 2014 (UTC)
Is there a reason that you can't make improvements to the help text now? Even though the errors are hidden from most editors, there are editors who have enabled the error display for hidden errors so whatever improvements you make can help those editors right now.
And why is that you're opposed?
Trappist the monk (talk) 23:52, 17 September 2014 (UTC)
TuxLibNit: the error messages used to be displayed. They were hidden after an RFC with one condition: that a bot designed to remove the errors in each category "run to sufficient completion". That condition has been met for the "dates" category. At this time, articles with date errors are being added to the error category at a rate of about 100 per day. If we continue to hide the error messages, some erroneous dates will continue to be added, but some editors will see the error, read the (improved?) help text, and fix the errors.
I reread the comments of TuxLibNit above, and I found it difficult to determine the editor's reasons for objecting. I saw a lot of questions and a link to an analysis of a sample of dates. What are the objections at this point? Do they outweigh the clear reasons for displaying the error messages, e.g. dates generating error codes are ambiguous, missing data, or otherwise not as clear as they should be? – Jonesey95 (talk) 06:21, 18 September 2014 (UTC)
OK, I'll try to be clearer. I was trying to be brief because I didn't want to generate a wall of text and because I'm aware I'm coming late to this discussion so I thought there was a good chance I'd be pointed to some conclusive argument or agreement that would render a more complete post pointless. As that does not appear to be the case, here we go:
  • A side point regarding the RFC: My understanding of the RFC is that "run to sufficient completion" merely signals the end of an agreement that this error should be disabled. While the closer does explicitly bless re-enabling the errors that the RFC hid, I beleive this is just to restore the status quo — with the messages enabled, but their status disputed. I didn't see a consensus that those arguing for the messages to be disabled more permanently had failed to make their case and that the messages should remain enabled despite their objection.
  • Trying to be clearer about my position:
    • I'm opposed to leaving the red error messages enabled on large numbers (say more than 400 pages/category) of minor errors (subjective, but an ambiguous date in a cite definitely qualifies for me) for long periods of time (really more than a month or so, but I'll be generous and say a year). I hope the principle at least is not too controversial — that the harm done by the presence of an error should be weighed against the harm done by displaying the corresponding error message and the latter harm is at its greatest for an error that does not get fixed for a long time. The messages are there to help ensure the errors are fixed, if they are not doing that job, then they are purely harmful (in the sense that they are defacing the article with no benefit) and should be turned off.
    • I'm not disputing that the overwhelming majority of dates flagged in this category are indeed errors that in an ideal world would be fixed. However I'm not convinced that simply enabling the errors as proposed is a fair or effective way to get them fixed.
    • Regarding fairness. The intended effect of an error message is to flag up an error that a grateful (or at least dutiful) editor will then voluntarily fix because it is an error. This should apply to all major errors and to recently introduced minor errors, but in the case of sufficiently minor errors this logic can be reversed so that the editor will fix the error primarily because the error message defaces the article not because they consider the error itself particularly worth fixing for its own sake. When the errors arise piecemeal (because the error has just been introduced) this can still be tolerable but if lots of minor errors appear all at once it becomes a problem, because the editor feels forced to expend a significant amount of effort right now simply to suppress the error messages, whereas the errors themselves could have been left until later (or arguably even left indefinitely). Forcing an editor to make trivial changes would be unfair.
    • Regarding effectiveness. There are already error categories with red messages enabled that contain huge numbers of members that are not declining significantly. Why should this error be any different (i.e. get fixed rapidly)? I also regard the history of Category:Pages_with_ISBN_errors as support for my position because although there has recently been some success in reducing it, it had remained large for a long period of time and the success has come from a (more or less) coordinated effort by a relatively small number of editors making many edits, and even then, every time that effort flags, the size of the category stabilises again (growing slowly). I think this means that people are mainly fixing these errors from the category and/or from WP:WikiProject_Check_Wikipedia/ISBN_errors, not because they happen to be viewing an article containing an error message.
  • Some final comments for now:
    • It appears that no-one here realised that when I was asking questions in my first post it was because I hoped for answers.
    • There's nothing stopping me editing the help, its on a to-do list but I haven't got to it yet. If the message enable was imminent it would bump that task up my list.
    • If it helps, I could for example support unhiding the error message for a single large category (eg the date errors with 60000 members) if there was also an agreement that if the category did not decline fast enough (say by 5000 messages in a calendar month) while it remained large (say over 400 members) then the message would be hidden again. That doesn't mean I'd be willing to put in much effort into fixing them myself.
    • On the other hand I realise there is no particular need for you to get hung up on one person's objections.
    • I hope that gives a clearer idea of the arguments and reasoning behind my objections. Apologies for the long post, but you did ask for it.
TuxLibNit (talk) 22:32, 18 September 2014 (UTC)
Thank you for the long post. I did ask for it, and your post was helpful. I will respond to a few of the points raised.
  1. I got involved in fixing citation errors because I saw a red error message and it made me curious. If all of the error messages had been hidden, I probably would not be here. Since I began fixing those errors in 2013, I have fixed errors in 20,000 to 30,000 pages. I typically fix 50 to 100 errors per day. There are other gnomes and bots who have fixed many more.
  2. There are currently 28 subcategories of CS1 errors. Of those, 19 are essentially empty or are close to empty due to diligent work by a handful of editors. A few more categories have been massively reduced in the last year; many categories had "huge numbers of members" (quoting the response above) recently, but they have been addressed in a reasonable amount of time. The wikilinks error category, for example, had something like 8,000 articles in it last year, and is empty now. The ISBN error category had about 7,000 articles in it just a few months ago, and it has been reduced to 650 during 2014 (it was at about 1,000 a month or two ago; progress has slowed, since the remaining cases are more difficult than the low-hanging fruit that was there initially). The "old style et al." categories had about 12,000 articles and are now down to about 1,400. My point is that significant progress has been made on almost all of the categories. Most of this work has been done by gnomes and bots, many of whom have been attracted to the work by the error messages. I have some evidence to show that individual reader/editors are motivated by the error messages to fix errors; sometimes I will queue up 20 articles to fix, only to find when I get to the 15th article that an editor actively working on the article has already fixed the error.
  3. The categories that have a stable and large number of members may need to be discussed here individually (e.g. accessdate without URL, format without URL). If nobody wants to fix these errors, maybe we need to set them aside. The "dates" and "deprecated parameters" categories are large, but not large and stable; they have both decreased significantly because editors are interested in fixing these errors.
  4. ReferenceBot has been configured to notify editors when they add some types of CS1 errors to articles. There are additional categories that need to be added to its notification list; I will help those categories to get on the bot's list one of these days. I have evidence that these notifications motivate editors to return to articles and clean up after themselves. Not everyone does so, of course, but many do.
  5. You suggest "more than a year" as a time to determine if progress has been made on reducing error messages in articles. There are currently about 60,000 date errors, with about 100 being added daily. That says that we need to fix about 260 articles per day for a year to clear out the category. Based on my own experience with fixing CS1 errors, I think it is reasonable to expect that focused effort by gnomes and bots on this category could achieve that. Adding ReferenceBot notification (optional, to be decided here) and showing the date errors make it more likely that we could reach that goal.
Thanks for your interest and the explanation. – Jonesey95 (talk) 23:38, 18 September 2014 (UTC)
The date is one of the critical elements of a citation that is used to identify the source. Error checking is intended to identify issues in the citations for quality assurance. Browsing a sample of articles, a large number of errors are due to ambiguous formats such as 9/10/2013 that a bot cannot resolve. Editors should be reviewing their work and if they see an obvious error message they should be able to use the help page and fix the problem quickly. This not currently happening since the messages are hidden by default. --  Gadget850 talk 11:34, 19 September 2014 (UTC)
ReferenceBot sounds particularly promising. Regarding recruitment of gnomes, I have to admit that I also started fixing ISBN errors as a result of the red text but I also can't help thinking that there must be a better way. TuxLibNit (talk) 21:41, 10 October 2014 (UTC)
This obscure discussion has suddenly generated '000s of error messages all over the wiki, many relating to very simple "errors" that are generated by current "official" template-filling tools inbuilt to the editing page that throw up dates like "2012 Mar". You can't fix those by bot??? Really? You can't change the tools first??? The whole situation is absurd. Look at Pancreatic cancer for example. Wiki CRUK John (talk) 14:03, 16 October 2014 (UTC)
@Wiki CRUK John: Hopefully you started at the top of this section and read about BattyBot, which has fixed over 50,000 articles with CS1 date errors, including running five times on the Pancreatic cancer article since December 2013. Making the errors visible to all users will hopefully keep people from adding additional incorrect dates into the article, and encourage people to do the necessary manual cleanup to reduce the number of articles with errors so the bot can run more often. GoingBatty (talk) 00:40, 17 October 2014 (UTC)
Fixing the official ref toolbar in the editing window so it doesn't continue to generate "errors", or indeed changing the MOS so that it accepts perfectly unambiguous forms like "2012 Mar", would seem more efficient and realistic approaches to this. Wiki CRUK John (talk) 12:14, 17 October 2014 (UTC)

OK, so what are we supposed to do with a date like 370 BC, then? That isn't an error, but one is reported. Chiswick Chap (talk) 19:48, 19 October 2014 (UTC)

If such citations are legitimate, they must be so rare that a hand-crafted citation will suffice. CS1 is a general purpose tool suitable to most applications but not all.
Just to satisfy my own curiosity, what are you citing that is nearly 2,400 years old?
Trappist the monk (talk) 21:24, 19 October 2014 (UTC)
"370 BC" could go in |origyear=, with the publication date of the source you are actually citing (and viewing with your own eyes), or to which you are referring readers, in |date=. – Jonesey95 (talk) 01:06, 20 October 2014 (UTC)

An ISO 639-1 language name test[edit]

Sometime ago somewhere there was a discussion about ISO 639-1 that led me to do a brief experiment with mw.language.fetchLanguageName(). Right now, Module:Citation/CS1/Configuration has a table of ISO 639-1 codes and their English meanings (["af"] = "Afrikaans"). It seems silly to me for us to be maintaining a list of these translations if there is some other place we can get the same information.

So, I've tweaked Module:Citation/CS1/sandbox to use mw.language.fetchLanguageName(). In the collapse box is a list of simple cites that use the names and codes from the list in Module:Citation/CS1/Configuration.

If the tweaked code works, and if mw.language.fetchLanguageName() agrees with Module:Citation/CS1/Configuration then the citation title and the parenthetical language element of the rendered citation should be the same.

{{cite book/new |title=Afar |language=aa}}
Afar (in Afar). 

I have been through the list and found a handful of code translations that do not match. Can anyone find others?

These codes produced results different from the current table of definitions in Module:Citation/CS1/Configuration. Each is accompanied by a link to SIL International (SIL) along with a list of language names. Presumably, where two or more language names are listed, the first listed should be preferred? Right? SIL is, according to the Library of Congress web site, is the keeper of ISO 639 (here is a table of ISO 639 codes at the Library of Congress).

  • Maldivian (in Divehi). code dv: Dhivehi; Divehi; Maldivian – neither CS1 nor mw.language.fetchLanguageName() use the preferred language name
  • Haitian Creole (in Haitian). code ht: Haitian; Haitian Creole – CS1 does use the preferred language name
  • Nuosu (in Sichuan Yi). code ii: Nuosu; Sichuan Yi – mw.language.fetchLanguageName() does not use the preferred language name
  • Gikuyu (in Kikuyu). code ki: Gikuyu; Kikuyu – mw.language.fetchLanguageName() does not use the preferred language name
  • Kwanyama (in Kuanyama). code kj: Kuanyama; Kwanyama – CS1 does use the preferred language name
  • Greenlandic (in Kalaallisut). code kl: Greenlandic; Kalaallisut – mw.language.fetchLanguageName() does not use the preferred language name
  • Central Khmer (in Khmer). code km: Central Khmer – mw.language.fetchLanguageName() does not use the preferred language name
  • Norwegian (in Norwegian). code no: Norwegian – bug in mw.language.fetchLanguageName()?
  • Ossetian (in Ossetic). code os: Ossetian; Ossetic – mw.language.fetchLanguageName() does not use the preferred language name
  • Tonga (Tonga Islands) (in Tongan). code to: Tonga (Tonga Islands) – mw.language.fetchLanguageName() does not use the preferred language name

With the exception of codes no and to, mw.language.fetchLanguageName() provides an approved language name for each of the codes I've tested here. If, at the very least, code no can be fixed then I think that Module:Citation/CS1 should discontinue use of it's current table of names in favor of mw.language.fetchLanguageName().

In the mean time I will adjust the existing table so that if it takes a while before code no is fixed at least CS1 will be doing correct code to language name translation for codes dv, ht, and kj.

Trappist the monk (talk) 23:06, 1 September 2014 (UTC)

Don't forget the parser function {{#language:}}, e.g. {{#language:dv}} → ދިވެހިބަސް and {{#language:dv|en}} → Divehi --Redrose64 (talk) 23:44, 1 September 2014 (UTC)
I'm going to hazard a guess that {{#language:}} and mw.language.fetchLanguageName() are intimately related: {{#language:no|en}} produces Norwegian (bokmål) which is the same wrong language name produced in the Norwegian language test above.
Trappist the monk (talk) 23:57, 1 September 2014 (UTC)
See also m:List of Wikipedias#Nonstandard language codes. --  Gadget850 talk 01:10, 2 September 2014 (UTC)

I have tweaked Module:Citation/CS1/sandbox to provide a workaround for the code no problem. This does not fix the {{#language:no|en}} magic word which relies on the same code in mw:Extension:CLDR. {{#language:no|en}} produces: Norwegian (bokmål). More at WT:LUA#Bug in mw.language.fetchLanguageName()?

With this tweak, I think we can remove the translation table from Module:Citation/CS1/Configuration.

Trappist the monk (talk) 16:59, 10 September 2014 (UTC)

If this is intended to appear in the citation, this would be way more helpful if the output was "(language: Langname)" instead of "(in Langname)", an ambiguous construction that relies entirely upon the reader's recognition of Langname as the name of a language. That is unworkable because of the large number of obscure languages on the planet, and the fact that some of them are not distinguished from geonyms or ethnonyms.  — SMcCandlish ¢ ≽ʌⱷ҅ʌ≼  14:49, 6 September 2014 (UTC)

I suspect that '(in language name)' is an artifact of published style manuals: this example at CMOS 14.110 for example.
Trappist the monk (talk) 14:45, 10 September 2014 (UTC)

I've been wondering if the categorization that Module:Citation/CS1 applies to pages with citations that assign ISO639-1 codes to |language= is correct. Right now, if |language=de is use then the page is added to Category:Articles with German-language external links. This may or may not actually be true. For example this citation provides no links except to Special:BookSources

{{cite book |last1=Busch |first1=Rainer |last2=Röll |first2=Hans-Joachim |title=Deutsche U-Boot-Verluste von September 1939 bis Mai 1945 |work=Der U-Boot-Krieg |volume=IV |publisher=Mittler |location=Hamburg, Berlin, Bonn |year=1999 |isbn=3-8132-0514-2 |language=de}}

Here, |language= is used to identify the language of the source which is in keeping with the CS1 documentation.

So my question is: Are we correctly categorizing these pages? If yes, then we're done; if no, how should these pages be categorized?

Trappist the monk (talk) 14:45, 10 September 2014 (UTC)

I added some comments about the 'no'/'nb' issue here. – Danmichaelo (talk) 20:03, 14 September 2014 (UTC)

Based on my conversation with Editor Danmichaelo, I have taken the decision that Module:Citation/CS1 will return the string '(in Norwegian)' when editors use |language=no.

I have removed the translation table from Module:Citation/CS1/Configuration/sandbox so language codes now get the language names from Mediawiki.

Still unanswered is the categorization question. Category:Articles with non-English-language external links appears to have been intended for just that: external links. Right now, the module is indiscriminate. Whenever a citation with |language=<ISO639-1 code> is used, we add the page to the 'language' sub category of Category:Articles with non-English-language external links regardless of whether the citation has an external link. We aren't the only violators of the category, editors use {{xx icon}} to identify a string of text as a particular language (see Borgund Stave Church at the top of the infobox for an example).

Trappist the monk (talk) 11:34, 15 September 2014 (UTC)

What about using the categorisation that the lang-xx templates use? For example, at Swansea, it has {{lang-cy|Abertawe}} which puts the page into Category:Articles containing Welsh-language text. --Redrose64 (talk) 15:07, 15 September 2014 (UTC)
I don't know. The text at the top of Category:Articles containing non-English-language text does say that articles added to those categories should be added by the {{lang}} family of templates. That restriction sort of disqualifies CS1, né?
Of course the question that I haven't asked yet is: do we need to categorize pages that cite foreign language sources? My guess is that there will be some sort of maintenance or other benefit that can be gained by doing so. As an aside, I've been thinking that CS1 needs a maintenance category any way so this could be the impetus to create one.
Trappist the monk (talk) 00:18, 16 September 2014 (UTC)
No opinions? Ok, there is a new category Category:CS1 properties into which we can put stuff like Category:CS1 foreign-language sources into which we put categories like Category:CS1 German-language sources (de), Category:CS1 Spanish-language sources (es), etc. There will be about 270 or so 185 of these language-specific subcategries. I will use AWB to create them all. So that I get it right, are these names acceptable? Is there a better way to name categories that contain citations to sources in languages other than English?
Trappist the monk (talk) 13:55, 26 September 2014 (UTC)
It is done: 183 individual subcategories ordered by ISO639-1 two-character codes in Category:CS1 foreign language sources which itself is a subcategory of Category:CS1 properties.
To get the list of individual language categories, I copied the table at the Library of Congress website into a spreadsheet; deleted ISO639-2, French, and German columns; sorted by ISO639-1 and removed all rows without a two-character code; used the spreadsheet to make a simple {{cite book/new}} template that used |language=xx where xx is the ISO639-1 code; copied the list of templates to an edit window in Wikipedia; Show preview to get the names associated with the codes as Wikimedia understands them; copied the result to Notepad++ where I used a regex search and replace to create category names and other info to feed to the AWB CSVloader plugin which did all of the actual work.
Trappist the monk (talk) 23:29, 27 September 2014 (UTC)

The other half is now implemented in Module:Citation/CS1/sandbox. The code will now categorize pages with CS1 citations containing |language=<lang> where lang is a ISO639-1 language name. The code does not check spelling but is immune to capitalization and will render language name correctly capitalized. Language names that aren't ISO639-1 names are allowed but are not categorized.

Trappist the monk (talk) 16:05, 29 September 2014 (UTC)

Does this test categorize pages only in the namespaces currently categorized by our error messages, or does it categorize pages in all namespaces? I recommend the former; categorizing User and Talk (et al.) pages is probably not useful, since fixing them is often not appropriate and they are not "encyclopedic" pages exposed to regular readers of WP. – Jonesey95 (talk) 18:35, 29 September 2014 (UTC)
Only categorizes mainspace and only if |language= is something other than en or English. This follows the precedence set by other language categorizes like the {{xx icon}} templates. These:
  • *{{code|{{cite book/new |title=Title |language=gd}}}}
  • *{{code|{{cite book/new |title=Title |language=Scottish Gaelic}}}}
don't categorize here:
  • <span class="citation book">''Title'' (in English).</span><span title="ctx_ver=Z39.88-2004&" class="Z3988"><span style="display:none;">&nbsp;</span></span>
  • <span class="citation book">''Title'' (in Scottish Gaelic).</span><span title="ctx_ver=Z39.88-2004&" class="Z3988"><span style="display:none;">&nbsp;</span></span>
but if you copy them to an article in mainspace, click Show preview, you'll see that they do.
But, your question did make me think that we should be applying the same mainspace limitation to categorizing articles with |language=en and |language=English into Category:CS1 maint: English language specified. We aren't right now so I'll fix that.
Trappist the monk (talk) 19:31, 29 September 2014 (UTC)

Should Template:Cite report be listed on this Help page?[edit]

I recently came across an invalid LCCN in a {{Cite report}} template, which uses Citation/core, so it did not report an error message. Should {{Cite report}} be listed in the table on this Help page? – Jonesey95 (talk) 00:07, 5 September 2014 (UTC)

See the talk page. --  Gadget850 talk 01:19, 5 September 2014 (UTC)
Sorry, I must be dense. I looked at Template_talk:Cite_report but did not see anything related to my question. I'll explain more fully. It appears to me that {{cite report}} is a CS1 template, so I'm asking if {{cite report}} should be added to the table at Help:Citation_Style_1#General_use. – Jonesey95 (talk) 04:29, 5 September 2014 (UTC)
Yes. Done.
Trappist the monk (talk) 09:58, 5 September 2014 (UTC)
Then we need to change the CS1 description:

There are a number of templates that use a name starting with cite; many were developed independently of CS1 and are not compliant with the CS1 style. There are also a number of templates that use one of the general use templates as a meta-template to cite a specific source.

To be compliant with CS1, a template must:

  • Be based on {{citation/core}}, Module:Citation/CS1 or one of the templates listed below.
  • Use a period as a punctuation mark to separate fields and end the citation.
  • Use a semicolon as a punctuation mark to separate authors and editors.
  • Format longer works in italics.
  • Format short works such as chapters in quotes.
--  Gadget850 talk 10:31, 5 September 2014 (UTC)
All of those criteria are met, are they not, with the exception of title formatting? It isn't clear to me why the |title= value isn't formatted in the same way as other titles throughout Wikipedia. The title value passed to {{citation/core}} is |Title= ''&#xFEFF;{{{title|}}}&#xFEFF;'' which undoes normal italic formatting and adds zero-width no break space characters. Why undo the italic formatting and what purpose do the unicode characters serve?
Trappist the monk (talk) 10:57, 5 September 2014 (UTC)
Exactly. This is discussed on the talk page. The template is for "Unpublished reports by government departments, instrumentalities, operated companies, etc" which I also failed to understand. It is redundant to {{cite journal}}. --  Gadget850 talk 11:02, 5 September 2014 (UTC)
And the manner of usage is also odd. See Area 51 for example. --  Gadget850 talk 11:06, 5 September 2014 (UTC)
I don't think that the case has been well made for removing title styling contrary to MOS. As for use of {{cite report}} in Area 51, that looks like a case of profound laziness on the part of the editors. A few minutes in Google turned up pdf copies of both documents:
The U-2's Intended Successor: Project Oxcart,1956-1968
A-12, YF-12A, & SR-71 Timeline of Events
That second document doesn't have the pristine provenance of the first, but editors have been satisfied with less.
Trappist the monk (talk) 12:25, 5 September 2014 (UTC)

The documentation for {{cite report}} is so bad and the title formatting is so non-standard that this template should not be considered part of CS1. I consider the template unfit for any use at all and if I come across a featured article that uses it, I will change it. If the change is reverted I will challenge the FA status of the article. Jc3s5h (talk) 12:54, 5 September 2014 (UTC)

Thanks for adding the template to the table. Should {{cite report}} be converted to use the CS1 Module? It looks like that would clear up problems with title formatting, lack of availability of some parameters, and documentation. Are there problems that it would create? – Jonesey95 (talk) 13:59, 5 September 2014 (UTC)
Do you intend to keep the existing formatting? As best I recall, the only difference between {{cite report}} and {{cite journal}} is the title presentation and the 'docket' id (which may not be in use). --  Gadget850 talk 14:20, 5 September 2014 (UTC)
It would seem that {{cite report}} is in many ways similar to {{cite thesis}}. Both support |docket= and both are 'unpublished', and each sets |type= to a default value. Similarly, one might claim that {{cite report}} is similar to {{cite techreport}}; |type= set to default values and 'unpublished' in the sense expressed at {{cite report/doc}}. In both {{cite thesis}} and {{cite techreport}}, titles are italicized.
{{cite report |title=Title |id=ID |docket=Docketname}}Title (Report). Docketname.
{{cite thesis |title=Title |id=ID |docket=Docketname}}Title (Thesis). Docket Docketname. ID. 
{{cite techreport |title=Title |id=ID |docket=Docketname}}Title (Technical report). ID. 
And what does 'docket' mean in this context? The general definition (wikt:docket) isn't much help. Perhaps 'docket' is a contraction of 'docket number'? {{cite thesis}} adds the word 'Docket' to the value assigned to the module's internal variable ID when |docket= is used with {{cite thesis}}. Is there a real value gained from this?
Trappist the monk (talk) 15:26, 5 September 2014 (UTC)
Where are we on this? Either we change cite report (which will make it redundant), we change the description of CS1, or we leave cite report as a different style. --  Gadget850 talk 20:52, 10 September 2014 (UTC)
I'm of the opinion that {{cite report}} should be modified to put report titles in at least quotation marks or italics (my preference being the latter). The template is otherwise no more redundant than {{cite thesis}} and should be included in the CS1 family. Imzadi 1979  21:31, 10 September 2014 (UTC)
At this point I am presuming that report titles should be unformatted. I will be updating the documentation as such and redirecting the talk page to the CS1 centralized talk. --  Gadget850 talk 14:20, 30 September 2014 (UTC)

Untitled work[edit]

I notice that CS1 does not support untitled works. Chicago would use a descriptive phrase where the title would normally go, and the phrase would not be in italics nor would it be enclosed with quotes. So if someone needs to cite an untitled work, what would they do, other than rewrite all the citations in the article to use a style other than CS1? Jc3s5h (talk) 13:26, 5 September 2014 (UTC)

Example? In my mind, untitled would mean unpublished, which would fail the reliability standard. --  Gadget850 talk 13:34, 5 September 2014 (UTC)
I will give an example from Chicago Manual of Style 16th ed. section 14.240:

42. Burton to Merriam, telegram, 26 January 1923, Charles E. Merriam Papers, University of Chicago Library.

Jc3s5h (talk) 13:46, 5 September 2014 (UTC)
That's arguably an unpublished source, per Gadget850's concerns, above, though some might consider it a primary source because it's publicly accessible and validated by the institution. While it's conventional in art-related publishing to treat untitled works as if their title was "Untitled" or Untitled, as appropriate, those are usually also unpublished works. The most common case for this where we'd care about it are small newspaper entries with no title, and these are actually quite common, or were, back in the day. I think the most common way of handling them has been to use their first few words and "..." as if it were a title: |title=Yesterday's Spinks vs. Schaefer match was... |at="Sports" section |department=Sports |work=[[New York Times]].... A |noquotes= parameter could be used to remove quotation marks from around descriptive entries that are not actually titles, per Chicago's usage (and properly documented as to why this parameter would ever be used). A corresponding |noitalics= could also be used, for even rarer cases, to deitalicize names of major works. Actually, the most common use for both cases might be where the title needs to ben manually given in the proper formatting and also be followed by an annotation that should not be quoted/italicized, e.g. |noitalics=y |title=''Serenity'' (original script) |at="Extras" section |work=[[Serenity (film)|Serenity]] |format=Blu-ray .... There are many cases of not-intended-to-be-published works which have subsequently been published as part of bigger collective ones, and which have no titles or have titles, have been assigned later, conventional names in critical lit that are not actually titles, or have titles that would be ambiguous or confusing without such notes. An obvious example is the enormous amount of draft and abandoned and incomplete work by J. R. R. Tolkien edited and published in collected volumes by his son, which includes examples of all three types of cases.  — SMcCandlish ¢ ≽ʌⱷ҅ʌ≼  14:36, 6 September 2014 (UTC)
Perhaps something like |description-in-lieu-of-title= (yeah, much too long as a parameter name but could be aliased to an initialism |dilot=). If |title= or |chapter= (or any of the appropriate aliases) is present in the template and has an assigned value, do not display |dilot=; else display the unstyled, unquoted content of |dilot=. Content of |dilot= would not be restricted; |url=, if present, attaches to |dilot= just as it does to |title=. CS1 templates that display |dilot= content would add the page to a maintenance category; it would be inappropriate to misuse this parameter for the purpose of affecting title style in a rendered citation.
Trappist the monk (talk) 16:01, 6 September 2014 (UTC)
But |title= is already mandatory. So, the simple way to get at this is to allow styling (quotes or italics) to be decoupled from that parameter when necessary. That's way simpler than that "dilot" stuff, the name of which no one is likely to remember. The real code to implement how you've described the behavior of such as |dilot= would be quite bloated; it's a bunch of nested ifs, and ifs in multiple sections of essentially unrelated code, etc. Let's go with the KISS principle on this.  — SMcCandlish ¢ ≽ʌⱷ҅ʌ≼  18:52, 6 September 2014 (UTC)
@SMcCandlish: Instead of |at="Sports" section I would recommend |department=Sports: "Regular department within the periodical. Displays after work and is in plain text." --  Gadget850 talk 16:32, 6 September 2014 (UTC)
Noted; that parameter must have been added since the last time I pored over the parameters list in any detail.  — SMcCandlish ¢ ≽ʌⱷ҅ʌ≼  18:52, 6 September 2014 (UTC)
I have seen sources that incorporated a bunch of other documents (usually memos, letters, or messages) that don't have titles. Should we establish some kind of recommended practice that such items be explicitly "titled" with something like "[untitled]"? I think the strongest reason for that is so that the lack of a title is acknowledged, rather than seeming like a possible omission. ~ J. Johnson (JJ) (talk) 20:58, 29 September 2014 (UTC)
The style manuals I've read use a description of such works, but don't give any special typographic treatment to the description (no quote marks, no italics) so readers will know it is a description rather than a title. Jc3s5h (talk) 21:12, 29 September 2014 (UTC)

Peculiar title formatting[edit]

When all three of |title=, |chapter=, and |work= (or their aliases) are used together in a CS1 citation, the normally (at least to me) expected formatting shifts. For example:

{{cite book |title=Title |chapter=Chapter}}"Chapter". Title. 

But when I add |work=, this:

{{cite book |title=Title |chapter=Chapter |work=Work}}"Chapter". Title. Work. 

If I take away |title=, this:

{{cite book |chapter=Chapter |work=Work}}"Chapter". Work. 

If I take away |chapter=, this:

{{cite book |title=Title |work=Work}}Title. Work. 

I suppose that these last two are correct if you assume that |title= or |chapter= are 'smaller' or subsidiary portions of |work=.

This is a new 'feature' to CS1 if you believe this comparison:

Cite book compare
{{ cite book | chapter=Chapter | title=Title | work=Work }}
Old "Chapter". Title.
Live "Chapter". Title. Work. 

Is the module's current rendering of citations that use all three |title=, |chapter=, and |work= correct?

Trappist the monk (talk) 17:38, 12 September 2014 (UTC)

That looks wrong to me. I think the value of |chapter= should always be in quotation marks, and the value of |title= should be italicized, at least in {{cite book}}. Under what circumstances would one want to use both |title= and |work=? Shouldn't they be aliases, as with |journal= and |work= in {{cite journal}} and {{cite news}}? The goal is to cite a chapter/article (in quotation marks) within a larger work (in italics), I think. What is the point of having three distinct parameters?
I notice that |work= is not a documented parameter in the documentation for {{cite book}}.
If three distinct parameters are needed, I think the display should be "Chapter". Title. Work. – Jonesey95 (talk) 20:30, 12 September 2014 (UTC)
|work= is a generic parameter name for |journal=, |encyclopedia=, |dictionary=, etc. This is probably a good example of the dark side of Module:Citation/CS1 supporting every parameter regardless of template name. Here are {{cite journal}}, {{cite encyclopedia}}, {{cite AV media}} which all give the same result:
{{cite journal |title=Title |chapter=Chapter |journal=Journal}}"Chapter". "Title". Journal. 
{{cite encyclopedia |title=Title |chapter=Chapter |encyclopedia=Encyclopedia}}"Chapter". Title. Encyclopedia. 
{{cite AV media |title=Title |chapter=Chapter |work=Work}}"Chapter". Title. Work. 
I think I agree with you that when this case exists, it should be as you described.
Trappist the monk (talk) 21:45, 12 September 2014 (UTC)

It works as documented, but this is because I documented it as it works:

  • work: Name of the source periodical; may be wikilinked if relevant. Displays in italics. Aliases: journal, newspaper, magazine, periodical.
    • issue: When the publication is one of a series that is published periodically. Alias: number.
When set, work changes the formatting of other parameters:
title is not italicized and is enclosed in quotes.
chapter is italicized and is not enclosed in quotes.
location and publisher are enclosed in parentheses.
page and pages do not show p. or pp.
edition does not display.
type does not display.

No one has ever been able to explain the formatting difference. I don't know why we even support chapter for journal and news. --  Gadget850 talk 00:49, 13 September 2014 (UTC)

In the code, the quote / italicize determination works this way:
if |work= is used then |title= is quoted.
if the citation is {{cite web}}, {{cite news}}, {{cite pressrelease}}, {{cite conference}}, or {{cite podcast}}; and |chapter= is not used then |title= is quoted.
|title= is italicized.
There is a similar construct for |chapter=:
if both |work= and |title= are used then |chapter= is italicized.
|chapter= is quoted.
I think that this is wrong and should be revised so that |chapter= is always quoted and (ignoring non-Latin languages for the time being) |title= is always italicized except when the citation is {{cite journal}}, {{cite web}}, {{cite news}}, {{cite press release}}, {{cite conference}}, or {{cite podcast}} in which cases |title= is quoted |chapter=, if used, is ignored. Further, |chapter= without |title= should cause a missing title error.
Trappist the monk (talk) 15:32, 14 September 2014 (UTC)
An addendum: |title= should not be italicized in {{cite journal}}. In those citations, |title= is always used for the title of the article within the larger work, at least in my experience of correcting citation errors in a zillion {{cite journal}} templates. |journal= or |work=, in italics, is used for the name of the larger work.
What about |title= in {{cite encyclopedia}}? I know that one is particularly complex.
I think the last suggestion about chapter without title may have unintended consequences and may merit a separate discussion. – Jonesey95 (talk) 15:39, 14 September 2014 (UTC)
Yep, missed that one; {{cite journal}} added. Did I miss any others? I can imagine that there might be encyclopediae with articles long enough to subdivide into chapter-like sections. |section= is an alias of |chapter= so the rule as I've proposed it would render something like this:
"Section Title". Article Title. Encyclopedia
But, the problem with this or any citation that uses the three parameters |work=, |title=, and |chapter=, is that |work= (which is |encyclopedia= in {{cite encyclopedia}}) doesn't make it into the COinS metadata so we've lost the most important identifier.
Trappist the monk (talk) 16:21, 14 September 2014 (UTC)

I've adjusted Module:Citation/CS1/sandbox to demonstrate how I think citations with |work=, |title=, and |chapter= should render:

Cite book compare
{{ cite book | chapter=Chapter | title=Title | work=Work }}
Live "Chapter". Title. Work. 
Sandbox "Chapter". Title. Work. 
Cite journal compare
{{ cite journal | chapter=Chapter | title=Title | work=Work }}
Live "Chapter". "Title". Work. 
Sandbox "Chapter". "Title". Work. 
Cite encyclopedia compare
{{ cite encyclopedia | chapter=Chapter | title=Title | encyclopedia=Encyclopedia }}
Live "Chapter". Title. Encyclopedia. 
Sandbox "Chapter". Title. Encyclopedia. 

Note that in the ({{cite journal}} example, |chapter= is ignored.

I found an undocumented snippet of code that maps |title= to |chapter= if |work= is set and |chapter= is not set. I have disabled that in the sandbox:

Cite book compare
{{ cite book | work=Work | title=Title }}
Live Title. Work. 
Sandbox Title. Work. 

I suspect it is there to force the title to render in quotes. If we adopt this change, I will leave it in the code, though disabled, in case it serves some other purpose that I haven't managed to noodle-out.

Trappist the monk (talk) 14:39, 16 September 2014 (UTC)

These all look right to me. Do we want to flag |chapter= in {{cite journal}} as an error of some sort, rather than silently discarding potentially useful information that has heretofore been displayed? – Jonesey95 (talk) 14:45, 16 September 2014 (UTC)
In the sandbox, |chapter= will be ignored by {{cite journal}}, {{cite web}}, {{cite news}}, {{cite press release}}, {{cite conference}}, and {{cite podcast}} when |title= and |work= are both set. For {{cite journal}}, {{cite web}}, {{cite news}}, and {{cite podcast}} this makes sense because |work= in those citation types is a journal or magazine, a newspaper, or a website. For {{cite press release}}, the quoted title is appropriate to the relative size of a press release. We should just get rid of {{cite conference}}. Most often, {{cite conference}} is used when {{cite journal}} or {{cite book}} or even {{cite speech}} would be more appropriate.
Certainly we should update the documentation for these particular templates to say that |chapter= is ignored. It isn't really an error to have a valid parameter that doesn't get used. Any reason that we shouldn't have |chapter= requires |title= (Help) when |chapter= is used without |title=?
Trappist the monk (talk) 15:24, 16 September 2014 (UTC)
In the past, the test for Missing or empty |title= looks for a value in any one of |chapter=, |title=, |work=, |conference=, |trans-title=, or |trans-chapter=. In the sandbox, I have removed the tests for |chapter= and |trans-chapter=. Either of these, without one or more of the others, is somewhat meaningless: |chapter=3. xxxx of what? With this change, I don't see a need for |chapter= requires |title=.
Cite book compare
{{ cite book | chapter=3. xxxx }}
Live "3. xxxx".  Missing or empty |title= (help)
Sandbox "3. xxxx".  Missing or empty |title= (help)
Trappist the monk (talk) 14:15, 24 September 2014 (UTC)
That makes sense to me. I have been trying to think of valid edge cases, but I haven't come up with any. – Jonesey95 (talk) 19:59, 24 September 2014 (UTC)

archivedate considered harmful[edit]

(in the tradition of Goto considered harmful)

The existence of Help:CS1_errors#.7Carchiveurl.3D_requires_.7Carchivedate.3D makes little sense to me. 99%[1] of archiveurl fields on en contain the archive date within the URL, so the archivedate field is redundant, and where the archivedate field conflicts with the archive date within the URL, it's always the former that is wrong. I therefore proclaim archivedate considered harmful, and propose it be deprecated. --{{U|Elvey}} (tc) 08:09, 17 September 2014 (UTC)

  1. ^ (guesstimate)
Cute, but... MOS:DATEUNIFY, even ignoring those archiveurls which don't "point to" the date. Can you make the template rendering of the date from the archiveurl context sensitive, so it matches the other dates in the article? LeadSongDog come howl! 13:38, 17 September 2014 (UTC)
I can't tell if the OP is joking or not. If so, the sarcasm is subtle.
URLs from typically have their archive dates in the URL, but those from some other archiving sites, like, do not (see, for example, this reference). In any event, someone who does not know how to decode the date in an URL (which requires hovering over the link with a mouse and reading the URL in the browser's status bar, if it is being displayed) is out of luck unless a human or bot editor converts it to a human-readable archive date. – Jonesey95 (talk) 16:05, 17 September 2014 (UTC)
For the Owen Edwards example, see [1] to see how timestamps are available. That raises the point though... couldn't a bot check archiveurl strings from major archives for matching archivedate values, and put them in the format appropriate for the article? LeadSongDog come howl! 18:33, 17 September 2014 (UTC)
I think so, yes... I meant to include a sentence in my OP suggesting that CS1 render the date from the archiveurl. I think CS1 could be made to match the other dates in the article; I think dates and templates like {{Use dmy dates}} and {{Use mdy dates}} can be detected too, but I'm not sure. --{{U|Elvey}} (tc) 21:26, 20 September 2014 (UTC)

────────────────────────────────────────────────────────────────────────────────────────────────────FYI—The other day I ran across several URLs that had bogus values in the date part; e.g., "15" in the location where the month is supposed to be. The URLs were functional, but IIRC they redirected to "valid" URLs. There is no way of knowing how widespread the problem is, but you can't rely on those values 100% of the time. ‑‑Mandruss (talk) 13:08, 27 September 2014 (UTC)

Odd. Can you link to an example of where you found one on-wiki?--{{U|Elvey}} (tc) 00:54, 2 October 2014 (UTC)

Wikimarkup and COinS metadata[edit]

One very common corrupter of the COinS metadata is wikimarkup when it is used to add bold and / or italic styling to |title= (or to undo the default italic styling). The problem arises because the markup gets copied into the metadata. In this example, all of the apostrophes used to make the title bold italic appear in the metadata as a string of %27 which are not part of the actual title:

  • '''''Bold italic''''' title → "Bold italic title". :
  • <span class="citation news">"'''''Bold italic title'''''".</span><span title="ctx_ver=Z39.88-2004&" class="Z3988"><span style="display:none;">&nbsp;</span></span>

In Module:Citation/CS1/sandbox, I have created a function get_coins_title(), that strips the most common legitimate markup from |title= before adding the title to the metadata:

the five-apostrophe cases
  • '''''Bold italic''''' title → "Bold italic title". :
  • <span class="citation news">"'''''Bold italic title'''''".</span><span title="ctx_ver=Z39.88-2004&" class="Z3988"><span style="display:none;">&nbsp;</span></span>
  • '''''Bold''' italic'' title → "Bold italic title". :
  • <span class="citation news">"'''''Bold''' italic'' title".</span><span title="ctx_ver=Z39.88-2004&" class="Z3988"><span style="display:none;">&nbsp;</span></span>
  • '''''Italic'' bold''' title → "Bold italic title". :
  • <span class="citation news">"'''''Bold'' italic''' title".</span><span title="ctx_ver=Z39.88-2004&" class="Z3988"><span style="display:none;">&nbsp;</span></span>
  • ''Italic '''bold''''' title → "Italic bold title". :
  • <span class="citation news">"''Italic '''bold''''' title".</span><span title="ctx_ver=Z39.88-2004&" class="Z3988"><span style="display:none;">&nbsp;</span></span>
  • '''Bold ''italic''''' title → "Bold italic title". :
  • <span class="citation news">"'''Bold ''italic''''' title".</span><span title="ctx_ver=Z39.88-2004&" class="Z3988"><span style="display:none;">&nbsp;</span></span>
simple bold
  • '''Bold''' title → "Bold title". :
  • <span class="citation news">"'''Bold''' title".</span><span title="ctx_ver=Z39.88-2004&" class="Z3988"><span style="display:none;">&nbsp;</span></span>
simple italics
  • ''Italic'' title → "Italic title". :
  • <span class="citation news">"''Italic'' title".</span><span title="ctx_ver=Z39.88-2004&" class="Z3988"><span style="display:none;">&nbsp;</span></span>
a combination of all three
  • '''''Bold''' italic'' title and ''italic'', '''bold''', '''bold again''', last ''italic'' → "Bold italic title and italic, bold, bold again, last italic". :
  • <span class="citation news">"'''''Bold''' italic'' title and ''italic'', '''bold''', '''bold again''', last ''italic''".</span><span title="ctx_ver=Z39.88-2004&" class="Z3988"><span style="display:none;">&nbsp;</span></span>

The above examples use {{cite news}} because that template renders titles in upright font.

Have I missed anything obvious?

With this fix, the work discussed at Module talk:Citation/CS1#non-italic titles may no longer be necessary.

Trappist the monk (talk) 12:46, 18 September 2014 (UTC)

Renamed function to strip_apostrophe_markup() and now also used to strip wikimarkup from |chapter=.
Trappist the monk (talk) 10:52, 19 September 2014 (UTC)
I don't know what COinS expects for data, or how resilient it is in the face of obviously bad data, but I would think that any kind of markup that is purely for display purposes is not proper data. Perhaps this is part of a larger problem, of editors trying to coerce the templates? Perhaps that should be flagged as an error? ~ J. Johnson (JJ) (talk) 20:40, 29 September 2014 (UTC)


At {{asin}} is this note:

If the ASIN begins with a number, it is a standard ISBN; please use "ISBN xxxxxxxxxx" instead of ASIN xxxxxxxxxx, as this will allow us to link to other sites as well as Amazon.

Using |isbn= links to Special:BookSources where there is some hope of finding information about the book so we aren't simply feeding readers to

We should add this to the documentation for |asin= shouldn't we? And that made me wonder if we shouldn't just treat such |asin= values as if they were |isbn=. Obviously we can't treat the alphanumeric asins as isbns so those would still render as they do now.

Here are a couple of (not very good) example citations. The first uses |asin=4081097011 and |

"幽☆遊☆白書 其之一 (1) 霊界死闘 編(SHUEISHA JUMP REMIX) (単行本)" [Yū Yū Hakusho (1) Spiritual Guide: Deathmatch (SHUEISHA JUMP REMIX) (Paperback)] (in Japanese). ASIN 4081097011. Retrieved September 2, 2013. 

The second uses |isbn=4081097011:

"幽☆遊☆白書 其之一 (1) 霊界死闘 編(SHUEISHA JUMP REMIX) (単行本)" [Yū Yū Hakusho (1) Spiritual Guide: Deathmatch (SHUEISHA JUMP REMIX) (Paperback)] (in Japanese). ISBN 4081097011. Retrieved September 2, 2013. 

The documentation for |asin-tld= needs to be updated to list or refer to all of the ccTLDs that can be found at

Trappist the monk (talk) 21:01, 21 September 2014 (UTC)

I was looking at this last week.
The Amazon ASIN should be only used for Amazon unique products. I know it is used quite a bit for video games published in Japan; I added the TLD codes to the CS1 core at the request of WikiProject Video games. But a quick sample shows that {{ASIN}} is often mis-used as a citation template; see The Highway Code for example. This lazy use needs to be replaced with a full citation template where it fits the established style.
The CS1 templates support ca, cn,,, de, es, fr, it, so we need to add (China), (India), (Mexico), (Australia) and (Brazil). --  Gadget850 talk 21:36, 24 September 2014 (UTC)
India is already supported. I don't see any reason to do anything about since it gets mapped to I have added support in the sandbox for Australia, Brazil, and Mexico.
  • Chandra, Bipan. History of Modern India. ASIN 8125036849.  – India already supported; this from the live version of the module
  • Corris, Peter. Silent Kill. ASIN B00G65FW0Q.  – Auatralia
  • Follett, Ken. Eternidade por um fio: Terceiro livro da trilogia O Século. ASIN B00KI2HDCS.  – Brazil
  • Jakovlevs, Valerijs. Mazatlan Travel Guide. ASIN B005JFBRFS.  – Mexico
Still, my main question remains unanswered. When |asin= is a number, that number is an ISBN. Should we treat it as an ISBN and link it to Special:BookSources?
Trappist the monk (talk) 22:53, 24 September 2014 (UTC)
I would rather see a bot or an AWB user convert ASINs that are ISBNs into ISBN parameters. That would follow the principle of linking to neutral sources for references instead of to specific vendors (see also WP:ADV). Linking an ASIN to Special:Booksources would be contrary to the principle of least surprise for readers. Converting ASINs to ISBNs would also allow us to validate those ISBNs and flag errors. – Jonesey95 (talk) 23:22, 24 September 2014 (UTC)
Point. So perhaps I'll make us a CS1 maintenance category and add code to categorize citations that use all numeric values for |asin=.
Trappist the monk (talk) 00:15, 25 September 2014 (UTC)
Don't forget that not all ISBNs are all-numeric: nine digits followed by the letter "X" is a valid ISBN format. --Redrose64 (talk) 07:05, 25 September 2014 (UTC)
Thanks, not forgotten, I merely misspoke. In Module:Citation/CS1/sandbox, the value in |asin= is passed to the code that validates ISBNs. If the asin value is a legitimate ISBN, the page is added to category Category:CS1 maint: ASIN uses ISBN.
Trappist the monk (talk) 14:26, 25 September 2014 (UTC)

Accessibility and COinS[edit]

There may be an accessibility issue with the COinS metadata that is appended to citations emitted by the {{citation}} template as well as all the Citation Style 1 templates. See Wikipedia:Village pump (technical)/Archive 130#Spurious text on source links. --Redrose64 (talk) 22:59, 24 September 2014 (UTC)


I have created some new categories. Category:CS1, Category:CS1 errors, and Category:CS1 maintenance. I have added Category:CS1 errors to all of the current error categories listed at Category:Articles with incorrect citation syntax with the exception of Category:Pages with DOI errors and Category:Pages with OL errors. Because these two categories are also populated by non-CS1 templates, I have modified Module:Citation/CS1/sandbox so that errors that previously categorized pages into these two categories will use two new categories Category:CS1 errors: DOI and Category:CS1 errors: OL when the module is next updated.

First use for Category:CS1 maintenance is likely to be to categorize citations that use |asin= where the assigned value is an ISBN. (see Help talk:Citation Style 1#Asin) Another use is to monitor pages that use |script-title=. (see Module talk:Citation/CS1#non-italic titles and Module talk:Citation/CS1#rtl language support in CS1 titles)

I plan to add a properties category which will get as its first subcategory something like Category:CS1 foreign language citations which will then list all pages that have CS1 citations that use |language=. This so we don't continue to improperly (I think) add pages to Category:Articles with non-English-language external links which was intended for other purposes. The conversation about this (such as it has been) is at Help talk:Citation Style 1#An ISO 639-1 language name test.

Trappist the monk (talk) 11:59, 25 September 2014 (UTC)

All of this is useful. I can imagine a bot checking the foreign language categories for various errors, like misspelled languages and |language=English. The maintenance categories could also be used to highlight other conditions that are not errors, but where improvements could be made, like the templates in |id= that Trappist the monk has recently been editing to use the dedicated parameters. – Jonesey95 (talk) 13:10, 25 September 2014 (UTC)
I have added a new maintenance category Category:CS1 maint: English language specified and supporting code in Module:Citation/CS1/sandbox that will categorize pages with citations containing |language=English or |language=en. These parameters are redundant in the English Language Wikipedia.
Trappist the monk (talk) 14:55, 26 September 2014 (UTC)
I wonder what sort of results we would get from adding "unrecognized" languages to a maintenance category. In other words, if the value of the language parameter does not match a list of known languages, we flag it to be checked. It should not be put in an error category, since some of the values like "English and French" might be OK, but I suspect there are misspelled languages or invalid two-letter codes out there. – Jonesey95 (talk) 21:35, 26 September 2014 (UTC)
Keep that thought in mind. The list of changes this go-round is getting rather long. I'd like to finish up on what has been started, update the live module, and then think about new stuff (and old stuff that I have been neglecting).
Trappist the monk (talk) 22:21, 26 September 2014 (UTC)
I'll keep an eye on the new Category:CS1 maint: English language specified - AWB should be able to fix these. GoingBatty (talk) 23:13, 26 September 2014 (UTC)

Updating the live module?[edit]

moved from Help talk:Citation Style 1#Categories

I support updating the live module. Has the missing editor/author condition been re-implemented in this round of updates? Also, do we feel that there is consensus on enabling any of the hidden error messages? (See discussion above.) If this comment is forking the discussion, feel free to move it to a new section or subsection about updating the module. – Jonesey95 (talk) 03:40, 27 September 2014 (UTC)

Missing name detection will be part of the next update.
I don't see much hew and cry about against unhiding date errors so I have set the bad date error hidden parameter to false.
Trappist the monk (talk) 10:49, 27 September 2014 (UTC)

Update to the live CS1 module weekend of 11–12 October 2014[edit]

Over the 11–12 October 2014 weekend I propose to update:

Module:Citation/CS1 to match Module:Citation/CS1/sandbox (diff)
Module:Citation/CS1/Configuration to match Module:Citation/CS1/Configuration/sandbox (diff)
Module:Citation/CS1/Whitelist to match Module:Citation/CS1/Whitelist/sandbox (diff)
Module:Citation/CS1/Date validation to match Module:Citation/CS1/Date validation/sandbox (diff)

This update changes these things:

in Module:Citation/CS1:

  1. Bug fix in get_coins_pages(); (discussion)
  2. Bug fix in extractnames(); (discussion)
  3. Change lccn() to require lower case alpha characters; (discussion)
  4. Change openlibrary() to emit error message when OL identifier contains leading alpha characters; (discussion)
  5. Change |language= support (discussion):
    1. get ISO639-1 language name from Wikimedia;
    2. Add support for right-to-left languages using new parameter |script-title=; (discussion) and (discussion)
    3. Categorize pages when |language=language name;
  6. Change listpeople() so that we don't link to the current page through |authorlinkn= or |editorlinkn=; (discussion)
  7. Add code to strip wikimarkup (italics and bold) from titles and chapters when adding those to COinS metadata; (discussion)
  8. Add Australia, Brazil, Mexico to list of countries supported by |asn-tld=; (discussion)
  9. Undo peculiar title and chapter format swap when |work= or any of its aliases set; (discussion)
  10. Categories:
    1. Add support for maintenance categories; discussion
    2. Change amazon() to add maintenance category when |asin= value is an isbn; discussion
    3. Add support for properties categories; (discussion)

in Module:Citation/CS1/Configuration:

  1. Updated JFM and ZBL prefixes; (discussion)
  2. deleted ISO639-1 table; (discussion)
  3. Changed error categories for doi and ol errors (discussion)
  4. Make date errors visible; (discussion)

in Module:Citation/CS1/Whitelist:

  1. add new parameter |script-title= (discussion) and (discussion)

in Module:Citation/CS1/Date validation:

  1. Add support for valid date formats Summer yyyy–yy and Summer yyyy–yyyy; (discussion)

There is enough here that there is a deal of documentaion to be done. I think that I'll begin that and not bother to hide it prior to the live update – it's just easier that way.

Trappist the monk (talk) 16:18, 30 September 2014 (UTC)

I think this update will also re-enable the missing author/editor errors. And wasn't there something about creating an error when |chapter= exists when there is no |title= or |work=? (See "I have removed the tests for" above.) – Jonesey95 (talk) 16:52, 30 September 2014 (UTC)
Missing author/editor is the extractnames() bug (#2); |chapter= is Undo peculiar title ... (#9).
Trappist the monk (talk) 17:25, 30 September 2014 (UTC)


Trappist the monk (talk) 14:20, 11 October 2014 (UTC)


Could the value of |script-title= not be underlined? Kanguole 17:51, 30 September 2014 (UTC)

Agree. The proper name mark is a Chinese convention that would not apply to other writing systems such as Hebrew or Arabic. --  Gadget850 talk 18:12, 30 September 2014 (UTC)
and an obsolete Chinese convention at that. But these are conventions for Chinese running text, which isn't the situation here. Kanguole 20:26, 30 September 2014 (UTC)
Looks like that was removed.
  • {{cite book/new |title=ABC |script-title=ar:العربية}}
ABC العربية. 


script-title: Title in the original writing system where it is not appropriate to italicize (e.g. Arabic, Chinese, Hebrew, Japanese). Displays after title in upright font and is isolated from the surrounding text-direction settings so that right-to-left scripts render properly. The language may be set by prefixing the value with the ISO 639-1 two-character language code followed by a colon. Unrecognized codes are ignored and will display in the rendered citation. Example: |script-title=ar:العربية.

--  Gadget850 talk 18:05, 1 October 2014 (UTC)

I'd suggest using Chinese or Japanese as the example language, as those are the two for which CMOS recommends also including the original form. Kanguole 20:50, 1 October 2014 (UTC)
Sample? --  Gadget850 talk 12:56, 2 October 2014 (UTC)
  • Wang, Li (1985). Hànyǔ Yǔyīn Shǐ 汉语语音史 [History of Chinese Phonetics] (in Chinese). Beijing: China Social Sciences Press. ISBN 978-7-100-05390-7. 
  • Morohashi, Tetsuji (1984–1986). Dai Kan-Wa Jiten 大漢和辞典 [Comprehensive Chinese–Japanese Dictionary]. Tokyo: Taishukan. 
Kanguole 00:45, 19 October 2014 (UTC)
Arabic is a good example because it is both non-Latin and rtl, and because support for rtl is a substantial reason for the existence of |script-title=.
Trappist the monk (talk) 13:15, 2 October 2014 (UTC)
Arabic is an unfortunate example, because it's one of the languages that CMOS recommends be transliterated, whereas Chinese and Japanese are the languages where it recommends using the original form in addition to romanization. Kanguole 00:45, 19 October 2014 (UTC)

Error checking[edit]

Unrecognized codes are not ignored:

  • {{cite book/new |title=ABC |script-title=zz:العربية}}
ABC zz:العربية. 

More than two characters or other than alpha characters do show the language code. --  Gadget850 talk 14:25, 2 October 2014 (UTC)

Thanks for that. Fixed.
Trappist the monk (talk) 14:38, 2 October 2014 (UTC)

Related discussions[edit]

See Template talk:Lang#Rtl-lang in citation titles. --  Gadget850 talk 12:49, 3 October 2014 (UTC)

Titles of journal articles[edit]

Titles of journal articles used to be in roman, wrapped in double quotes, as in {{cite journal}}:

  • {{cite journal | author = Author | year = 2000 | title = Article title | journal = Journal | ref = none }}
  • Author (2000). "Article title". Journal. 

but now in {{citation}} they're in italics:

  • {{citation | author = Author | year = 2000 | title = Article title | journal = Journal | ref = none }}
  • Author (2000), Article title, Journal 

However {{cite journal}} still does it the old way. Kanguole 00:56, 19 October 2014 (UTC)

It does seem to have changed, yes. This might be the same issue as Template talk:Citation#Formatting of journal article_title. --Redrose64 (talk) 08:39, 19 October 2014 (UTC)
Oversight on my part when I undid the peculiar title formatting. Fixed in the sandbox so that now if the using {{citation}} with one of the |work= aliases then |title= is upright and quoted.
Author (2000), "Article title", Journal 
Trappist the monk (talk) 10:07, 19 October 2014 (UTC)

Date when no date supplied[edit]

This webpage does not list a date, but does state the copyright, "© 2009-2014 Concordia Theological Seminary", at the bottom. Should I leave |date= blank or use |date=2009-2014? Thanks! - Location (talk) 23:16, 1 October 2014 (UTC)

Trappist the monk (talk) 13:16, 2 October 2014 (UTC)

Date for ArXiv[edit]

I'd like to be able to add a separate date for the ArXiv link specified with {{{arxiv}}} in {{cite journal}} and related templates. For those not familiar with this, ArXiv is a site where academic papers (I think overwhelmingly scientific) can be uploaded for free access and peer review before and while being submitted to regular journals. The ArXiv version remains (freely!) available regardless of subsequent publication elsewhere, and the date of submission can be very different from that of eventual publication. I know that {{{orig-year}}} is available but it doesn't mean the same thing. I would therefore like to add {{{arxiv-date}}} to allow this distinction to be displayed. HTH HAND —Phil | Talk 17:48, 3 October 2014 (UTC)

  • publication-date: Date of publication when different from the date the work was written. Displays only if year or date are defined and only if different, else publication-date is used and displayed as date. Use the same format as other dates in the article; do not wikilink. Follows publisher; if work is not defined, then publication-date is preceded by "published" and enclosed in parenthesis.
--  Gadget850 talk 19:21, 3 October 2014 (UTC)
If you want to specify a version of the arxiv link, include the v1, v2, etc... part to the arxiv identifier and that's all. If you want to cite the arxiv link specifically, use {{cite arxiv}}. Headbomb {talk / contribs / physics / books} 20:18, 3 October 2014 (UTC)

Date template[edit]

Just curious - why doesn't the {{date}} template generate a CS1 error, even though the template's documentation says not to use it within citation tempates? For example:

  • {{cite web|url=|title=test|accessdate={{date|22 mar 2013}}}} generates
  • "test". Retrieved 22 March 2013. 

Thanks! GoingBatty (talk) 15:29, 5 October 2014 (UTC)

Umm, you added {{COinS safe|n}}, perhaps you should tell us?
Trappist the monk (talk) 15:47, 5 October 2014 (UTC)
Self-reverted my edit to the documentation. GoingBatty (talk) 15:51, 5 October 2014 (UTC)
I don't see any issues with {{date}} except for the template overhead. But I really don't see the use unless you are importing citation data. --  Gadget850 talk 16:02, 5 October 2014 (UTC)
I always use {{date}} in the hope that some time in the future it will be recognised by the renderer, and show date in user's preferred style and language. Overhead, schmoverhead. John of Cromer (talk) mytime= Mon 17:41, wikitime= 09:41, 6 October 2014 (UTC)
If the MediaWiki parser ever gets that feature we will add support to the templates. In that case, {{date}} will probably interfere. --  Gadget850 talk 12:27, 6 October 2014 (UTC)
I think it's unlikely that User:Johnmperry's hope for the renderer to show dates in the user's preferred style and language will ever be fulfilled. During the extensive debates about date linking, it was shown that, in running text, it is impossible to automatically change from the dmy format to the mdy fomat or vice versa without causing usage errors, because the former does not use commas and the latter usually uses 2 commas. So the renderer would have to limit its activity to tables and other structured text which might be predictable enough to automate the change in rendering. Any such change would clash with the unchanged dates in the running text. Jc3s5h (talk) 16:40, 6 October 2014 (UTC)

Author alias[edit]

I have a work by someone with an English name and a Chinese name. What to do?

What I did was put his English name as author, and his alias as others= alias....

Anyone got a better idea? John of Cromer (talk) mytime= Mon 17:37, wikitime= 09:37, 6 October 2014 (UTC)

Translated title error fails to display red error message[edit]

I came across an article with the following citation, which caused the article to be categorized in Category:Pages with citations using translated terms without the original but did not display a red error message.

Citation example
Markup Renders as
{{cite journal | author=Staff |date=5 June 1998|trans_chapter=Coxon, Sachi | title=インタビュー 坂口 博信 |language=Japanese |trans_title=Interview with Hironobu Sakaguchi | url= | journal=[[Famitsu Weekly]] | accessdate=2006-07-15}} 
Staff (5 June 1998). [Coxon, Sachi] |trans_chapter= requires |chapter= (help). "インタビュー 坂口 博信" [Interview with Hironobu Sakaguchi]. Famitsu Weekly (in Japanese). Retrieved 2006-07-15. 

Any ideas about why the error message did not display? I expect that the problem is related to the code changes that just went into effect, since the category has been empty for a long time and the article had not been edited recently. – Jonesey95 (talk) 02:05, 12 October 2014 (UTC)

The category is added because of the unmatched |trans_chapter=Coxon, Sachi (which should be |others=Trans. Sachi Coxon) but it's also quashing the url. I'll look into it in the morning.
Trappist the monk (talk) 02:47, 12 October 2014 (UTC)
When we accepted the notion that |chapter= is inappropriate in {{cite web}}, {{cite news}}, {{cite journal}}, {{cite press release}}, {{cite conference}}, and {{cite podcast}}, I didn't get the code right. When |trans-chapter= doesn't have a matching |chapter=, an error message is created and the appropriate category is added to the list of error categories. Metaparameter Chapter is a concatenation of |chapter= (an empty string in this case) and |trans-chapter=. The Chapter value is then made into an external link with either |chapter-url= or |url=. Next, the error message is tacked onto the end of Chapter so we get this:
[ Coxon, Sachi] |trans_chapter= requires |chapter= (help).
But, we then look at the citation type. If the citation type is one of those I listed above, we set Chapter to an empty string so neither chapter nor error message is in the rendered citation.
I have tweaked Module:Citation/CS1 so that the error message is restored and will spend some time pondering the issue.
Trappist the monk (talk) 14:14, 12 October 2014 (UTC)
In your pondering, you might consider the case of an erroneous |accessdate=, for which the error message displays even if the accessdate is hidden because there is no value in |url=. It seems like a similar case, from a programming perspective.
Citation example
Markup Renders as
{{cite journal | author=Staff |date=5 June 1998| title=Title |language=Japanese | url= | journal=[[Famitsu Weekly]] | accessdate=2006-07-15}} 
Staff (5 June 1998). "Title". Famitsu Weekly (in Japanese). 
For clarification, I am not asking for changes to how the above accessdate condition is displayed. I have no objection to that. – Jonesey95 (talk) 18:01, 12 October 2014 (UTC)

Deprecated parameters?[edit]

Asked at Wikipedia:Help desk#Module:Citation/CS1.

Module:Citation/CS1/Configuration has a number of parameters commented with "remove after 1 October 2014". I am guessing that the use would be updated, the documentation updated and the parameters removed. --  Gadget850 talk 10:46, 12 October 2014 (UTC)

I thought about doing that with yesterday's update but there was enough in the update that I just didn't want to add more. The 1 October date is not a hard and fast date. I put it there as a reminder to myself to remove those deprecated parameters that it makes sense to remove. Clearly, |coauthors= isn't ready for removal yet but most of the others can be so will likely be removed at the next update.
Trappist the monk (talk) 12:52, 12 October 2014 (UTC)

Date range[edit]

Hi, I just noticed a broken cite journal date in an article. The publication was "date = January-March 1969" but the system doesn't like this, so I changed it to "date = January 1969". But how should a date range be specified? Thanks, Esowteric+Talk 14:30, 13 October 2014 (UTC)

Use an endash instead of a hyphen. Did the link in error message not answer your question? If not, how can we improve it?
Date with hyphen. January-March 1969.  Check date values in: |date= (help)
Date with endash. January–March 1969. 
Trappist the monk (talk) 14:46, 13 October 2014 (UTC)
Ah, thanks! Sorry, I was in too much of a dash and didn't correctly read the help. Esowteric+Talk 14:48, 13 October 2014 (UTC)

cite journal and quarterly publications[edit]

I took a quick search through the archives but I don't see mention of this. Some of the historical journals that I reference for railroad history are issued quarterly and their issue dates read like "First quarter YYYY" through "Fourth quarter YYYY". When I put that information into the date parameter, the page gets added to CS1 errors: dates even though this specification is valid according to WP:SEASON. Typing "the first quarter of YYYY" instead of "first quarter YYYY" seems nonsensical for a citation, so I've used the shorter of the two versions.

So what I guess I'm asking is this: how should I be entering journal issue dates when the journal is issued quarterly and specifies the quarter number in its official issue date rather than the month name? Thanks! Slambo (Speak) 16:33, 13 October 2014 (UTC)

CS1 doesn't support quarterly dates per se; and WP:DATESNO, from which CS1 takes its date format guidance, is mute on the topic. You might write |date=January–March YYYY.
Trappist the monk (talk) 16:48, 13 October 2014 (UTC)

Error messages across the entire project[edit]

Please stop playing around with {{cite web}} immediately. For the first time in years, almost all preformatted dates and accessdates are suddenly "rejected" by this template with a nasty error message. Please put it back where it was. These are not "errors". Thanks, Poeticbent talk 18:09, 13 October 2014 (UTC)

@Poeticbent: Please give a few examples. --Redrose64 (talk) 18:34, 13 October 2014 (UTC)

  • I really have no time for this but here are a few examples: 1.[1] 2.[2] 3.[3] 4.[4] 5.[5]

  1. ^ "Stutthof (Sztutowo): Full Listing of Camps, Poland" (Introduction). Jewish Virtual Library. Retrieved 2014-10-7. "Source: "Atlas of the Holocaust" by Martin Gilbert (1982)."  Check date values in: |accessdate= (help)
  2. ^ Grossman, Vasily (1946), The Treblinka Hell [Треблинский ад] (PDF file, direct download 2.14 MB), Moscow: Foreign Languages Publishing House (online version), retrieved 5 October 2014., "original in Russian: Гроссман В.С., Повести, рассказы, очерки [Stories, Journalism, and Essays], Воениздат 1958."  Check date values in: |accessdate= (help)
  3. ^ Sereny, Gitta (1974, 1995, 2013). Into That Darkness: From Mercy Killing to Mass Murder (Google Books preview). Random House. pp. 54–. ISBN 144644967X. Retrieved 5 October 2014.  Check date values in: |date= (help)
  4. ^ Burian, Michal; Aleš (2002). "Assassination — Operation Arthropoid, 1941–1942" (PDF file, direct download 7.89 MB). Ministry of Defence of the Czech Republic. Retrieved 5 October 2014..  Check date values in: |accessdate= (help)
  5. ^ Browning, Christopher R. (1992; 1998). "Arrival in Poland" (PDF file, direct download 7.91 MB complete). Ordinary Men: Reserve Police Battalion 101 and the Final Solution in Poland. Penguin Books. pp. 52, 77, 79, 80. Retrieved October 5, 2014. "Also: PDF cache archived by WebCite."  Check date values in: |date= (help)

— Preceding unsigned comment added by Poeticbent (talkcontribs) 18:53, 13 October 2014 (UTC)

See the linked help page, which references MOS:BADDATEFORMAT

  1. "2014-10-7" is missing a leading 0
  2. "5 October 2014." includes a period
  3. "1974, 1995, 2013" my swag is this should use |origyear=
  4. "5 October 2014." includes a period
  5. "1992; 1998" is an ambiguous date; this should use |origyear=

--  Gadget850 talk 19:03, 13 October 2014 (UTC)

(edit conflict) OK, here we go. In (1) there is |accessdate=2014-10-7 this should be |accessdate=2014-10-07 - a zero is missing; in (2) there is |accessdate=5 October 2014. this should be |accessdate=5 October 2014 - the full stop is invalid; in (3) there is |year=1974, 1995, 2013 this should be |origyear=1974|year=2013 - only one year per parameter should be given, and the 1995 is superfluous since all we need are the year of the edition that was actually used plus (optionally) the original year of publication; in (4) there is |accessdate=5 October 2014. - full stop again; in (5) there is |year=1992; 1998 this should be |origyear=1992|year=1998 as with (3). --Redrose64 (talk) 19:05, 13 October 2014 (UTC)
  • Please stop defacing articles with this nonsense. A zero missing in 7: 07? A full stop again after 2014? How many thousands of articles have them... and, for God's sake, who cares if the full stop is invalid! The dates are there. They are readable and easy to trace. Unless your intention is to fix all these "errors" yourself, please let go of scare tactics. Thanks, Poeticbent talk 19:24, 13 October 2014 (UTC)
The intent is to improve the quality of the citations by providing error checking that complies with the Manual of Style. We have had the date checking for several months, but have not displayed the error. This has allowed bots to repair thousands of bad dates. But now we are down to dates that cannot be automatically fixed and need human intervention, thus the errors now show. --  Gadget850 talk 19:34, 13 October 2014 (UTC)
  • The "errors" do not show, only the "error message" shows like a sore thumb. The actual numbers (which are dates added by users in a matter-of-factly way) remain hidden from the reader. This whole affair is ripe for some kind of intervention. I wonder how many articles are affected by this new red monster across the entire project. Poeticbent talk 20:11, 13 October 2014 (UTC)
What you're asking is that a programmer should (1) be able to imagine every possible way that an editor could screw up a date in a way that doesn't rise to the level of "error", and (2) spend the time to include toleration of those screw-ups in the code. Have you done much programming? That's simply not practical short of something approaching artificial intelligence. Feel free to start an RfC to ask whether the community would rather have the errors or the red errors reminding editors to correct the errors. I'll be there to negate your !vote. ‑‑Mandruss (talk) 20:36, 13 October 2014 (UTC)
You can see the articles with date errors by looking at Category:CS1 errors: dates, which currently has 60,874 articles in it, which is a decline on what it was the last time I looked before the error message was turned on. This is good as it was steadily rising before that and the BOTs could only fix some of the new article errors. Keith D (talk) 23:54, 13 October 2014 (UTC)
  • Do you guys hear me at all... or you're just to wrapped up in your own chain of sorrow. Twenty four hours went by, and the number of articles affected by this new red monster has gone from 60,874 articles to 60,698 total. Some 176 articles were fixed... sixty thousand six hundred ninety eight articles remain defaced. At this rate, all silly pseudo-errors will be fixed in 344 days providing that no new articles get written and no new references added with even more of those. I'm taking this page off my watchlist. Poeticbent talk 23:25, 14 October 2014 (UTC)
    • Whilst I want to remain neutral on the actual issue myself, my overall feeling is that consensus is against you on this thread, at least. I think that, if you want to overturn this, the best way to go is to follow others' advice here and start an RfC. It Is Me Here t / c 14:03, 15 October 2014 (UTC)
Part of me says this thread should be allowed to die a natural death. According to the OP, s/he has given up trying to get through to us morons. But I'm the only one who has mentioned RfC, so if we're encouraging the OP to start one I guess I should clarify what I meant. For starters, I said "feel free to", which is not the same as "advice" to do so. To the contrary, I think it would be a really bad idea, and a waste of people's time. I was not referring to an RfC to ask whether CS1 should flag some some undefinable set of date errors that don't actually prevent the reader from understanding the date. As I said, that is not practical, and we don't ask programmers to do impractical things. That RfC would be a snow close before it got started. What I meant was an RfC to ask whether CS1 should flag any date errors at all. I was only half serious about that, I know it would fail, and I don't think it's what the OP wanted. ‑‑Mandruss (talk) 14:34, 15 October 2014 (UTC)
An obscure discussion higher up the page has suddenly generated '000s of error messages all over the wiki, many relating to very simple "errors" that are generated by current "official" template-filling tools inbuilt to the editing page that throw up dates like "2012 Mar". You can't fix those by bot??? Really? You can't change the tools first??? There is in any case no ambiguity there, so no actual need to fix anything. I don't see why a bot can't fix other errors like filling leading zeros to numbers between 1-9. The whole situation is absurd. Look at Pancreatic cancer for example. No-one is ever going to fix all all these errors, and realy why should they bother. I would certainly support an RFC. Let me know if one launches. Wiki CRUK John (talk) 14:03, 16 October 2014 (UTC)
Wiki CRUK John: BattyBot is fixing the errors you describe. It runs about once a month and looks like it will stop by to edit Pancreatic cancer in the next day or two. "2012 Oct" is not a valid date per the MOS, so it is flagged as invalid so that someone, or some bot, will fix it in order to present valid dates to our readers.
The date error messages were exposed for a while in 2013, then hidden for about a year after this RFC resulted in their being hidden until a bot could go through the error category and fix as many as were feasible. We did that, and the vast majority of remaining date errors require human intervention. We have solid evidence that at least some human editors are motivated by the red error messages to bring the dates into compliance with the MOS, which is linked from the help page that you see when you click on "help" immediately following the error message.
We have already alerted the maintainers of some citation-filling tools that their tools were being used to generate citations with formatting errors, and many of those tools have been fixed. If you use a particular tool that is generating citations with errors, contact the maintainer of that tool, and feel free to refer them to this Talk page for assistance. – Jonesey95 (talk) 14:37, 16 October 2014 (UTC)
When you wrote No-one is ever going to fix all all these errors, I expected a sea of red error messages. I found 13 in a list of 107; I fixed the five date errors and the six accessdate errors. The two url-missing-title errors (currently 100 and 102) are left for editors more knowledgeable on the topic to fix. It took me about five minutes.
Trappist the monk (talk) 14:52, 16 October 2014 (UTC)
Thanks both, but @Trappist the monk, you know what you are doing, which most editors won't. Factor in a trip to the MOS to try to work out what is wrong, and the initial fix time rises hugely. Even at 5 minutes, with 68K pages to fix, and presumably new ones being added all the time, that's still 6,000-odd working hours. @Jonesey95, I have no idea who, if anyone, maintains the drop-down cite tool in the standard editing window, which I think isn't even a preference option nowadays. Do you? Wiki CRUK John (talk) 18:06, 16 October 2014 (UTC)
That would be Wikipedia:RefToolbar. --  Gadget850 talk 18:50, 16 October 2014 (UTC)
Thanks, I've raised it there, & we will see what, if anything happens. It still seems bizarre that in one part of the forest there is a tool busy generating "errors" and in another a team busy fixing them, without fixing the tool. Although some of the people seem the same. Wiki CRUK John (talk) 12:29, 17 October 2014 (UTC)
Yep, that's a lot of hours. But those errors were not created overnight; it has taken us multiple years to make them all and we didn't even know that we were doing it. Now the foot is in the other shoe.
Trappist the monk (talk) 19:09, 16 October 2014 (UTC)
I worked error tracking cats for some time before the messages were turned on. I could see the messages, but only because I made some modification to my local configuration that I can't remember—something that would be beyond most editors even if they could find the instructions on how to do it. I remember thinking, "of course there are a lot of errors, the editors can't see the messages." I think being a quality encyclopedia, which is the goal, includes consistency in these date values. Since citations are at the core of what WP is about, their quality is as important as anything else an editor does. As for the backlog, see WP:WIP. We'll get there. ‑‑Mandruss (talk) 18:45, 16 October 2014 (UTC)
Regarding the error-fixing rate, my experience has been that once I get going, I can fix about 50–100 errors per hour, depending on how easy they are to fix and whether I need to click through to additional web pages to locate an unambiguous date. 60,000 errors will take a while to fix, but luckily, there is no deadline. Most of the erroneous dates are still human-readable and more or less understandable, so often it is a question of fixing formatting rather than content. – Jonesey95 (talk) 19:01, 16 October 2014 (UTC)
Help:CS1_errors#Controlling_error_message_display. Pretty much just copy and paste.
Trappist the monk (talk) 19:09, 16 October 2014 (UTC)
@Wiki CRUK John: You posted the same comment in two sections on this page. See my response to you at the end of the #Time to show date error messages? section above. GoingBatty (talk) 00:46, 17 October 2014 (UTC)
similar comments - replied above. Wiki CRUK John (talk) 12:29, 17 October 2014 (UTC)
Answered at Wikipedia talk:RefToolbar#Toolbar generating date errors. --  Gadget850 talk 14:14, 17 October 2014 (UTC)

Similar template with cite web with date format in YYYY-MM-DD[edit]

Hello, I am looking for a template similar with {{cite web}}, that is capable to accept the date in the YYYY-MM-DD format and then to transform it into the desired output.

Imagine for example a template named {{cite web US}}, where, when I specify the date like "date= 2014-01-07", it will show it like "January 7, 2014" (or "January 7th, 2014").

And another template named {{cite web UK}}, where, when I specify the date like "date= 2014-01-07", it will show it like "7 January 2014" (or "7th of January 2014").

And then, each Wikipedia will translate that template into their own language. So when you create a reference (manually or automatically), you only have to specify the data in the universally YYYY-MM-DD format, so you don't have to bother to understand how that language outputs the date. For example, at Spanish Wikipedia, "date= 2014-01-07", will output "7 de enero de 2014". Therefore, the same reference can be translated and adapted everywhere (in any other Wikipedia), and then when you use it, you don't need to know how each language outputs the date.

To be more specific, the following reference:

will output it like:

on Spanish Wikipedia, it ({{cite web YMD}}) will output it like:

Is there such a template, or can anyone create such template? Thanks —  Ark25  (talk) 10:45, 14 October 2014 (UTC)

There isn't such a template as far as I know. There was a time when something similar was done based on a user preference setting. That functionality went away. The history of that in in the {{citation}} or {{citation/core}} archives I think. The problem as I see it is that here at, each article can have its own date format style which may or may not be specified by templates like {{use dmy dates}} and {{use mdy dates}}. Templates can only deal with information that they are given so unless each template includes a date format parameter of some sort, it will have no way of knowing what the article requires.
Trappist the monk (talk) 11:18, 14 October 2014 (UTC)
This has had a lot of discussion over the years. See User:Gadget850/FAQ/YYYY-MM-DD dates, especially reference 4. Bottom line: write the article and the references with the date styles desired. --  Gadget850 talk 11:41, 14 October 2014 (UTC)
The various Citation Style 1 templates did format dates according to user pref setting (known as "dynamic dates") - for about six weeks in December 2008-January 2009. Not only was it technically complicated, it was also unpopular. In more general terms, Wikipedia deprecated the use of dynamic dates in late 2009, and the facility for performing that function was removed from the MediaWiki software itself with the release of MediaWiki 1.21 in March 2013. It's not coming back. --Redrose64 (talk) 16:38, 14 October 2014 (UTC)
Wait, I don't need something so deep as a special MediaWiki functionality or article-wide settings for how to output the data. I don't want the template to interfere with any settings. I just need a new template, with this tiny feature: to accept the data field specified in YYYY-MM-DD format and to output it in a certain way. Such a functionality would make it much easier to translate Wikipedia articles from one language to another, because you won't have to bother to translate the dates of every single reference you have in the translated articles. —  Ark25  (talk) 17:31, 14 October 2014 (UTC)
I suggest User:Ohconfucius/script/MOSNUM dates. When you edit it gives you options on the side bar to convert all dates to the desired format. This will ensure the entire article is unified, not just the citations. If you add it to your global.js, it should be available cross wiki. --  Gadget850 talk 18:13, 14 October 2014 (UTC)

──────────────────────────────────────────────────────────────────────────────────────────────────── Sorry, I don't need anything like unifying an entire article. I just need the template to output the date as I say. I tried it other way, using the {{date}}, it works, but it doesn't work with subst:

This one works:

  • <ref name="MyUser_BBC_October_15_2014c">{{cite web |url= |title=How do zoos prepare for dangerous animal escapes? |newspaper=BBC |date= {{date|2014-04-18|mdy}} |author=Laurence Cawley |accessdate= {{date|2014-10-15|mdy}}}}</ref>

And this one doesn't work:

  • <ref name="MyUser_BBC_October_15_2014c">{{cite web |url= |title=How do zoos prepare for dangerous animal escapes? |newspaper=BBC |date= {{subst:date|2014-04-18|mdy}} |author=Laurence Cawley |accessdate= {{subst:date|2014-10-15|mdy}}}}</ref>

 Ark25  (talk) 22:34, 14 October 2014 (UTC)

subst: never works inside <ref>...</ref>, this is a known problem, and is not specific to one template (or group of templates). There's nothing that we can do about it except wait for the existing bugzilla: reports to be actioned. --Redrose64 (talk) 00:13, 15 October 2014 (UTC)
Thanks, it looks like there is already a bug report for it: Ark25  (talk) 09:09, 16 October 2014 (UTC)

Two months in the date parameter[edit]

See also cite journal and quarterly publications]]

I want to use a {{harvnb}} with a year parameter of 1828. See here, the date in the publication is given as "February & June MDCCCXXVIII" which before the introduction of CS1 could be dealt with as in a {{cite book}} with the parameters set to the following values "month=February & June |year=1828" or as "year=1828 |date=February, June 1828" How to deal with it now as CS1 barfs on "date=February, June 1928" (Help:CS1 errors#bad date). Of course one can use "date=February–June 1828" or even "date=June 1828", but that is a distortion of date in the source.

 {{cite book |title=Date months with an ndash: |date=February–June 1928}}
Date months with an ndash:. February–June 1928. 
{{cite book |title=Two months fails: |date=February, June 1828}}
Two months fails:. February, June 1828.  Check date values in: |date= (help)

-- PBS (talk) 19:12, 14 October 2014 (UTC)

It looks like that source is the February and June issues of The Foreign Quarterly Review in a single binding. The June issue appears to begin on page 403 – if you 'search in this book' for "list of the principal new works", google will show you two search results at pages 395 and 750. I think that simplifies the problem: cite the issue that contains your source material. If you want to cite the whole book, use the "Published in... as a subtitle: |title=The Foreign Quarterly Review: Published in February & June MDCCCXXVIII and set |date=1828.
Trappist the monk (talk) 20:27, 14 October 2014 (UTC)
Rather bizarre case, especially if there were any intervening issues. Perhaps they published three times a year, and combined two issues? Something like a single "January—February" issue would be ordinary enough, while two separate issues bound together would cited individually. This seems more like a book which was compiled from two issues, and the "February & June" seems more like a note of the original source than a subtitle. But if you go with this as a date I'd say use "date=February—June 1828". ~ J. Johnson (JJ) (talk) 21:25, 14 October 2014 (UTC)

Date with "onwards"[edit]

There are a number of web sites which give their preferred citation in the form "YEAR onwards" (e.g. the Angiosperm Phylogeny Website – see [2]) or "YEAR+" (e.g. the Euro+Med database – as one example see [3])). I recall pointing this out when the date processing in the cite/citation templates was updated. Yet dates of this form produce errors. Peter coxhead (talk) 16:11, 15 October 2014 (UTC)

Because WP:DATESNO does not support open-ended date ranges (2001–, or 2001+, or 2001 onward, etc.), and in some cases discourages them, CS1 does not support open-ended date ranges.
Trappist the monk (talk) 16:59, 15 October 2014 (UTC)
But if these are the standards recommended by major websites (and I can list several more) then they should be supported. They are really part of the "external titles and quotes" given by WP:DATESNO: using them is just using the citation given in the source. I use several of these sources very regularly and am very reluctant to have to give up the template in favour of plain text, which is what I will have to do if they aren't supported. Peter coxhead (talk) 17:08, 15 October 2014 (UTC)
The CS1 templates are designed to comply with the MoS, including WP:DATESNO. Thus the place to start this is WP:DATESNO. --  Gadget850 talk 17:26, 15 October 2014 (UTC)
My understanding/memory of the Wikipedia guidance on citing websites is that those sites are in the wrong (in the sense that "YEAR onwards" is unnacceptably ambiguous as to the version of the source consulted) and in this case one can ignore the given date and must use |accessdate= to identify the date you consulted the source. Just because a source asks us to cite in a particular way does not mean we have to, particularly if the format is not fit for our purposes.
--TuxLibNit (talk) 18:31, 15 October 2014 (UTC)
@TuxLibNit:, @Gadget850: so to be clear you would not use |date= at all for such websites, just |accessdate= (which of course I always give)? Peter coxhead (talk) 10:17, 16 October 2014 (UTC)
@Peter coxhead:Well I can't speak for Gadget850 but I expect I would omit |date= in this case even if it was a free text field that could accept "<year> onwards" without the red warnings. For me, "<year> onwards" contains almost no useful information, and none at all once "Retrieved on <date>" is given.
TuxLibNit (talk) 16:49, 18 October 2014 (UTC)

Cite Journal/magazine[edit]

Is there any particular reason why {{Cite journal}} displays page numbers in a format unlike the rest of the cite x series of templates? Why is p. or pp. missing here? Resolute 17:27, 15 October 2014 (UTC)

Cite journal
Markup Renders as
{{cite journal|author=Author|title=Title|journal=Journal name|pages=56-62}} 
Author. "Title". Journal name: 56–62. 
Cite book
Markup Renders as
{{cite book|author=Author|chapter=Chapter name|title=Book Title|pages=56-62}} 
Author. "Chapter name". Book Title. pp. 56–62. 
Examples above. – Jonesey95 (talk) 17:35, 15 October 2014 (UTC)
Don't know. Perhaps it's because the various templates that eventually became CS1 were independently developed. You might research the the template histories and report back.
Trappist the monk (talk) 18:17, 15 October 2014 (UTC)
  • "test". test: 5. 
  • "test". test: 5. 
  • "test". test: 5. 
  • "test". p. 5. 
    • Looks like the work or journal (or other alias) parameter suppresses this for some reason. That's... just odd, especially when I want to cite a magazine's title. But I don't know enough about how the CS1 modules work to go further than that. This actually came up in a GA review because the reviewer didn't see that the page number is included, just formatted inconsistently compared to book and newspaper sources. I think my immediate solution is to simply use Cite News instead of Cite Journal for my magazine cites. Resolute 18:56, 15 October 2014 (UTC)
Or use |pages=pp. 56-62. ‑‑Mandruss (talk) 19:01, 15 October 2014 (UTC)
Not recommended; the extra pp. will be injected into the citation's COinS metadata. Very often, {{cite journal}} includes |volume= and |issue= parameters so perhaps the all-numeric formatting arises from that association.
The {{cite journal}} page handling has been done the way it has been for a long time: Compare the old {{citation/core}} version with the current version:
Cite journal compare
{{ cite journal | pages=56–62 | title=Title | volume=25 | issue=10 | author=Author | journal=Journal name }}
Old Author. "Title". Journal name 25 (10): 56–62.
Live Author. "Title". Journal name 25 (10): 56–62. 
Trappist the monk (talk) 19:22, 15 October 2014 (UTC)

────────────────────────────────────────────────────────────────────────────────────────────────────{{Cite journal}} states that the p. or pp. is displayed (i.e., added) unless |work= or |nopp=y is present. That appears to be incorrect, and either the code or the doc should be fixed. ‑‑Mandruss (talk) 19:36, 15 October 2014 (UTC)

The documentation is correct:
{{cite journal|title=test|page=5}}"test". p. 5. 
{{cite journal|title=test|journal=Journal|page=5}}"test". Journal: 5. 
|journal= and |work= are aliases of each other.
Trappist the monk (talk) 19:47, 15 October 2014 (UTC)
Interesting. I am fairly certain the templates once displayed the p. and pp., but I can't really argue against that that cite comapre tool. And I have used cite magazine for years. Rather surprising that I haven't picked up on this before. Thanks! Resolute 19:51, 15 October 2014 (UTC)
{{cite journal}} is primarily used for citing peer-reviewed academic journals ({{cite magazine}} has redirected to {{cite journal}} for some years, but I don't know why), and the people who mainly use peer-reviewed academic journals prefer to omit abbreviations like "vol.", "no." and "p.", preferring instead a system of boldface and parentheses. --Redrose64 (talk) 19:56, 15 October 2014 (UTC)
That is along the lines of what I figured, and also why I seriously dislike the idea of template consolidation in many cases. It seems that a possible solution is to split the journal prameter out as its own option which disables those abbreviations, but leave them on by default for the work/magazine/other aliases. Resolute 22:47, 15 October 2014 (UTC)
People who mainly cite peer-reviewed academic journals normally use a style manual that treats all sources in a reasonably unified manner, and which does not switch between using "p." or "pp." or not depending on what kind of work is being cited. Such style manuals also don't radically change the position of the publication date depending on what kind of source is being cited; an RFC concluded the date behavior should be fixed, but developers have preferred to fix other things. Jc3s5h (talk) 22:54, 15 October 2014 (UTC)
That's quite a stretch, Ttm, to say that the doc is correct if you've been around long enough to know the aliases of |work= (or to know that you have to check for possible aliases to make effective use of the doc). But I've been down this road before at Wikipedia. ‑‑Mandruss (talk) 19:58, 15 October 2014 (UTC)
Pretty sure that I did not write anything that was not true. But if you believe otherwise, perhaps you can back up your claim and show me where I went wrong. This is a quote from the {{cite journal}} documentation:
  • work: Name of the source periodical; may be wikilinked if relevant. Displays in italics. Aliases: journal, newspaper, magazine, periodical.
page and pages do not show p. or pp.
Trappist the monk (talk) 21:02, 15 October 2014 (UTC)
Volume was bolded in January 2006.[4] I don't see it ever used the page abbreviations. {{cite magazine}} was created as a redirect in 2006 with the summary of "creating redirect, 'cause I keep forgetting what the real template is called."
The documentation is correct, but I wrote it based on how the template worked. --  Gadget850 talk 21:08, 15 October 2014 (UTC)
Ttm: Yes, but elsewhere on that very long page, it says the following:
page: The number of a single page in the source that supports the content. Use either |page= or |pages=, but not both. Displays preceded by p. unless |nopp=y or |work= is defined.
[followed by similar language for pages:]
There is nothing there that tells the user, Don't take this literally, dummy. When we say "work=", what we really mean is, "work=, or any of its aliases".
The user should not be expected to have read the entire very long page and absorbed everything on it. Inexperienced users know nothing of aliases, and good doc is written for inexperienced users. They are the users who have the greatest need for the doc. ‑‑Mandruss (talk) 21:26, 15 October 2014 (UTC)

──────────────────────────────────────────────────────────────────────────────────────────────────── Here is the complete documentation section:

  • work: Name of the source periodical; may be wikilinked if relevant. Displays in italics. Aliases: journal, newspaper, magazine, periodical.
    • issue: When the publication is one of a series that is published periodically. Alias: number.
When set, work changes the formatting of other parameters:
title is not italicized and is enclosed in quotes.
chapter is italicized and is not enclosed in quotes.
location and publisher are enclosed in parentheses.
page and pages do not show p. or pp.
edition does not display.
type does not display.

As noted, if and only if work or one of its aliases is defined, then the formatting changes are applied. If you use {{cite journal}} and don't define work then the formatting is not applied. --  Gadget850 talk 21:50, 15 October 2014 (UTC)

I understand that, but the point I am struggling to make is that a user will not go to the doc for the work parameter for information about the page parameter. Instead, he will go to the doc for the page parameter, and it says nothing about work aliases there. Again, the doc is correct only if taken in its entirety, which is not how doc is used. ‑‑Mandruss (talk) 21:57, 15 October 2014 (UTC)
So allow me to state the obvious: If the documentation that we have is unsatisfactory, change it until it is. If you would like assistance in understanding what something does, ask, I'll help. I can't necessarily answer the why, but I can probably tell you the what where and when. We know that documentation is never good enough; help make it better.
Trappist the monk (talk) 23:12, 15 October 2014 (UTC)
If I'm understanding how the pieces fit together, the element that needs to be changed is Template:Citation Style documentation/pages. It currently contains the phrase: or work is defined. This phrase needs to read: or work or one of its aliases is defined. If possible I would like to link to the subsection that defines those aliases, which is currently titled "journal", as or '''work''' or one of [[#journal|its aliases]] is defined. Will this work there, and is there any reason the link shouldn't be done that way? ‑‑Mandruss (talk) 01:43, 16 October 2014 (UTC)
Can't hurt to try. I would suggest an alternate text: or '''work''' (or an [[#csdoc_work|alias]]) is defined.
Trappist the monk (talk) 11:00, 16 October 2014 (UTC)
The master template is {{citation Style documentation}}; each section has an [edit subtemplate] link. Every entry has an anchor, in this case it is "csdoc_work". --  Gadget850 talk 14:13, 16 October 2014 (UTC)
Done. Two edits, here and here. {{Cite journal}} looks good and the links work. Doc for other Cite templates displays as before, and that's the desired result as I understand it. Thank you. ‑‑Mandruss (talk) 15:38, 16 October 2014 (UTC)
@Gadget850: I don't think that it's quite true to say "every entry has an anchor" - the one in question was added by myself just a few days ago and a few minutes later I added another. I expect there are plenty more that I've not got around to yet. --Redrose64 (talk) 21:06, 16 October 2014 (UTC)

Multiple publishers in cite book[edit]

Does anyone know how to go about adding more than one publisher to a book citation? I've got a scenario where a source has two publishers in the same country who jointly published the book - but I don't know how to put the second publisher into the cite template. Thanks! Miyagawa (talk) 16:54, 16 October 2014 (UTC)

There may not be a great solution here. CS1 is not designed to accommodate multiple values except in the |author= and |editor= parameters. If you can, choose one publisher and use that one, especially if that one is a much better known publisher, presumably the one listed first on the book's title page. You could then add a note following the {{cite book}}: also Location:Second Publisher. Alternately, if the two publishers are in different locations, leave out |location= and then do something like this: |publisher=First Publisher, Second Publisher (jointly).
Trappist the monk (talk) 18:02, 16 October 2014 (UTC)
Great, thanks for the advice. Miyagawa (talk) 18:47, 16 October 2014 (UTC)

Chapter and its associated parameters[edit]

I've been wondering about how CS1 handles |chapter= and its associated parameters. There are some sixty possible combinations of |chapter=, |chapterlink=, |chapterurl=, |trans-chapter=, |title=, and |url=. These last two are included in this list because they can change how |chapter= is rendered.

Some of the things that I've noticed are:

  1. Like |chapterurl=, |chapterlink= should be applied to |trans-chapter= even in the absence of |chapter=
    • {{cite book |chapterlink=Abraham Lincoln |trans-chapter=Trans Chapter}}
      • [Trans Chapter] |trans_chapter= requires |chapter= (help).  Missing or empty |title= (help)
    • {{cite book |title=Title |chapterlink=Abraham Lincoln |trans-chapter=Trans Chapter}}
      • [Trans Chapter] |trans_chapter= requires |chapter= (help). Title. 
    • {{cite book |title=Title |chapterlink=Abraham Lincoln |trans-chapter=Trans Chapter |url=//}}
      • [Trans Chapter] |trans_chapter= requires |chapter= (help). Title. 
    • {{cite book |chapterlink=Abraham Lincoln |trans-chapter=Trans Chapter |url=//}}
      • [Trans Chapter] |trans_chapter= requires |chapter= (help). //  Missing or empty |title= (help)
  2. |chapterurl= and |chapterlink= are mutually exclusive; currently |chapterlink= has priority; is this correct? The template is citing a source so shouldn't a link to the source take precedence over a Wikipedia article about the source?
    • {{cite book |chapter=Chapter |chapterlink=Abraham Lincoln |chapterurl=//}}
    • {{cite book |title=Title |chapter=Chapter |chapterlink=Abraham Lincoln |chapterurl=// }}
    • {{cite book |title=Title |chapter=Chapter |chapterlink=Abraham Lincoln |chapterurl=// |url=//}}
    • {{cite book |chapter=Chapter |chapterlink=Abraham Lincoln |chapterurl=// |url=//}}
  3. When a citation contains either or both of |chapter= and |trans-chapter=, these parameters are linked by |url= regardless of the presence of |title=. Is this correct? Shouldn't |url= be limited to |title= just as |chapterurl= is limited to |chapter= and/or |trans-chapter=?
    • {{cite book |title=Title |chapter=Chapter |url=//}}
    • {{cite book |title=Title |trans-chapter=Trans Chapter |url=//}}
    • {{cite book |chapter=Chapter |url=//}}
    • {{cite book |trans-chapter=Trans Chapter |url=//}}

Trappist the monk (talk) 11:49, 17 October 2014 (UTC)

I do not see item 1 as an error. Once the missing |chapter= is added, the citation will display correctly. I don't think we should bend over backwards to render citations that are missing key parameters.
Item 2 is a new one to me. What is the purpose of |chapterlink=? It does not appear to be documented in any of the CS1 templates. As a reader, I expect a linked chapter title to point me to the source so that I can verify that the referenced information is in the source. |chapterlink= seems unlikely to do that. I would lean toward eliminating it. Do we know how often that parameter is used?
I think I brought up item 3 sometime in the past. Because |chapterurl= exists, |url= should always go with |title=. Given the complexity of CS1's handling of titles, chapters, entries, journal names, and similar parameters, I expect that fixing this without breaking anything will require very careful programming and testing. – Jonesey95 (talk) 00:37, 18 October 2014 (UTC)
The point of #1 is to note that |chapterlink= doesn't act the same way that |chapterurl= does. Simplified, with |chapter= present:
{{cite book |chapter= Abe's chapter |chapterlink=Abraham Lincoln |trans-chapter=Trans Chapter}}
"Abe's chapter" [Trans Chapter].  Missing or empty |title= (help)
|chapterlink= should link both |chapter= and |trans-chapter= like |chapterurl= does in this version:
{{cite book |chapter=Abe's chapter |chapterurl=// |trans-chapter=Trans Chapter}}
"Abe's chapter" [Trans Chapter].  Missing or empty |title= (help)
When |chapter= is missing, |chapterurl= still links |trans-chapter= similar to #3. |chapterlink= should do the same I think.
What is the purpose of |chapterlink=? I don't know; perhaps it is the |chapter= equivalent of |titlelink=. I don't recall ever having seen it in the wild so maybe it can/should get deprecated and removed.
By the time we get to the point of deciding to link |chapter= with |url= all of the various aliases have been reduced to the metavariable Chapter. I don't think that it is as complex as you are making it out to be. Of course I say this fully recognizing that I have of late missed the obvious.
Trappist the monk (talk) 01:14, 18 October 2014 (UTC)
A CirrusSearch using insource: syntax did not find any use of the search term chapterlink in any of the ways I could think of to format it. I was able to find four instances of the term chapter-link and chapter link none of which were |chapterlink= or an alias. I propose to deprecate this parameter because it is unused.
Trappist the monk (talk) 22:29, 22 October 2014 (UTC)
Support. It will be difficult to detect instances of this parameter's deprecation amid the population of deprecated parameter errors, unless we could persuade one of the local bots or an AWB user to search through the category's articles. I suppose deprecating the parameter is the logical first step, in any event. – Jonesey95 (talk) 03:37, 23 October 2014 (UTC)

Nowrap for accessdate[edit]

I was just thinking a bit about this yesterday and then overnight, a ping from Editor GoingBatty caused me to do this experiment. In Module:Citation/CS1/sandbox I have added <span class="nowrap">...</span> to the value assigned to |accessdate=. The code looks for the three MOS compliant date formats and wraps accordingly. YYYY-MM-DD is completely wrapped and for the other two, Mmm dd, yyyy and dd Mmm yyyy, the code wraps everything but the last space and the year. Invalid date formats aren't wrapped.

Live Sandbox
"Title". Retrieved 2014-10-18. 
"Title". Retrieved 2014-10-18. 
"Title". Retrieved 18 October 2014. 
"Title". Retrieved 18 October 2014. 
"Title". Retrieved 18 October 2014. 
"Title". Retrieved 18 October 2014. 
"Title". Retrieved 18 October 2014. 
"Title". Retrieved 18 October 2014. 
"Title". Retrieved October 18, 2014. 
"Title". Retrieved October 18, 2014. 
"Title". Retrieved October 18, 2014. 
"Title". Retrieved October 18, 2014. 
"Title". Retrieved October 18, 2014. 
"Title". Retrieved October 18, 2014. 

I have seen ISBNs that wrapped inappropriately so those and perhaps other identifiers are candidates for nowrap. Is there anything else that should be prevented from wrapping in inappropriate places?

Trappist the monk (talk) 11:30, 18 October 2014 (UTC)

I'm glad this is getting some attention. But am I understanding correctly? You think it's OK to wrap between October and 18, but not between 18, and 2014?
Lack of sensible wrapping in ISBNs, DOIs, and so on is one of my biggest peeves about the citation templates, so I hope those can be rationalized as well. But they're way too long to be simply nowrapped. EEng (talk) 14:36, 18 October 2014 (UTC)
Better table.
Trappist the monk (talk) 15:27, 18 October 2014 (UTC)
Yeah, but what you said is, "the code wraps everything but the last space and the year." That sounds wrong. The effect should be October{{nbsp}}18, 2014 or 18{{nbsp}}October, 2014, but it sounds like you're saying something like October 18,{{nbsp}}2014. Have I got it wrong? EEng (talk) 23:02, 18 October 2014 (UTC)
@EEng: I read that as, "the code wraps everything [in the span tags] but the last space and the year". Imzadi 1979  23:13, 18 October 2014 (UTC)
This: <span class="nowrap">October 18,</span> 2014, <span class="nowrap">18 October</span> 2014, and <span class="nowrap">2014-10-18</span>.
Trappist the monk (talk) 00:07, 19 October 2014 (UTC)
Just curious: 1. Why is this about accessdate and not the other CS1 dates? 2. Are we talking about appearance in the reflist and the citation tooltip? ‑‑Mandruss (talk) 23:39, 18 October 2014 (UTC)
1. No doubt there are other things that could/should be given the nowrap css class. |accessdate= is convenient for testing. And it is at the end of the rendered citation so perhaps has a greater chance of wrapping improperly. 2. Yes.
Trappist the monk (talk) 00:07, 19 October 2014 (UTC)
Ok. I would ask how publications with high MOS standards would handle this. I'm thinking they would break a date only at a comma, but it should be possible to verify that with a style manual. Or the question could be asked at WP:Reference desk/Language. ‑‑Mandruss (talk) 00:19, 19 October 2014 (UTC)
What is a high MOS standard?
Trappist the monk (talk) 00:25, 19 October 2014 (UTC)
I mean close attention to matters of form as opposed to substance, such as is used at newspapers like NYT. They have what I would call a high MOS standard, whereas my local town newspaper has a lower MOS standard. Btw, I don't see any guidance on the question in WP:DATES. ‑‑Mandruss (talk) 00:41, 19 October 2014 (UTC)
Don't know about them. We have MOS:NBSP and if you read between the lines at WP:DATESNO (see the ranges section at MOS:DOB) you can see how dates are formatted to prevent line-wrapping.
Trappist the monk (talk) 01:00, 19 October 2014 (UTC)
  • This is very important... take a look at WP:Manual_of_Style/Dates_and_numbers#Non-breaking_spaces. You can't infer much if anything from the absence of nbsp at any given point in any given example. For whatever reason, there hasn't been enough trouble about where nbsp is/should be used for MOS to give comprehensive treatment of the question (with respect to dates or, generally, anywhere else, with a maybe a few exceptions). EEng (talk) 01:47, 19 October 2014 (UTC)
I went ahead and asked the question at WP:Reference desk/Language, without mentioning the reason for asking, as that might affect the replies. There are a few smart language guys watching that page. ‑‑Mandruss (talk) 01:57, 19 October 2014 (UTC)
Sorry, but I think that's an extremely poor idea. Most editors will say, with some justice, that this kind of machinery should follow MOS, and if MOS needs clarification it should be done in a MOS discussion. By asking at Refdesk you're just inviting a lot of personal opinions or old remembered style rules. EEng (talk) 03:45, 19 October 2014 (UTC)
Did you read what I posted there? I specifically said I was not looking for personal opinions. If any are given, I will ignore them. If our MOS were clarified as to this question, I suspect it would be made consistent with major style manuals—why wouldn't it?—which is what I asked for in my post. The major style manuals are regularly updated per modern trends, so they don't represent "old remembered style rules". That said, it wouldn't be my first extremely poor idea! ‑‑Mandruss (talk) 03:54, 19 October 2014 (UTC)
I just consulted my three style guides, (APA 6th ed., CMOS 16th ed., and MLA 7th ed.), and all three are silent on this issue. That's not too surprising to me since this is more a matter of typesetting. MOS:NBSP does say, "It is desirable to prevent linebreaks ... within expressions such as 17 kg, AD 565, 2:50 pm; in other places where breaking across lines might be disruptive to the reader ...." I've always interpreted that to mean that it is undesirable to have a number separated from other content to which is explicitly related. I edit a lot of highway articles, and we encourage fellow project members to use non-breaking spaces in things like "exit 326", "Route 66" or "US 41" to keep the number together with the text because that is one name. That name forms a unit that should be parsed together. This follows the advice and review commentary I've received in working on Featured Articles over the years, and I would say that the day of a month and the month are similarly a single unit. Now in WP:MOS#Years and longer periods, and advice I've been given over the years, tells me that years can be parsed as a separate unit.One of the examples in the list following that quotation from MOS:NBSP above shows "May 2014" with a non-breaking space to keep them together. Sadly, no examples deal with full dates or months and days taken together. Imzadi 1979  05:40, 19 October 2014 (UTC)
Ok, thanks for the research and the rest. That satisfies me, and it sounds consistent with Ttm's opening post. Ttm, you could collapse beginning with my 00:19, 19 October comment, if you like, as everything after it followed from it. ‑‑Mandruss (talk) 06:19, 19 October 2014 (UTC)

Cite4Wiki Phoenix released[edit]

Cite4Wiki Phoenix is a Firefox extension intended to make it easier to generate citations for a page which you are currently viewing. It has a number of features which are configurable in order to generate citations formatted as desired for the article which you are working on. The point of view is that the tool should do a good job at generating values for parameters, but ultimately the user is in control of what actually goes in the citation.
Some feature highlights:

  • Extensible page scraping for any/all parameter(s)
There is not yet a GUI to make it easy to extend the page scraping Profiles, but page scraping configurations are read from a file with each Profile (a domain, or subset of a domain) in JSON format. The user can specify a file containing their own definitions.
  • Includes fleshed-out profiles for many different websites (many news outlets, JSTOR, HighBeam, Questia, PubMed, arXiv,, WebCite, Google Books, etc.)
  • Includes a decent default page scraping definition for domains for which there is not a specific one available.
  • Automatic and semi-automatic preemptive archiving ( and/or WebCite) with easy display of the archive page for user verification.
  • Automatic detection of some failures and fallback to the other archiving site.
  • Automatic and semi-automatic case formatting for parameters (user selectable, both what type of casing change and what parameters)
  • Uses TemplateData (when available) for determining available parameters and if there are duplicated aliased parameters (warns user).
  • "Picker": single click selection of text elements within a webpage and applied to the parameter.
  • Automatic parsing of author and editor names into first/last (can be overridden by user).
  • Unlimited number of authors and/or editors, or user can specify a maximum number to use.
  • Available user selection of the contents of all <meta /> tags on the page.
  • Automatic selection of the appropriate citation template (based on dynamic or static selection in the Profile used).
  • User selection of any Citation template (including all CS1 templates)
  • The list of templates is user definable, or free-form while citing a page.
  • User selection of citation layout format: horizontal/vertical/vertical with "=" lined up
  • Option to display in vertical format and automatically use horizontal when copying to the clipboard
  • Date formats in DMY/MDY/YMD
  • Automatic and semi-automatic naming of references (user configurable, name defined in the Profile used for the site or a random number component is generated).

A couple of example citations:
<ref>{{cite news |url= |title=Ebola Outbreak Boosts Odds of Mutation Helping It Spread |last1=Langreth |first1=Robert |last2=Cortez |first2=Michelle Fay |last3=Lauerman |first3=John |date=October 15, 2014 |work=[[Bloomberg News]] |publisher=[[Bloomberg L.P.]] |accessdate=October 19, 2014 |others=Photographer: John Moore/Getty Images |editor1-last=Gale |editor1-first=Reg |editor2-last=Putka |editor2-first=Gary |editor3-last=Pollack |editor3-first=Andrew |location=[[New York, NY]] |archiveurl=// |archivedate=October 19, 2014 |deadurl=no}}</ref>[1]

<ref name="PubMed-22012269">{{cite journal |url= |title=Requirements for Empirical Immunogenicity Trials, Rather Than Structure-Based Design, for Developing an Effective HIV Vaccine |last=van Regenmortel |first=M. H. |date=January 2012 |journal=Arch Virol |publisher=Springer |accessdate=19 October 2014 |issue=1 |doi=10.1007/s00705-011-1145-2 |volume=157 |pages=1–20 |pmid=22012269}}</ref>[2]

  1. ^ Langreth, Robert; Cortez, Michelle Fay; Lauerman, John (October 15, 2014). Gale, Reg; Putka, Gary; Pollack, Andrew, eds. "Ebola Outbreak Boosts Odds of Mutation Helping It Spread". Bloomberg News. Photographer: John Moore/Getty Images (New York, NY: Bloomberg L.P.). Archived from the original on October 19, 2014. Retrieved October 19, 2014. 
  2. ^ van Regenmortel, M. H. (January 2012). "Requirements for Empirical Immunogenicity Trials, Rather Than Structure-Based Design, for Developing an Effective HIV Vaccine". Arch Virol (Springer) 157 (1): 1–20. doi:10.1007/s00705-011-1145-2. PMID 22012269. Retrieved 19 October 2014. 

This follow-on to Cite4Wiki is a work in progress. This release should be considered to be an alpha or beta level release. I have been using Cite4Wiki Phoenix to make citations for some time now and found it to be quite useful.

It would be helpful to have input from others on any and all aspects of the tool. This includes anything (e.g. bug reports, desired/missing features, GUI changes, page scraping issues, lack of CS1 compliance, etc.). You can add it to Firefox from Mozilla Add-ons. — Makyen (talk) 23:56, 19 October 2014 (UTC)

Duplicate template parameters[edit]

bug 69964

On the October 23 update, templates with duplicate parameters will be tracked by a category defined in MediaWiki:duplicate-args-category.[5][6] --  Gadget850 talk 14:00, 20 October 2014 (UTC)

It looks like this new category will apply to all templates, not just citation templates. That will be fun. I could not tell from the code diffs whether there will be an error message associated with this categorization, to make it easier for editors to identify the location of the offending template. Can you tell whether there will be such a message, or whether we'll be able to add one here on the English Wikipedia?
For what it's worth, Citation Bot finds duplicate parameters in citation templates and converts the first instance of each duplicate to |DUPLICATE_parameter-name= (e.g. |DUPLICATE_title=). If there is no error message, we can run Citation Bot on the new category and it will create CS1 "unsupported parameter" errors for us. – Jonesey95 (talk) 03:02, 21 October 2014 (UTC)
Yes, this applies to all templates. The category defaults to Category:Pages using duplicate arguments in template calls. Duplicate parameters will only put the page into the tracking category; there is no provision to display an error. --  Gadget850 talk 17:07, 21 October 2014 (UTC)

How is spring 1969 a date error if a publication dates an issue that way?[edit]

I seem to be seeing dates indicated as "spring 1969" or "fall 1993" or "Nov./Dec. 1983" on the original publications showing an error when the date= field of the cite template inserts that value. That seems silly; we ought to be able to declare the exact date of a dated publication that is issued quarterly or bimonthly, for careful bibliographic work. You can find examples by page-searching for "(help)" in my user bibliography for updating articles. This brand-new error message isn't helpful to editors. -- WeijiBaikeBianji (talk, how I edit) 17:03, 21 October 2014 (UTC)

The template enforces the Manual of Style, and none of the dates you specified meet the MOS:DATEFORMAT format standard:
  • Title. spring 1969.  Check date values in: |date= (help)
  • Title. Spring 1969. 
  • Title. fall 1969.  Check date values in: |date= (help)
  • Title. Fall 1969. 
  • Title. Nov./Dec. 1983.  Check date values in: |date= (help)
  • Title. November–December 1983. 
--  Gadget850 talk 17:15, 21 October 2014 (UTC)
@WeijiBaikeBianji: I agree that it is critical for the Wikipedia citation clearly provide accurate date information so users can find the source for careful bibliographic work. Note that this source uses "Fall", while you used "fall" on your userpage. Wikipedia has its own house style for capitalization and punctuation that might be slightly different than some publishers.
BattyBot is fixing incorrect dates like those you mentioned in articles (e.g. this edit from June), but not on user pages. Also note that abbreviated months are also OK if that is consistent with the reference format of the Wikipedia article:
  • Title. Nov–Dec 1983. 
Thanks! GoingBatty (talk) 17:34, 21 October 2014 (UTC)
In most cases, we are not at a point where the information in a citation can be, or should be, matched character-for-character to the corresponding data provided by the publisher. Like all style manuals I have ever seen, the data from the publisher is adjusted to conform to Citation Style 1. Jc3s5h (talk) 17:57, 21 October 2014 (UTC)
Thanks for the speedy replies. I used the examples I just encountered to expand the help page for this error message. -- WeijiBaikeBianji (talk, how I edit) 18:01, 21 October 2014 (UTC)
  • "The template enforces the Manual of Style, and none of the dates you specified meet the MOS:DATEFORMAT format standard" - perhaps the template does enforce the MOS itself, but if so it is wrong to do so. That MOS section begins: "These requirements do not apply to dates in quotations or titles. Special rules apply to citations; see Wikipedia:Citing sources § Citation style." - aka WP:CITESTYLE. That section begins "While citations should aim to provide the information listed above, Wikipedia does not have a single house style, though citations within any given article should follow a consistent style.". If "Spring 1969" is how the source dates itself, that is the correct and only correct dating to use, and it the template's job to allow that - ideally including capitalization at least, but not punctuation. It is entirely wrong to attempt to twist citation dates to meet the general MOS article text dating rules. Wiki CRUK John (talk) 11:34, 22 October 2014 (UTC)
I'm somewhat baffled by what you wrote. CS1 is one of several citation styles in use at Wikipedia. Each of the standard styles listed at WP:CITESTYLE no doubt has its own particular rules for date format. Why should CS1 be any different? CS1 chooses to follow the style rules enumerated in Wikipedia's Manual of Style. If you wish to create articles that comply with APA, or with The Chicago Manual of Style, or whatever, then to be in compliance with the chosen style, you must render dates and other details as the style specifies. This same is true if you wish to use CS1 to style your citations. If you wish to create articles that mimic the styles used by your sources, you are of course free to do so. In that case, CS1 may not be useful to you.
CS1 allows |date=Spring 1969 but not |date=spring 1969. I'm not sure I understand what you mean by If "Spring 1969" is how the source dates itself, that is the correct and only correct dating to use, and it the template's job to allow that - ideally including capitalization at least, but not punctuation.
Trappist the monk (talk) 12:11, 22 October 2014 (UTC)
I concur with Trappist the monk. The citation in question from your user page:
The linked source uses "Spring 1969" so I don't see the issue at all.
I have done a bit of cleanup on dates just to see the problems; most are simple typos that we now hope that editors will catch and repair quickly. --  Gadget850 talk 12:55, 22 October 2014 (UTC)
Yes, Wiki CRUK John, that's a good point. If the Wikipedia manual of style itself doesn't impose a house style on how dates are mentioned in original sources (as some other style guides, such as those I used to work with in professional editing, do), then Wikipedia's citation templates might just as well allow as valid date formats any date format actually encountered in reliable sources. In any event, once a citation throws an error message and a Wikipedian like me follows the link to the help page, I hope the help page is informative (through specific examples) about what the error is and how to fix it. Thanks for the further comments here. -- WeijiBaikeBianji (talk, how I edit) 18:00, 22 October 2014 (UTC)
CS1 forms a distinct citation style which is described at "Help:Citation Style 1". The "CS1 compliance with Wikipedia's Manual of Style" section of that document states "CS1 uses Wikipedia:Manual of Style/Dates and numbers §§Dates and years (WP:DATESNO) as the reference for all date format checking performed by Module:Citation/CS1." So although Wikipedia in general does not apply the date rules in Wikipedia's "Manual of Style" to citations, CS1 does. Jc3s5h (talk) 18:13, 22 October 2014 (UTC)
But if CS1 enforces date styles more strictly than WP:CITESTYLE requires, the unfortunate consequence is that editors may choose not to use templates for those citations which are acceptable under WP:CITESTYLE but not to the CS1 templates. This means that metadata, etc. is lost, and is surely undesirable? This seems a classic case of the best being the enemy of the good: by being too strict, usefulness may be lost. Peter coxhead (talk) 19:18, 22 October 2014 (UTC)
The bots have reached a point of diminishing returns. Should we remove the date checking altogether? This would allow dates such as "posted on Tuesday, September 16, 2008", "05 tháng 08 năm 2014", "2nd cent.", "31 November<newline>2009" and the like. IF we just relax the error checking, what is allowable? --  Gadget850 talk 20:00, 22 October 2014 (UTC)
@Gadget850: conceptually it's easy for me to answer: in citations, we should allow editors to use date styles according to WP:CITESTYLE, i.e. those which editors have deliberately used and for which there is support in style manuals or published sources. Turning this into an algorithm is a different issue. It does seem to me that anything which is bot-fixable is almost certainly an error and so should be corrected. I myself wouldn't leave "red marking" of supposed date errors in place. At the least it should be possible for an editor to turn it off for individual references.
My real objection is to what seems to be Jc3s5h's view, namely that the cite templates should be restrictive rather than trying to encourage their use by supporting a range of styles where this is possible. Peter coxhead (talk) 09:00, 23 October 2014 (UTC)
We should leave date error messages in place until obviously wrong dates like "12/4/11" and "1975, 1986" have been fixed. – Jonesey95 (talk) 17:00, 23 October 2014 (UTC)
Historically, citation templates were created to produce consistently formatted citations, just like most style guides demand. Any value the parameter values might have as metadata was a by-product. To say that what we really care about is the metadata, and we no longer care that templates tend to produce a consistent style, is a reversal of the original purpose of citation templates, so would require a well-advertised RfC to be accepted. Jc3s5h (talk) 17:12, 23 October 2014 (UTC)

@WeijiBaikeBianji: our MOS does actually specify which date formats can be used for citation dates in an article. Under MOS:DATEUNIFY, it says:

Publication dates in an article's citations should all use the same format, which may be

  • any format from the "Acceptable date formats" table, or
  • formats required by the citation style being used.
  • However, all-numeric date formats other than yyyy-mm-dd must still be avoided.


Access and archive dates in an article's citations should all use the same format, which may be:

  • yyyy-mm-dd, or
  • the format used for publication dates in the article, or
  • the format required for the citation style adopted in the article.

In other words, if a Wikipedia article is citing a newspaper article dated today in the APA style, "2014, October 23" would be the appropriate date format. However, if that same Wikipedia article is using CS1 in its references, then it needs to follow the "Acceptable date formats" table, and the templates used for CS1 style are enforcing those formats.

@Wiki CRUK John, Peter coxhead: CS1 has its own style guide, which is Help:Citation Style 1. If editors want to use APA instead of CS1, and WP:CITESTYLE and MOS:DATEUNIFY both allow that option, then they need to consult the current edition of the Publication Manual of the American Psychological Association. However, the CS1 templates are their own style, so editors should not mix and match thinking it's acceptable. Imzadi 1979  15:17, 23 October 2014 (UTC)

Of course people have no idea there is such a thing as "Citation Style 1" or that they are using it, they just use the tool provided in the default editing window, which gives no indication of this. Wiki CRUK John (talk) 17:06, 23 October 2014 (UTC)
@Imzadi1979: no citation style manual – and I've both them in my academic career and taught postgraduate students how to cite and reference – ever covers all the possible complications in dealing with real publications rather than the idealized ones in the examples in the manual, and I've never come across a journal editor who didn't understand this. No-one is arguing that obviously incorrect dates should not be flagged dealt with, only that there needs to be some way of flexibly dealing with less straightforward cases which do occur (open-ended dates and discontinuous dates are two examples). Peter coxhead (talk) 17:26, 23 October 2014 (UTC)
For me, it's enough to have a help page that lists the actual "error" I'm encountering as I use the templates, with examples that suggest what "problem" has been identified by the automatic check, and what I can do as a careful editor to fix the problem. I attempted to fix the help page that the error message links to as soon as I saw the first part of the helpful discussion here. I have a lot of habits that carry over from years of editing print publications according to the Chicago Manual of Style that I've gradually had to unlearn as I do more editing on Wikipedia. I can unlearn old habits and practice new habits more efficiently if a help page linked to an error message takes me directly to the answer to my problem. (Thanks to all of the editors here for your helpful comments.) -- WeijiBaikeBianji (talk, how I edit) 17:42, 23 October 2014 (UTC)
Peter coxhead, above you seemed to argue that CS1 templates should accommodate a range of choices. Many style manuals do mandate certain date formats; a publication date of "July 12, 1962" used as a publication date in APA style would be an obvious error. If someone who will submit a manuscript to a journal that requires APA asked me to copy-edit the manuscript, it would be my job to fix it. But in your earlier post, you seemed to say CS1 should only flag date formats that no English-speaker would ever use. But your later post seems to indicate that a style manual cannot anticipate every situation, and should allow flexibility when it is impossible to both comply with the style manual and correctly convey the date information (which I agree with). So which is your position? Jc3s5h (talk) 18:33, 23 October 2014 (UTC)
They are different but related issues. The issue immediately above, on which it seems we agree, is independent of Wikipedia policies. I would go further within the English Wikipedia, which explicitly allows an extremely wide range of citation styles. The citation templates obviously can't support all of them, but I believe the value of the templates – including, but not limited to, some degree of consistency, error-checking, and providing metadata – is such that they should try to support more variation than would be permitted within in a "normal" publication. Where exactly to draw the line is open to discussion. I assume we would all agree that "31 February 2014" should be flagged as an error. However, although it's desirable to standardize "spring" to "Spring", it's perfectly meaningful to a reader, and I'm not convinced that it's sensible to generate a red error notice in an article in such cases. Peter coxhead (talk) 21:39, 23 October 2014 (UTC)
The advantage of limiting the choices in CS1 is when an editor (or sometimes, even a bot) comes an article that uses CS1 templates, but the style of the parameters is inconsistent, the editor immediately knows what to change them to, rather than having to go through the article history to see which style was first, or count the instances of different usages to see which one has a plurality. Jc3s5h (talk) 21:49, 23 October 2014 (UTC)