Template talk:Cite journal/Archive 4

 Done. עוד מישהו Od Mishehu 06:07, 12 March 2008 (UTC)

29-March-2008: I have grouped older topics above using headers "Topics from 2007" (etc.) to emphasize age of topics. Older topics might still apply, but using the tactic of yearly headers to note the age helps avoid rehashing old news, without archiving any ongoing issues. Also, new topics are more likely to be added to the bottom, not top. -Wikid77 (talk) 21:56, 29 March 2008 (UTC)

After correspondence with PubMed Central the site now displays the PMC ID prominently on both PubMed abstracts and the PMC pages themselves. Isn't it time that we integrate the {{PMC}} template with this template, or am I wrong here? JFW | T@lk 11:29, 31 March 2008 (UTC)

Ok, but how to structure ? Either as a link after the PMID link, or alternative for linking the paper's title:

1. As
   PMC 67890".
   This is though rather wordy and we do not explain that the linked PMID number is to an abstract webpage. So, do we want "|pmid=12345 |pmc=67890" to show as:
   PMID 12345 PMC 67890
2. However, what if a url link is available to the full article at the original journal's website, do we really need links to both versions of the full article ? Personally I think the original article should be prefered (as being the original, often better integration of figures & tables than at PubMed Central, and may have list of articles citating the paper).
   • So do we code this that if url is defined, then the pmc link is not displayed after the PMID ?
   • OR does a defined pmc get used as the link for the paper's title if url is not otherwise specified - i.e. coded as (dropping the multiple curley brackets for clarity):
     {#if {url} | [{url} {title}] | {#if {pmc} | [http://www.pubmedcentral.gov...={pmc} {title}] } }
     • Finally if the latter option then should the format parameter be prefixed with some indication that this is not the original journal being linked to, eg show "Title (Reprint.)", "Title (PMC reprint.)" or "Title (PMC reprint.)"

Lots of options here I know, but I only really want to try and code this the once :-) David Ruben ^Talk 19:19, 11 April 2008 (UTC)

Well zero response to above implementation options, so far :-) Just to reiterate, proposal is to use a "pmc" parameter as link (instead of url if that is not available) for the title. However if url is provided, then just appears as a PMC number after the PMID number. Any objections ? David Ruben ^Talk 20:54, 3 June 2008 (UTC)

Fine by me. I, for one, will be glad to see PMC integrated into {{cite journal}} regardless of how the output looks :) Fvasconcellos (t·c) 23:18, 3 June 2008 (UTC)

Integrating the PMC into cite journal is certainly a good idea. There has been some opposition to using the DOI to form the title link when no URL parameter is specified (although not from me); using the PMC to form the link seems no different, and could cause problems if the DOI (which is more universal—it's used for essentially all modern journals, not just medical ones) is made to operate this way as well. Either way, I suggest that the formats of the DOI, PMID, and PMC should all be similar, which is not the case now:

• Doe, John (2008). "Made up article".
  PMID 1234. {{cite journal}}: Check |doi= value (help); Cite journal requires |journal= (help
  )

ASHill (talk | contribs) 23:50, 3 June 2008 (UTC)

Agree that url should take priority, but doi often provides link just to an abstract of an article whereas PMC is always to the full whole article. Therefore I think the doi should not automatically be used by this template as a title link if there is no url specified, whereas using pmc as a title link would definitely be to a whole article (ie title link on url else pmc as lesser option, with no title link as standard for doi value but with it being like the pmid as a separate item at the end of citation).

That as it may be, certainly there now seems a stated consensus to have as a minimum "pmc" as a parameter like DOI & PMID included in some manner, so lets do this step first - we can then continue discussing whether or not to title link as a formating/style issue :-)

Style of displaying can also be played around with as ASHill notes. Currently we have the template showing PMID as per the mediawiki rendering of PMID 123, ie PMID 123 a single linked item rather than say

PMID: 123 which is how this template shows the doi value. However I think PMID needs be left how mediawiki currently styles it, whereas we can make pmc behave like doi & issn with the mnemonics separately wikilinked from teh external link of their value, so helping readers who do not know what these are. There is a difference between doi & issn as to whether to separate mnemonic from teh value with a space or a semicolon, but for now i'll add pmc to initially show as PMC:123 David Ruben ^Talk

01:46, 4 June 2008 (UTC)

Just to throw another layout style into the ring, PubMed Central itself seems to use no separating punctuation see [2] where "PMCID: PMC23456" style is used - I'll prevote with ASHill not to add ever new formats to this citation template :-) David Ruben ^Talk 01:57, 4 June 2008 (UTC)

Amen to not introducing yet another format. The options make my head spin. I would vote to change the formatting of PMIDs in this template to look like PMCs if possible, given MediaWiki consistency. Capitalizing 'DOI' might be a good idea, but only if done in other templates that use it; see

Template talk:Cite conference#DOI capitalization

.

I should note that I rarely edit medical articles, so I'm not familiar with the practical use of PMIDs and PMCs in these templates, but I am quite familiar with DOIs. I didn't realize that PMCs always link directly to full text; that makes using them to form the link more attractive than DOIs. However, others (not me!) have raised privacy concerns about relying on a third-party server to resolve links to a large number of articles. ASHill (talk | contribs) 02:30, 4 June 2008 (UTC)

So if I'm referencing a journal with a date of "Winter 2005", how would I put that? Right now I'm setting "month" to "Winter", which seems weird but the only real option that I can see.—Chowbok ☠ 21:10, 31 March 2008 (UTC)

It is what I have always done... David Ruben ^Talk 21:52, 31 March 2008 (UTC)

It's been a few days now, and D. Trebbien's request for edit in

 Done I see Happy-melon already made the changes yesterday - thanks David Ruben ^Talk 12:15, 6 April 2008 (UTC)

complaint - please verify that this is not the case, or provide a fix. Thanks, Happy‑melon 22:36, 6 April 2008 (UTC)

I don't see a problem:

*{{cite journal |author=Fred Bloggs |title=Title |journal=Journal |volume=1 |pages=12-23 |year=2008 |month=April}}
*{{cite journal |author= |title=Title |journal=Journal |volume=1 |pages=12-23 |year=2008 |month=April}}

Gives:

• Fred Bloggs (2008). "Title". Journal. 1: 12–23. {{cite journal}}: Unknown parameter |month= ignored (help)
• "Title". Journal. 1: 12–23. 2008. {{cite journal}}: Unknown parameter |month= ignored (help)

David Ruben ^Talk 23:18, 6 April 2008 (UTC)

I was testing out various combinations on my sandbox page (using Show preview only), but could not duplicate the problem. « D. Trebbien (talk) 23:20 2008 April 6 (UTC)

The only issue I see is if month is present & year is not:

{{cite journal |author=Fred Bloggs |title=Title |journal=Journal |volume=1 |pages=12-23 |month=April}}

• Fred Bloggs. "Title". Journal. 1: 12–23. {{cite journal}}: Unknown parameter |month= ignored (help)

--Karnesky (talk) 23:51, 6 April 2008 (UTC)

See also

Template_talk:Citation#The_month_is_not_displayed. I think that Adoniscik wasn't using the right template.--Karnesky (talk

) 23:53, 6 April 2008 (UTC)

I don't think funding information should be added to this very common template, as it doesn't help with locating an article (the chief reason to use this template) & would be used on a small minority of articles. It might make sense to have a "note" or "extra" field on this and other citation templates, as

WP:NOR. --Karnesky (talk

) 21:15, 9 April 2008 (UTC)

I'm talking about confirmed funding like the one on this study here "Acknowledgment. This study was partly funded by Homint." [3], this means that there is no need to make OR to determine the funding, since many will disclose it. A "extra" field could do the trick. Mind you, I could also just go and add the text "funding by xxx" inside the ref tags after the template, of course, and this would have the same problems as an "extra" field. (damn, I just noticed that cite journal already has "pmid" param for pubmed) --Enric Naval (talk) 16:46, 10 April 2008 (UTC)

In this study, it was uaed to compare the use of various journals across citations. Circeus (talk) 03:27, 14 April 2008 (UTC)

The documentation says:

Specify the DOI to provide a permanent and direct link to the article from the publisher

{{cite journal |last=Giles |first=Jim |title=Preprint Server Seeks Way to Halt Plagiarists |url=http://www.nature.com/nature/journal/v426/n6962/full/426007a.html |journal=Nature |date=6 November 2003 |pages=7 |doi=10.1038/426007a }}

→ Giles, Jim (6 November 2003). "Preprint Server Seeks Way to Halt Plagiarists". Nature: 7.

doi:10.1038/426007a

.

But why should we add both DOI and URL, since DOI is essentially a permanent URL? --Bender235 (talk) 18:08, 28 April 2008 (UTC)

Because the Doi is not necessarily a link to a freely accessible version of the article. In such cases URLs are often used to link to a personal or institutional archive copy of the paper. Otherwise, it is indeed pretty pointless. URLs can also be used alongside a PMID (e.g. to a freely accessible journal where the PMID only has the abstract, and no DOI exists). Circeus (talk) 18:25, 28 April 2008 (UTC)

You're definitely right, if there's a freely accessible version of a paper we should use URL additionally to DOI. But in that case above, with that Nature article? We should probably change the example in the documentation. --Bender235 (talk) 22:11, 28 April 2008 (UTC)

I suspect the example was intended to show that the DOI resolves to the same url, but that information should be separate fro the example. Circeus (talk) 22:48, 28 April 2008 (UTC)

Agreed, a URL is only useful in addition to a DOI if a) it is not the same as you get when you follow a DOI; b) it is free to access. Smith609 Talk 15:40, 30 April 2008 (UTC)

But if there is a DOI, and no URL, then the title of the article should link to the DOI. This is for the benefit of the user. It's intuitively clear what clicking on a hypertext title means. Most readers will not know what a DOI is (I work for a high tech company, and this is my experience anyway. Random readers are probably even less likely to know.) Furthermore, the doi "looks" random, and users are taught to be wary about clicking on random looking links (these are often phishing or spam tracking links). So I'd say if there is a URL, link the title to that; if not and there is a DOI, then link the title to "dx.doi.org/DOI". This would IMO be in the best interest of the readers. LouScheffer (talk) 21:35, 30 April 2008 (UTC)

I agree that that should be the behavior (and that's what the editprotected request below is asking for). ASHill (talk) 21:45, 30 April 2008 (UTC)

No, the request below says "if no title is supplied". This should be the behaviour even if a title is supplied, I think. LouScheffer (talk) 21:49, 30 April 2008 (UTC)

I am absolutely opposed to hot-linking a DOI in a Wikipedia citation to an URL that redirects to a for-pay journal archive site. Wikipedia should not be the place to advertise commercial services like this, (especially when I can log in to my local public library's website and request a copy of the same article for free via Inter-Library Loan.) In fact, I am confused about the fundamental purpose of the DOI scheme. Is it (A) to conveniently identify a particular journal article (or other citation) online, or (B) to conveniently identify a website that allows one to purchase a copy of a particular journal article (or other citation) online from a particular vendor? I am asking this because I see DOIs for articles that long predate any computerized identification schemes. And from doi.org's FAQ #43,

What is being identified need not be the same as what is resolved to (for example, if I assign a DOI name to a physical book, I cannot resolve to that, but I can resolve to, e.g., an Amazon page or a MARC bibliographic record).

So whoever buys a DOI (and apparently they do cost money to buy) is free to link it to a for-pay site that offers for sale a copy of the resource identified. It looks to me more like a money-making scheme than a library classification/identification scheme. I would suggest trying to find a way to use the identification of the underlying resource of the DOI without linking to these commercial sites. Deepmath (talk) 08:11, 20 August 2008 (UTC)

No. You have to buy ISSNs (for periodicals) and ISBNs (for books), too. Still we use them, because there is no better identifier for a book than ISBN.

As for "commercial sites": yes, most scientific publishers (such as Wiley or Elsevier) request subscription/payments in order to read more than the abstract. But that doesn't mean Wikipedia shouldn't link to those websites. ––Bender235 (talk) 09:55, 20 August 2008 (UTC)

It depends on the publisher—not all DOI links require purchase. And if you are on a university campus (with many subscriptions), most of these link to the full article. For my collection of references, about 20% can link to the full article without paying, and when browsing at home. I personally find DOIs extremely functional, as they point you to the correct place to find the article abstract and metadata.+mt 15:11, 20 August 2008 (UTC)

Hi,

I've just configured my

To make it as simple as possible to add references to articles, could you please replace the source with the contents of Template:Cite journal/Edit?

This amendment has the following action:

• IF no title parameter is supplied, BUT a DOI is,

• THEN provide a link, in place of {{{title}}}, to a URL that will make DOI bot fill out the reference from the DOI.

Thanks,

Smith609 Talk 15:40, 30 April 2008 (UTC)

• A further update, as requested on my talk page: Please replace

{{{url}}}

with:

{{{url|{{#if:{{{doi|}}}|http://dx.doi.org/{{{doi}}}}}|{{#if:{{{pmid|}}}|http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=retrieve&db=pubmed&list_uids={{{pmid}}}&dopt=Abstract}}}}}

This will make the title wikilinked with a DOI, or PMID, if either parameter exists.

Thanks, Smith609 Talk 22:16, 30 April 2008 (UTC)

Clarification

Request one would allow editors to create a citation by simply pasing "cite journal | doi = 10.1092/whatever" into the article. The citation would then appear with a link to DOI bot, which would then dutifully pad out the citation with the article's title, author, journal etc.

Request two would wikilink the title with:

• If it exists, the URL parameter supplied
• If there's no URL parameter, the DOI destination
• If there's neither of these, the PMID destination

This is in addition to the DOI and PMID links appearing at the end of the citation.

Sorry not to be clearer in the first instance!

Smith609 Talk 22:37, 30 April 2008 (UTC)

No please do not enact this suggestion, for multiple reasons:

• PMID is always to just an abstract and DOI usually so, whereas the implication is that url is used only to link to a copy of the full article.
• To create a linked title to an abstract would hide that there might be a freely accessable online access to the whole paper that could be added as the url.
• PMID gives more than just an abstract, but also allows searching on authors other papers, related topics and what other papers have cited the paper subsequently. The PMID link is therefore useful even if title is linked to full copy by the url parameter
• Also would be silly to have title linked to DOI and then also at the end of the template show the DOI value, eg Smith A Example paper BMJ 123:10; 12-4
  doi:10.123
  duplicates the same link
  Likewise for the PMID link.
• Finally what bot is going to go through correcting "cite journal | doi = 10.1092/whatever" or "cite journal | pmid= 123456/whatever" ? That's what Diberri's tool is for surely ? David Ruben ^Talk 22:41, 30 April 2008 (UTC)

Hi, thanks for your input! Just to clarify, this suggestion would not remove the PMID or DOI links from the end of the citation. It was suggested that the duplicated link be added above, and on my talk page - I'm remaining neutral on that subject for now! The bot that's causing all the fuss is

User:DOI bot, and Diberri's tool is very limited in its scope - it did a poor job on a couple of articles I tested, and doesn't provide scope for DOIs. Also, I'd never heard of it before!! Smith609 Talk

22:51, 30 April 2008 (UTC)

Diberri's tool does give doi's where known to PubMed, eg see this random example. David Ruben ^Talk 22:58, 30 April 2008 (UTC)

Oops you mentioned your DOI bot above (but still I'd rather know if I'm going for an abstract or a carefully located full article link), but there is no equivalent for PMID. David Ruben ^Talk 22:48, 30 April 2008 (UTC)

PMIDs are never touched by the bot. And in my experience (alas), the URL parameter has very rarely been selected to point to a full article page! I guess the needs of medics and scientists are probably quite different. Smith609 Talk 22:53, 30 April 2008 (UTC)

Hmmmm sad if url so rarely used to a full version then. What happens in such citations, is the format parameter used to specify "abstract" then ? David Ruben ^Talk 22:58, 30 April 2008 (UTC)

... and if not then should that be a direction in the documentation on using this template ? David Ruben ^Talk 23:00, 30 April 2008 (UTC)

The format parameter is very rarely used at all, in my experience! I think that people are probably more prone to copy existing templates than to check the documentation. I am trying to work out whether I could adapt the bot to detect the appropriate "format" automatically - it may not be too difficult. Smith609 Talk 07:52, 1 May 2008 (UTC)

In astronomy articles at least, the url parameter often (though not always) points to the

template:arXiv

}} template is often used in the cite template to point to the arXiv entry as well. The lesson, I think, is not to make generalizations about the use of the template across fields, as it appears to me that the convention of using the URL parameter to point to a freely available, full text version of the article is common in medical articles (which I rarely edit). In general, if the url is specified, it really ought to point to the version that the editor used to get the information in the wikipedia article—sometimes, the abstract is enough as it does contain the cited information.

Using the format parameter might be a good indicator. However, I have two concerns. 1) Is that a proper use of the format field? 2) Saying 'abstract' for a doi varies depending on where you're accessing the internet; if you're at a library with access to the journal, the doi link is usually full text.

A doi link can at least be read in full if you go to a library or university with a subscription or site license to the journal. ASHill (talk) 14:52, 2 May 2008 (UTC)

• Additional thought http://pmid.us/ is not the true link for PubMed which gets redirected, see {{PMID}} which gives a fuller url address and parameters.David Ruben ^Talk 22:58, 30 April 2008 (UTC)

 Done: I've amended the URL to the one used at {{PMID}}. Smith609 Talk 07:46, 1 May 2008 (UTC)

I think the primary consideration should be what's best for the typical/casual user. By far the most natural place to click to get more information is the title (and DOI links, with their odd syntax, are the worst). If the full paper is available, that's best, but if not they'd presumably like to read at least the abstract. In the case where the full paper is available on-line for free, then the following works: the convention is that URL is only used when it points to a full paper, and the title links to URL, then DOI if no URL, then PMID if no URL or DOI, then does what the user wants - it points to the best known publicly available material. Nothing then prevents an editor who finds a full version on-line from adding a URL link to make it available, but until that point at least the reader can see the abstract.

The case is much less clear when the full-text version is only available on a pay-to-access site. Ideally what the user would like is a link to the full paper if they have access, otherwise only the abstract. The best way to do this would be to have the for-fee web sites default to the abstract if the user is not allowed full article access, but clearly we cannot count on this. Another might be to add another ID, such as 'pay-url='. Then the link could appear, but with a '$' symbol, and the abstract could be a separate link. Any other ideas?? LouScheffer (talk) 23:24, 30 April 2008 (UTC)

(Umm, $-symbol isn't entirely international. Subscription Required, Payment Required or Pay might be better.)

I've been going through "my" articles (the watched ones) and adding

• format= Abstract
• format= PDF: Full text
• format= HTML: Full text

as appropriate. I do wish everyone would do this. Perhaps to add as a suggestion in the instructions?

I prefer linked title for "ordinary" users. People at universities will know at a glance if DOI will give them full text. --Hordaland (talk) 23:07, 12 May 2008 (UTC)

Solution

Okay, I think I've got it. What if:

• If the template uses a DOI or PMID to link the title, and no format parameter is specified, it sets the format to display "DOI link" or something similar?

I think that solves all the qualms. Any problems with that implementation? Smith609 Talk 08:03, 2 May 2008 (UTC)

I like this implementation (assuming it will use the URL field for the title link if the URL field is provided). ASHill (talk) 14:55, 2 May 2008 (UTC)

It will do. As there haven't been any other issues with this, I'll request the edit. Smith609 Talk 07:57, 4 May 2008 (UTC)

I'm not sure I understand. If the url param specifies something like http://dx.doi.org/10.000/123p423, then it will display Synthesis and characterization of thingamajig (DOI link) (...)

doi:10.000/123p423

? So, you have effectively two hyperlinks with the same target in one line? If that's the case, I would rather your bot or anyone else simply not specify the URL at all, and add format = free/open access.

I would also think it is irresponsible to cite something off the abstract of a paper without having read it in full. There could be important caveats or exceptions listed in the full text.--Rifleman 82 (talk) 08:29, 4 May 2008 (UTC)

Certainly true usually, but sometimes for a very broad point that's common knowledge in the field, an abstract is OK to verify the point on Wikipedia. More importantly, if the writer of the Wikipedia point had access to the full article, an abstract is often enough for another reader to verify the citation, and as Smith says, an abstract link is surely better than nothing.

I personally would favor eliminating the displayed DOI in favor of having the title alone link to the DOI when no URL parameter is specified. However, there was a consensus against that when I proposed it here a few months ago. I'm agnostic on whether the title should link to the DOI if the DOI is specified and displayed; I just think having the URL parameter explicitly set to link to the same page as the DOI is silly, redundant, and prone to breaking. ASHill (talk | contribs) 15:00, 4 May 2008 (UTC)

Yes, there would be two hyperlinks; the advantage to having the title hyperlinked is that it is intuitive for readers to click on the title to view the article, but not so obvious what the DOI link means. I agree that it doesn't seem entirely necessary, but a lot of people seem to feel quite strongly that it is useful, and I see no point in denying them their request on personal/aesthetic grounds - it seems there's a deal of agreement on my talk page that the title should link to something.

I'm not sure where your second point comes in; a DOI link is not necessarily free or open access; I agree that the full text should always be read before citing a source, but sometimes only abstracts are available online / for free, and a link to the abstract of a paper is much more useful than no link at all. Smith609 Talk 11:47, 4 May 2008 (UTC)

{{

{{{url}}}

}}{{#if:{{{format|}}}
  |  ({{{format}}})

{{{url|{{#if:{{{doi|}}}
  |http://dx.doi.org/{{{doi}}}
  |{{#if:{{{pmid|}}}|http://www.ncbi.nlm.nih.gov/sites/entrez?cmd=retrieve&db=pubmed&list_uids={{{pmid}}}&dopt=Abstract}}
}}}}}

}}{{#if:{{{format|{{#ifeq:{{{url|{{{doi|{{{pmid}}}}}}}}}|{{{doi|{{{pmid}}}}}}|has no URL but a DOI or PMID|}}}}}
  |  ({{{format|{{#if:{{{doi|}}}|DOI link|PMID link}}}}})

 Done You removed a URL-encoded space that I suspect was there for a reason, so I re-added it: if it was genuinely unnecessary, just say the word and I'll take it out. Happy‑melon 19:08, 4 May 2008 (UTC)

Hmm; this has added the (DOI link), but did not make the title a DOI link. e. g.

• J. Jeff Hester (2004). "The Cradle of the Solar System" (DOI link). Science. doi:10.1126/science.1096808. 

It also incorrectly says (PMID link) when none of the url, DOI, or PMID parameters are included:

• J. Jeff Hester (2004). "The Cradle of the Solar System" (PMID link). Science. 

ASHill (talk | contribs) 19:17, 4 May 2008 (UTC)

Reverted, until you work out what's gone wrong and how to fix it. Happy‑melon 19:22, 4 May 2008 (UTC)

Given the recent concern raised at

WP:AN#DOI bot blocked for policy reconsideration, I think we should hold off on making this change (properly implemented) pending the outcome of that policy discussion. ASHill (talk | contribs

) 21:23, 4 May 2008 (UTC)

I've now corrected the code - I misplaced a brace. Smith609 Talk 07:54, 5 May 2008 (UTC)

As links now automatically display whether they point to a PDF, is the "format=PDF" necessary? My sense from recent discussion is that people would find it more useful if they knew where the link was pointing: for example, to an abstract, a freely available full text, and so on. I won't amend the doc page here until I'm sure there's consensus! Smith609 Talk 09:48, 2 May 2008 (UTC)

I agree that format=PDF does not help much (if at all - if there was a PDF format file without the .pdf extension, I don't think the browser would know what to do with it anyway). I also agree that it would be an excellent idea to know what the link points to (pay article, free article, abstract, etc.), and display these differently (as some web sites add $$$ to links that point to pay pages - I think yahoo financial news used to do this). I'd vote for making a new parameter, though, and not re-using 'format=', since 'format' is not very intuitive for this use. LouScheffer (talk) 13:54, 2 May 2008 (UTC)

Okay. We could easily retain backwards compatibility too until a bot corrects them all. Any suggestions for an intuitive parameter name? Smith609 Talk 14:54, 2 May 2008 (UTC)

The documentation for Template:cite news explicitly suggests using the format parameter for identifying 'fee required' links, so maybe format is the right parameter to use. ASHill (talk) 15:00, 2 May 2008 (UTC)

They are doing this because it does what they want, not because it's the right thing to do, IMO. They are using 'format' as 'string-to-appear-after-link'. There is nothing wrong with the function, but it's not intuitive to overload a parameter this way, and it will take new editors longer to figure out how this works. (For example, I've been an editor for several years and was unaware of this use of format until you told me.) Also, if you do it this way it will appear differently in different articles. Since there are really two issues - what type of thing the link points to, and whether it requires payment, I'd actually prefer two fields, each with a more descriptive name. Perhaps 'url_type={abstract, paper, pre-print}' and 'cost={free, fee required}', or something similar, with defaults of 'paper' and 'free'.

There is also the case (commonly in physics) where the article is pay, but the pre-print is free (from arXiv, typically}. The way I handle this now is <ref>{{cite journal |url= |doi=...} Pre-print [http://arXiv.org/... here].</ref> but it would be nice if there was a standardized way of doing this as well. However this requires 2 URLs. LouScheffer (talk) 15:50, 2 May 2008 (UTC)

Perhaps the thing to do is to replace the "url" parameter with something along the lines of "free-url" and "fulltext-url"? Smith609 Talk 16:12, 2 May 2008 (UTC)

As long as the url parameter still works for backwards compatibility and simplicity, of course. However, that only works if a URL is being provided, which isn't necessary when using doi, arXiv, and PMID. ASHill (talk) 16:59, 2 May 2008 (UTC)

Fair enough; I wasn't aware of that usage either until I looked while composing the above comment. (Meta-comment: The variety of cite templates, each with their own quirks, are a huge mess!)

I agree that format= isn't intuitive; url_type sounds good. However, saying 'abstract' is difficult (as I mentioned above) as, for DOIs, it varies based on whether you're at a library or university with a site license to the journal. Using url_type=ADS for

NASA ADS entries would also be a good use of that field, for example. I like the cost= proposal very much, except that a doi link does provide useful information (the abstract) even if you can't read the whole article for free. — Ashill

16:23, 2 May 2008 (UTC) — continues after insertion below

If no URL was specified (but a DOI was), url_type could default to "DOI"? I think a DOI always links to the same page anyway - but you may get a "no entry" page if you don't have access. Smith609 Talk 18:05, 2 May 2008 (UTC)

For every journal I know, you at least get the abstract from a DOI. That default would be fine by me, although I'm concerned that this proposal will make references rather crowded. ASHill (talk) 22:23, 2 May 2008 (UTC)

This is my experience as well. I don't know any sites that charge for abstracts, or block access if you have not paid. (Of course if anyone knows differently, please speak up!) I think we can count on a DOI link to give at least the abstract. LouScheffer (talk) 22:33, 2 May 2008 (UTC)

Incidentally, for arXiv preprints, using {{cite journal | author,title,etc | id={{arXiv|0805.0001}} }} is a nice way to produce consistent output, and won't require fixing links if the arXiv decides to change their URL format. ASHill (talk) 16:23, 2 May 2008 (UTC)

{{

Last, first (1977). "Title". The Journal of Stuff. 5 (2). The Publishers of Stuff: 22–23. {{cite journal}}: |access-date= requires |url= (help)

Current rendering:

...."Title". The Journal of Stuff5 (2):22-23. The Publishers....

Correct rendering:

...."Title". The Journal of Stuff. 5(2):22-23. The Publishers....

--emerson7 00:05, 3 May 2008 (UTC)

 Not done, the spacing looks okay in all of my tests on Firefox and IE; if you give me the code for an example that doesn't work I'll take a look again and try to fix it. Also, I did a quick look at APA and MLA styles for journal citations, and neither uses a full stop (though this form doesn't seem to fit either style perfectly). MLA uses nothing and APA uses a comma, so I'm just going to leave it alone for now until we see some consensus to change it. --CapitalR (talk) 17:53, 3 May 2008 (UTC)

I left a note at

We already do - check the talk page archives or template source code for more information. Smith609 Talk 11:58, 6 May 2008 (UTC)

Really? Besides of <cite>…</cite> I can't see no microformat. Where is it? —Bender235 (talk) 22:47, 6 May 2008 (UTC)

There is no agreed-upon "true" microformat for citations. hCite is being developed, but very slowly & there's almost no adoption. WP citation templates use COinS, which are more separated from the displayed text than so-called "true microformats," but are machine-readable (try out Zotero and/or LibX). --Karnesky (talk) 00:18, 7 May 2008 (UTC)

Okay, so I guess we wait until there's something like hCite. —Bender235 (talk) 08:58, 7 May 2008 (UTC)

BTW: I've just spotted Wikipedia:WikiProject Microformats#Citations. Explains everything. —Bender235 (talk) 10:30, 9 May 2008 (UTC)

{{

Please replace {{{doi|{{{doilabel|}}}}}} with {{urlencode:{{doi}}}}}.

This renders the doilabel parameter obsolete, allowing the template to generate a URL-friendly DOI itself.

Smith609 Talk 12:57, 6 May 2008 (UTC)

 Done, -CapitalR (talk) 13:18, 6 May 2008 (UTC)

This function doesn't quite work for DOIs containing square brackets. In solving one problem, it's exposed another. There are four solutions:

1. We simply undo the edit, returning tot he bad old days of having to specify DOI labels, and having lots of broken links where editors don't realise this;
2. We don't link the text of the DOI, but instead provide an invisible link after it. example: doi:10.1082/asduy39871 
3. Alternatively, if a DOI is specified, but a URL is not, we link the title to the DOI; if a URL is specified then we would have to either not display the DOI link, or have the "invisible" link in 2.
4. Or, we put up with working links which look messy, like this one: doi:10.1012/example[squarebrackets]

Which is preferable? Smith609 Talk 18:36, 6 May 2008 (UTC)

It turns out that these DOIs didn't work, even in the days of doilabels. I've implemented a fix at {{doi}} which will do the job here for now, at least until anyone comes up with a more elegant solution!

{{

Replace:

[[Digital object identifier|doi]]:[http://dx.doi.org/{{urlencode:{{{doi}}}}} {{{doi}}}]

With:

[[Digital object identifier|doi]]:{{#ifeq:{{doi/encode|{{{doi}}}}}|{{{doi}}}|[http://dx.doi.org/{{urlencode:{{{doi}}}}} {{{doi}}}]|{{{doi}}}[http://dx.doi.org/{{urlencode:{{{doi}}}}}  ]}}

Result:

Most "normal" dois remain the same:

Before: doi:10.1010/test[squarebrackets]

After: doi:10.1010/test[square]brackets 

Thanks, Smith609 Talk 10:24, 7 May 2008 (UTC)

 Done, please double check that the new code is correct. I think you intended for an "nbsp;" to be in place of one of the spaces so I made that change; let me know if that wasn't what you intended. --CapitalR (talk) 10:48, 7 May 2008 (UTC)

Argh, the hazards of copying and pasting. Didn't realise the nbsp didn't show up in the pre code. I also forgot to replace {{{1|{{{id}}}}}} with {{{doi}}} in the "new" code. I've corrected it above, would you mind amending to the new version, as it renders abover? Thanks. Smith609 Talk 11:22, 7 May 2008 (UTC)

Okay, should be all set now. Let me know if there's anything else. --CapitalR (talk) 11:35, 7 May 2008 (UTC)

This doi doesn't have a square bracket and yet shows up with a postlink. Why? Is this avoidable? ASHill (talk | contribs) 15:59, 9 May 2008 (UTC)

• M. J. Fogg. "Long title".
  doi:10.1051/0004-6361:20066171. {{cite journal}}: Cite journal requires |journal= (help
  )

It's the colon that does it - unfortunately, wikipedia's parser functions are not really designed for text manipulation, so any DOIs containing special characters (except slashes and dashes) will display with a postlink. I can't think of any way around this, unfortunately! Smith609 Talk 16:12, 9 May 2008 (UTC)

I don't understand why people apparently believe these "post-links" are acceptable. They're inaccessible to the reader. This issue is not limited to "square brackets" as claimed above. Parentheses cause the same issue:

{{cite journal|first=R.|last=Lahana|title=How many leads from HTS?|journal=Drug Discovery Today|year=1999|issue=4|pages=447-448|doi=10.1016/S1359-6446(99)01393-8}}

produces:

Lahana, R. (1999). "How many leads from HTS?". Drug Discovery Today (4): 447–448.

doi:10.1016/S1359-6446(99)01393-8

.

--

Talk

00:08, 10 May 2008 (UTC)

This should be corrected by changing:

{{#if:{{{doi|}}} |. [[Digital object identifier|doi]]:{{#ifeq:{{doi/encode|{{{doi}}}}}|{{{doi}}}|[http://dx.doi.org/{{urlencode:{{{doi}}}}} {{{doi}}}]|{{{doi}}}[http://dx.doi.org/{{urlencode:{{{doi}}}}}  ]}}

to:

{{#if:{{{doi|}}} |. [[Digital object identifier|doi]]:{{#ifeq:{{doi/encode|{{{doi}}}}}|{{{doi}}}|[http://dx.doi.org/{{urlencode:{{{doi}}}}} {{{doi}}}]|[http://dx.doi.org/{{urlencode:{{{doi}}}}} {{urlencode:{{{doi}}}}}]}}

I find it possible to use postlinks by clicking on the link icon. Does this not work for you? A possible workaround would be to instead postlink a very small word "go" after the DOI, thus: doi:10.1039/te[eph]1298.as/aw34 go

The alternative solution is percent-encode the characters <>[] when they appear in a DOI, and remove the postlink code.

A bot could update existing DOIs. Then all DOIs would work in the conventional style. Smith609 Talk

13:48, 10 May 2008 (UTC)

That's ugly; it means you can't copy and paste the doi from the paper you're citing or a BibTeX file, and typing character symbols is a good way to introduce human error. Couldn't the template do that encoding automatically, on the fly? ASHill (talk | contribs) 13:58, 10 May 2008 (UTC)

Alas, there's no way for the template to do it. I think the best solution would be to encourage editors to enter links manually, and for DOI bot to keep watch for non-encoded DOIs (which would still work, the link would just look ugly) and replace them with the percend-encoded versions as soon as it notices them. Smith609 Talk 14:11, 10 May 2008 (UTC)

OK. I guess that is a job for DOI bot. (Should that be explicitly added to the request for approval while it's still open, or does that fit within "The bot may fix a syntactically broken DOI or PMID reference"? ASHill (talk | contribs) 15:39, 10 May 2008 (UTC)

No harm in noting it, I guess. A shame it has to be done by bot, but I suspect the MediaWiki software will be upgraded at some point in the future... Smith609 Talk 19:01, 10 May 2008 (UTC)

Confusion with DOIs containing parentheses

An editor recently complained that DOIs with parentheses were not working. I tried it and also thought they were not working, because they were displayed as black. I didn't notice the tiny hyperlink after them, and I suspect most readers won't notice them either.

Why are parentheses troublesome in DOIs? I can see some other characters being troublesome (e.g., <) but I don't see why parentheses are troublesome.

If this can't be fixed in the template, what can an editor do to work around the problem? Use doilabel=, perhaps? It's pretty confusing as it stands. Eubulides (talk) 21:57, 19 May 2008 (UTC)

See the discussion above this post. A bot is almost ready to get to work percent-encoding the parameters so that all DOIs can be rendered correctly by the template; there didn't seem to be any feeling either way about putting a mini "go" link after character-including DOIs, but that could be employed for the moment if you think it a good idea. Smith609 Talk 07:49, 20 May 2008 (UTC)

{{

While you're making an edit (to reduce server load, edits should be performed at the same time if possible) the change listed below will do the following:

• If a citation is left without a title (the one mandatory parameter)
• And the citation has a a DOI or URL parameter specified

• Then instead of displaying {{{title}}} in place of the missing title parameter, the citation will display a clickable link that will bring the page to a bot's attention; when this link is clicked, the bot will complete the missing information.

Replace:

{{{title}}}

With:

{{{title|{{#if:{{{doi|{{{url|}}}}}}|<span class=error>Citation is missing a {{{title}}}.
Either specify one, or [http://tools.wikimedia.de/~verisimilus/Bot/DOI_bot/doibot.php?edit=on&user=Cite_journal&page={{FULLPAGENAMEE}} click here]
and a bot will complete the citation details for you.}}}}}</span>

Note: The double E in FULLPAGENAMEE is not a typo! It URL-encodes the page name.

Thanks a lot!

Smith609 Talk 17:46, 22 May 2008 (UTC)

Addendum: Another change that needs making is to place the DOI link in a neverexpand tag, as it looks incredibly messy in print view. So instead of undoing the first edit, please:

Replace

{{#ifeq:{{doi/encode|{{{doi}}}}}|{{{doi}}}|[http://dx.doi.org/{{urlencode:{{{doi}}}}} {{{doi}}}]|{{{doi}}}[http://dx.doi.org/{{{doi}}}  ]

With

<span class="neverexpand">[http://dx.doi.org/{{urlencode:{{{doi}}}}} {{{doilabel|doi}}}}}}]</span>

Cheers,

Smith609 Talk 10:34, 23 May 2008 (UTC)

Hm. A couple of users have pointed out that even with the above edit, a doilabel or postlink will still be required. As the post link didn't please everyone and a bot can add a doilabel, I'll opt for that alternative, and have updated the code above accordingly. Smith609 Talk 16:11, 24 May 2008 (UTC)

Editprotected clarification

I thought I'd post a clarification as the above has become a little confusing. Please: {{

Replace

{{{title}}}

With

{{{title|{{#if:{{{doi|{{{url|}}}}}}|<span class=error>Citation is missing a {{{title}}}.
Either specify one, or [http://tools.wikimedia.de/~verisimilus/Bot/DOI_bot/doibot.php?edit=on&user=Cite_journal&page={{FULLPAGENAMEE}} click here]
and a bot will complete the citation details for you.}}}}}</span>

Note: The double E in FULLPAGENAMEE is not a typo! It URL-encodes the page name.

Replace

{{#ifeq:{{doi/encode|{{{doi}}}}}|{{{doi}}}|[http://dx.doi.org/{{urlencode:{{{doi}}}}} {{{doi}}}]|{{{doi}}}[http://dx.doi.org/{{{doi}}}  ]

With

<span class="neverexpand">[http://dx.doi.org/{{urlencode:{{{doi}}}}} {{{doilabel|{{{doi}}}}}}]</span>

 Not done for now. The first change will result in an empty string being legitimately output as the title when no url or doi is specified, instead of an obvious error. There's a brace imbalance in the second change - is this intentional? Happy‑melon 10:40, 25 May 2008 (UTC)

Fixed the brace imbalance. Those things have a habit of disappearing, no matter how carefully you check...

To address the "blank title is absent" issue, the below code would display an error on an empty string:

{{#ifeq:{{{title|}}}||{{#if:{{{doi|{{{url|}}}}}}|<span class=error>Citation is missing a {{{title}}}.
Either specify one, or [http://tools.wikimedia.de/~verisimilus/Bot/DOI_bot/doibot.php?edit=on&user=Cite_journal&page={{FULLPAGENAMEE}} click here]
and a bot will complete the citation details for you.|{{{title}}}}}|{{{title}}}}}</span>

Like any good workman, I'm blaming Notepad++ if these braces don't balance!

Thanks for looking over these edits. Smith609 Talk 19:52, 25 May 2008 (UTC)

 Done There is still a brace imbalance in that second change, but I think I worked out what you're trying to change :D. I hacked the second change around a bit to avoid redundancy (it was checking for a |url= parameter when the code would only be executed if such a parameter was defined). Any problems, revert and let me know. Happy‑melon 19:20, 26 May 2008 (UTC)

Please contribute to this discussion at Citing sources: Wikipedia talk:Citing sources#Retrieval dates for online versions of old printed sources, again --EnOreg (talk) 16:18, 26 May 2008 (UTC)

Consensus: We have a consensus that access dates for online copies of offline sources, while helpful as a comment in the source, should be hidden from the reader. Could somebody who is competent to adapt the citation templates please do so? The idea is to keep the access date as a template parameter but remove the code that displays it. Thanks, --EnOreg (talk) 09:21, 5 June 2008 (UTC)

Seems to be fixed now. Please check and let me know if there are any problems. --Rifleman 82 (talk) 15:32, 5 June 2008 (UTC)

That works. Many thanks! I've updated the template documentation. --EnOreg (talk) 16:06, 5 June 2008 (UTC)

I tried using the new pmc= parameter on

I like the citation style where the article's title is linked if the article's body is freely readable, and otherwise the title is not linked (readers can follow the DOI or PMID instead). This makes it easier for the reader to tell when the article is free. Assuming this style, please consider this citation currently in

• Volkmar FR, Chawarska K (2008). "Autism in infants: an update". World Psychiatry. 7 (1): 19–21.
  PMID 18458791
  .
• {{cite journal |journal= World Psychiatry |date=2008 |volume=7 |issue=1 |pages=19–21 |title= Autism in infants: an update |author= Volkmar FR, Chawarska K |pmid=18458791 |url=http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2366821}}

• Volkmar FR, Chawarska K (2008). "Autism in infants: an update". World Psychiatry. 7 (1): 19–21.
  PMID 18458791
  .
• {{cite journal |journal= World Psychiatry |date=2008 |volume=7 |issue=1 |pages=19–21 |title= Autism in infants: an update |author= Volkmar FR, Chawarska K |pmc=2366821 |pmid=18458791 |url=http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=2366821}}

• Volkmar FR, Chawarska K (2008). "Autism in infants: an update". World Psychiatry. 7 (1): 19–21.
  PMID 18458791
  .
• {{cite journal |journal= World Psychiatry |date=2008 |volume=7 |issue=1 |pages=19–21 |title= Autism in infants: an update |author= Volkmar FR, Chawarska K |pmc=2366821 |pmid=18458791}}

Here is a suggestion for improvement. Can you please alter Template:Cite journal so that if pmc= and title= are specified but url= is not, the template applies the link to the article title rather than outputting text of the form "PMC:2366821"? That way, I'd get the following behavior:

• Volkmar FR, Chawarska K (2008). "Autism in infants: an update". World Psychiatry. 7 (1): 19–21.
  PMID 18458791
  .
• {{cite journal |journal= World Psychiatry |date=2008 |volume=7 |issue=1 |pages=19–21 |title= Autism in infants: an update |author= Volkmar FR, Chawarska K |pmc=2366821 |pmid=18458791}}

Thanks. Eubulides (talk) 21:24, 5 June 2008 (UTC)

Agree (see also above

Template talk:Cite journal#Pubmed Central

).

I'll mention, and immediately vote-down an extra possibility for hiding the pmc at the end of the citation: if both the url & pmc are specified, then there are two links to free full text versions of the article (one the url-linked title, the other the pmc value at the end) - one could in this circumstance hide the end pmc? However this gets overly complicated very quickly and we have no problem having a url-linked title and later a doi value that may link to a free-full text verion.

In my original series of proposals I also asked if the format parameter should be part-filled when the title is linked to the pmc (ie where the link is not to the primary source of the journal itself but to an offsite repository of PubMed Central). I can now show how that might work:

{{cite journal |journal= World Psychiatry |date=2008 |volume=7 |issue=1 |pages=19–21 |title= Autism in infants: an update |author= Volkmar FR, Chawarska K |pmc=2366821 |pmid=18458791}}

→Volkmar FR, Chawarska K (2008). "Autism in infants: an update" (

PMID 18458791

.

This would still allow an editor to use the format parameter itself:

{{cite journal |journal= World Psychiatry |date=2008 |volume=7 |issue=1 |pages=19–21 |title= Autism in infants: an update |author= Volkmar FR, Chawarska K |pmc=2366821 |pmid=18458791 |format=PDF}}

→Volkmar FR, Chawarska K (2008). "Autism in infants: an update" (PDF

PMID 18458791

.

David Ruben ^Talk 23:29, 5 June 2008 (UTC)

Thanks. Re the last proposal, I'd rather not see the string "(PMC copy)". The reader won't care whether the title links to a PMC copy or to some other copy, and it's better to not waste so much valuable screen real estate on an irrelevant detail. Eubulides (talk) 06:36, 6 June 2008 (UTC)

I usually include "free full text" when necessary in the "format" parameter. How about that? --Rifleman 82 (talk) 10:19, 6 June 2008 (UTC)

That is a reasonable style for some articles, but it's not the compact medical style I prefer, as it clutters up the screen. Please see

Autism is not using pmc= because of the abovementioned problems. Eubulides (talk

) 17:57, 6 June 2008 (UTC)

So is this a consensus for linking the title to the full free-text of PMC provided that the url has not been specified, making no special comment in format parameter, and then hiding the PMC:xxxx at the end so as not to duplicate displayed links (PMC:xxxx only shown therefore if the title has been linked with a specified url) ? This will be easy for me to code :-) David Ruben ^Talk 18:16, 6 June 2008 (UTC)

Yes, that's what I'm asking for. I'm glad it's easy to code. I don't know about "consensus" but I don't hear any dissents either. Eubulides (talk) 01:32, 8 June 2008 (UTC)

PS Diberri just allowed pmc value itself to be used by his Wikipedia template filling tool, see example :-) David Ruben ^Talk 22:40, 6 June 2008 (UTC)

 Done <undent> Ok title linked by pmc if url not specified, in which case PMC not then stated at end of the citation (as would a duplication of the link).

Hover mouse over the linked title of "test" to see if to setting url=example.org or pmc=123456

1. "test". {{cite journal}}: Cite journal requires |journal= (help) - links to url, pmc blank
2. "test".
   PMC 123456. {{cite journal}}: Cite journal requires |journal= (help
   ) - links to url, so pmc listed separately
3. "test".
   PMC 1234565. {{cite journal}}: Cite journal requires |journal= (help
   ) - links to pmc, url is blank
4. "test".
   PMC 1234565. {{cite journal}}: Cite journal requires |journal= (help
   ) - links to pmc, url empty
5. "test". {{cite journal}}: Cite journal requires |journal= (help) - links to url, pmc empty
6. "test". {{cite journal}}: Cite journal requires |journal= (help) - no link as url & pmc blank
7. "test". {{cite journal}}: Cite journal requires |journal= (help)- no link as url & pmc empty

David Ruben ^Talk 03:53, 8 June 2008 (UTC)

Here is a full example

{{cite journal |author=Dworkin J, Losick R |title=Does RNA polymerase help drive chromosome segregation in bacteria? |journal=Proc. Natl. Acad. Sci. U.S.A. |volume=99 |issue=22 |pages=14089–94 |year=2002 |month=October |pmid=12384568 |pmc=137841 |doi=10.1073/pnas.182539899 |url=}}

Dworkin J, Losick R (2002). "Does RNA polymerase help drive chromosome segregation in bacteria?". Proc. Natl. Acad. Sci. U.S.A. 99 (22): 14089–94.

Let me know if any problems David Ruben ^Talk 03:57, 8 June 2008 (UTC)

apparently the last changes have broken the template's accessdate field.

Arthur Writer (1 January 2000). "Naming names". The Journal of Journals. 30 (3): 0. {{cite journal}}: |access-date= requires |url= (help) ← ???

--emerson7 00:13, 9 June 2008 (UTC)

No not the last change of linking title to pmc parameter (see edit history), but instead accessdate coding was commented out (<!-- -->) by preceeding edits of Rifleman 82 with edit summary "comment out accessdate per talk" as per preceeding request thread of

Template talk:Cite journal#Retrieval dates for online versions of old printed sources, again David Ruben ^Talk

01:22, 9 June 2008 (UTC)

oic....sheeesh, what a dumb decision, consensus or not. thanks anyway, cheers! --emerson7 20:02, 13 June 2008 (UTC)

This change by the DOI bot to

DOIs are supposed to be stable and I suppose they are worth documenting, but as this example shows, they do not always work as a link. For Philitas of Cos I temporarily worked around the problem with this hack, which comments out the DOI and disables the DOI bot for the entire article (so that it doesn't re-add the broken DOI later). But this is overkill: I want the DOI bot to work in general and I don't mind if the DOI displays; I just don't want the DOI to display as a link that can be followed (because the link won't work).

I asked about this on

An alternative way of implementing this would be to leave the doi parameter as the "doi", but to add a parameter called something like "don't_link_the_doi" which could be set in the instances where the doi parser doesn't recognise the DOI,. Smith609 Talk 17:23, 10 June 2008 (UTC)

Yes, that was the intent for the doi_broken proposal. Sorry I wasn't clear. The idea is that doi= still specifies the DOI as before, but doi_broken=foo says "Please don't link from the DOI text, and put a comment saying 'foo' after the DOI text". Eubulides (talk) 17:33, 10 June 2008 (UTC)

Rather than showing a dead doi at all (it is of no practical use to the general reader), but accepting it is of interest to us editors/bots who might seek to provide a doi link, why not have a dummy value for the doi parameter that the template ignores and the bot does not overwrite (vs the "hack" mentioned above of disabling the doi bot from the article as a whole). No need then for an additional doi_broken parameter. This would work (I presume) if the bot does not overwrite the doi parameter where a value has already been provided. So I propose that a dead doi link is coded as "deadlink" and then in hidden tags the old doi value is noted for editors.

{{... |doi=deadlink<!-- 10.1093/cq/46.1.308 --> }}

• The bot does not overwrite the specified parameter value (and so leaves the "deadlink" entry as it is).
• We get this template to treat "deadlink" as if there is no doi value specified at all (a simple coding change).
• Readers would then not be shown a deceased doi value nor its link, yet editors can see what has occured.David Ruben ^Talk 21:04, 13 June 2008 (UTC)

This suggestion also fixes the problem, and would be fine with me as well. I mildly prefer the doi_broken proposal, though, as it defers to the template the question about whether to display broken DOIs. Eubulides (talk) 21:24, 13 June 2008 (UTC)

Broken DOIs will hopefully be fixed in the long run; including them will help editors who copy the text of a reference from Wikipedia to another work. Smith609 Talk 15:31, 15 June 2008 (UTC)

Smith609, I think you slightly misunderstand the issue – it is not of the doi values being bad (they are themselves correctly written out), but that the doi-links are nolonger supported by their original web-target. If a journal nolonger supports DOIs, there may be no expectation that they will be "fixed in the long run" and allowing editors to copy the reference details to another work is not helping anyone. What is needed is to indicate what the DOI used to be, and somehow indicate that it is a broken value (therefore just as useless for our readers as for any other work). The question is how to flag up broken DOIs – with a separate doi_broken parameter being set, or use of dummy "broken" value for the doi parameter ? David Ruben ^Talk 18:04, 15 June 2008 (UTC)

• The impression I get from CrossRef, who maintain the DOI database, is that all DOIs will eventually to be fixed; the bot now notifies them of any "broken" DOIs for their attention. As you say, the issue is how to reflect that a DOI still exists, that machines can still use it to deduce the metadata for a citation, but that humans cannot use it to find a URL. One rational way would be to provide the DOI but not hyperlink it (although this may confuse users). However, it is then more laborious to check at a later date whether or not the repair has been made. I don't really mind what you settle on, so long as the "broken" DOI can remain visible, as I find this very useful. Smith609 Talk 09:02, 16 June 2008 (UTC)

Thanks Smith609 for explanation (so if another company takes over a journal's back catalogue, would the same doi's become valid or would new doi values be generated to link to a new location for the old papers?). Thanks also for observation that even broken links useful to be displayed for some readers (vs above discussion tending to think of ways of hidding broken doi values), and just needing to remove the hyperlinking – this would then need to use a separate "doi_broken" parameter.

Quick question then, for non-hyperlinked broken-doi's should the template display an explanation, say "(broken doi)" or better a less agressive "(inactive doi)" or "(inactive)" ?

eg doi: 10.1093/cq/46.1.308 (inactive) David Ruben ^Talk 18:31, 16 June 2008 (UTC)

Before I code up for "doi_broken", do we want it to take a fixed affirmation value (eg "broken") or for editors to set it as the date they become aware that the doi was broken (eg "2008-06-17"), presumeably identical to the date they edit the article setting the "doi_broken" parameter. This would help inform editors/bots when not to retest whether the doi link is still broken (a link noted as being broken 2 years ago is worth looking into, that set as broken 2 days ago is almost certainly still so). A date value would help future editors who would know which day to look at in the article's edit history to find teh edit that had set the "doi_broken" parameter – one would hope the relevant edit summary might give further information. If we do go with the date option, we probably should specify in the template documentation to use fixed ISO dates yyyy-mm-dd (as per "accessdate"). David Ruben ^Talk 22:28, 17 June 2008 (UTC)

Great idea; perhaps you would also consider automatically including the article within a category of "DOIs which have been broken since July 2008" etc, so bots can easily find the oldest broken DOIs and re-test them? Smith609 Talk 22:41, 17 June 2008 (UTC)

Ok 2 more queries on developing the necessary code

1. It struck me if doi_broken is to be date that an editor notes that the doi is broken, would it not be better to be more explict and name the parameter "doi_brokendate"?
   Hence: |doi=10.1093/cq/46.1.308 |doi_brokendate=2008-06-18
   Gives: doi:10.1093/cq/46.1.308 (inactive 2008-06-18)
   (the doi value not being linked as it has been specified that the link is broken)
2. Do we want to categorise on full month & year (eg "July 2008" as per your example), or just the year (ie "2008") ? David Ruben ^Talk 23:46, 17 June 2008 (UTC)

• I like having the name end in "date", so "doi_brokendate=" sounds like a win.
• doi_brokendate= is closer in function to accessdate= than it is to the date= parameter of (say)
  Template:Fact
  , because it is talking about access or lack-of-access to an external resource, not about an editing issue. So I think its value should use ISO format (e.g., "doi_brokendate=2008-06-18") rather than year or month-year format.

Eubulides (talk) 03:09, 18 June 2008 (UTC)

 Done "doi_brokendate" added, takes ISO yyyy-mm-dd dates. Results in doi value not being linked, adds comment when inactive and adds to "Category:Pages with DOIs broken since YYYY" using just yearly-groupings (for the moment). Do let me know if I've messedup, especially re the Category coding.

Here are some test examples for the displayed text.

• Note examples check for emty and blank use of doi_brokendate parameter
• I've included issn parameter as this is next item that gets coded and need make sure no trailling extra spaces or carriage returns from my coding
• The adding to "Category:Pages with DOIs broken since YYYY" needs to be separately demonstrated, as set up only to so add if in mainspace article:

David Ruben ^Talk 12:37, 18 June 2008 (UTC)

For a mainspace example to show the addition of a category, lets use the example Eubulides gave at the start of this thread: Philitas of Cos. So see this edit removing need to inactivate the doi-bot, use of the parameter and adding to a yearly (for now) category. Let me know if people would prefer the categorisation to be by "Month YYYY". David Ruben ^Talk 12:49, 18 June 2008 (UTC)

Thanks very much for getting this to work; it should solve the problem I encountered in

Philetas of Cos and have encountered elsewhere (now I just have to remember where). I don't have a preference about whether the month is in the category; on general laziness principles, perhaps we should just leave things be, and insert the month later if the category proves to be too populated and unweildy (which I hope doesn't happen). Eubulides (talk

) 15:04, 18 June 2008 (UTC)

I would argue that if the category becomes so populated that a month is needed, then the DOI probably isn't living up to its promise as a permanent identifier of the content. Fortunately, broken DOIs are pretty rare. Although this is considerably harder to implement (and probably not worth the effort), categorizing by journal might make fixing broken DOIs easier in the future (on the assumption that all DOIs for a given journal will either work or not work). ASHill (talk | contribs) 15:42, 18 June 2008 (UTC)

Coding-wise very easy to place into "Category:Pages with DOIs broken from {{{journal|}}}", with the doi_brokendate used if no journal name specified. Reasoning for that is a good one – so people let me know now if this is thought a better system... but I can think of 2 problems without simple solutions:

1. Everyone would have to use the same style of specifying the journal name (?full name, partial or compact abbreviations, with/without full stops). Using Diberri's tool that uses PubMed will give consistant abbreviations, but anyone coding manually is likely to use full journal name. e.g. "Australian Pharmacist" and "Aus.Pharm." and "Aus. Pharm." are separate string values.
2. Even if consistant journal name format, would still be hard to find the category pages. Currently by year is easy (we can search on Category....2008 and next year Category....2009 etc), but how would one know to try looking for "Category:Pages with DOIs broken from Am J Psychol Rehab" or "Category:Pages with DOIs broken from Australian Pharmacist" – presumeably would need a bot to constantly search for "Category:Pages with DOIs broken..." and to each of these add a master category of "Category:Pages with DOIs broken". David Ruben ^Talk 19:08, 18 June 2008 (UTC)

1. It was particularly the first issue (inconsistent journal names) that I was thinking about in dismissing my own idea. However, thinking about it more, that's not the end of the world; whatever citations do use a consistent journal title would be categorized together, which is still helpful.
2. We could use both year and journal name categories, and the template could certainly put everything in a catch-all category up front; no need for a bot to be involved. Thus, the template would put a single article in "Category:Pages with broken DOIs", "Category:Pages with broken DOIs since 2008", and "Category:Pages with broken DOIs from Australian Pharmacist".

However, as a non-bot user, I ask if it is a speed improvement to have these categories, or is it more efficient to have a bot periodically check all the articles with broken DOIs, which is probably a relatively small number anyway? ASHill (talk | contribs) 19:25, 18 June 2008 (UTC)

Sensible suggestions, but likewise as a non-bot user I'm unsure how helpful multiple categorisation would be – need opinions of others at this point :-) Also should these categories be set as hidden on article page ? Finally I've raised a thread at {{citation}} to get their input into this doi_brokendate. David Ruben ^Talk 22:21, 18 June 2008 (UTC)

• I vote we wait until someone requests further categorisation rather than speculate about the best way to do something that there doesn't seem to be a clear benefit for... Smith609 Talk 13:27, 19 June 2008 (UTC)

Progress report: 4 days in and 126 pages currently in Category:Pages with DOIs broken since 2008, mostly I think located by DOI bot. I'm not sure of that bot's workload whether it has in the last few days checked a sizeable fraction of all articles using cite journal (and the DOI broken category is likely increase only slightly more), or if only a small fraction of all articles have been checked, in which case the category list is about to swell hugely :-) David Ruben ^Talk 17:39, 22 June 2008 (UTC)

Sometimes they are broken simply by being misspelled, with the publisher not being at fault at all. E.g. [4]. --Rifleman 82 (talk) 17:53, 22 June 2008 (UTC)

According to the bot, it's worked through 90% of its ~70,000 page workload (all pages transcluding either Cite Journal or Citation) in the last couple of days. However, it's done a couple of rather fast bursts which I suspect may reflect its losing its place. Nonetheless, it should be taken as encouraging that so few broken DOIs have been detected so far. Martin (Smith609 – Talk) 17:59, 22 June 2008 (UTC)

{{

 Done But for future reference, interwikis should go on the template documentation, which anybody can edit. Best, PeterSymonds (talk) 09:00, 15 June 2008 (UTC)

Thanks, I will remember that. guillom 11:29, 23 June 2008 (UTC)

The template adds a period at the end of the title. Now there are journal articles whose title ends in a "?" and then we have "title?". appearing. Also the quotes=no does not seem to have an equalent in Template cite book - leading to inconsistencies. Shyamal (talk) 01:33, 16 June 2008 (UTC)

So do we just remove the final "." from title parameter display ? David Ruben ^Talk 18:36, 16 June 2008 (UTC)

Personally, I would think that no alteration should be made on the title as given in the original and so adding a period should be avoided. In addition, the use of quotes which is found only in the MLA style encloses the period. Ideally "quotes=no" should be the default for journals. See for instance here http://www.liu.edu/cwis/cwp/library/workshop/citation.htm Shyamal (talk) 06:46, 17 June 2008 (UTC)

Based on a quick test at User:Ilmari Karonen/sandbox/Cite journal, it seems that the need for the "doilabel" parameter can be safely eliminated by using the {{#tag:}} parser function. The change is very simple: just replace "{{{doilabel|{{{doi}}}}}}" with "{{#tag:nowiki|{{{doi}}}}}". Can anyone think of any reason not to do this, or should I just go ahead with it? —Ilmari Karonen (talk) 18:46, 18 June 2008 (UTC)

How often is "doilabel" parameter already used in articles. We can always depreciate it from the documentation. But is it not already superfluous with the "urlencode" used in the coding (I'm not sure how that really works) ? David Ruben ^Talk 19:12, 18 June 2008 (UTC)

What a handy new function! I spent ages playing with urlencode to no avail (see above) – this is just what was needed to solve the problem! I've given it a test run and it appears to be working brilliantly – super! The edit should definitely be made; to this page as well as template:doi and template:citation. Smith609 Talk 11:13, 19 June 2008 (UTC)

{{

 Done PeterSymonds (talk) 20:43, 19 June 2008 (UTC)

Just 1123 per this search instances of it used it articles to remove – can we get a bot to do that ? David Ruben ^Talk 20:50, 19 June 2008 (UTC)

DOI bot is already on it (-: Martin (Smith609 – Talk) 21:36, 19 June 2008 (UTC)

Accessdates are not showing for cite journal, but do show for other templates such as {{cite web}}:

|{{cite journal |last=Wiki |firstBob |title=Templates |journal=Wikipedia |url=http://en.wikipedia.org/wiki/Template:Cite_journal |accessdate=2008-06-21}}

Wiki, Bob. "Templates". Wikipedia. Retrieved 2008-06-21.

--—— Gadget850 (Ed) ^talk – 11:09, 21 June 2008 (UTC)

See #Retrieval dates for online versions of old printed sources, again. Martin (Smith609 – Talk) 11:54, 21 June 2008 (UTC)

I'm familiar with that discussion. The only result as yet is to wrap the date in <span class=reference-accessdate> so that editors can hide the date; no global change has been applied to MediaWiki:Common.css. {{cite news}} has had the css class applied so the date can be hidden, but it currently shows the date by default, as does {{cite web}} which has not been updated:

• Wiki, Bob. "Templates". Wikipedia. Retrieved 2008-06-21.

• Wiki, Bob. "Templates". Wikipedia. Retrieved 2008-06-21.

--—— Gadget850 (Ed) ^talk – 13:44, 21 June 2008 (UTC)

The result as I remember it was that access dates are of little importance in unchanging documents, such as journal articles, which are the same whenever they are accessed (but they may help editors repair links which have died). However, web pages change with time, so providing an access date gives a record of the exact page that was viewed. Martin (Smith609 – Talk) 14:16, 21 June 2008 (UTC)

Ah- I see it now. The accessdate was commented out on 5 June. That is not the solution, the idea is to wrap the date in a css class that gives editors the option of seeing the date. --—— Gadget850 (Ed) ^talk - 15:40, 21 June 2008 (UTC)

The central discussion for this is at: Wikipedia talk:Citing sources#Retrieval dates for online versions of old printed sources, again. The only consensus is to enclose the accessdate in a class so that it has the capability of being hidden. I'm going to uncomment the accessdate here and add the class. --—— Gadget850 (Ed) ^talk - 16:44, 23 June 2008 (UTC)

Other cite templates such as cite web and cite news automatically wikilink dates for the date tag in the ISO 8601 format, and allow wikilinked dates in other formats as well, so that user date preferences can be used. To me this seems like a really useful feature, would it be possible to add it to cite journal too? Thanks Rjwilmsi 22:38, 21 June 2008 (UTC)

A citation that uses doi= generates text that goes off-wiki, but it's missing the little box-with-an-arrow that warns the reader that following the link will exit Wikipedia. It'd be better to be consistent with the rest of Wikipedia, and warn readers by using the box-with-an-arrow. Here's an example of a citation and what it currently generates:

• {{cite journal |journal= Nat Rev Genet |date=2008 |volume=9 |issue=5 |pages=341–55 |title= Advances in autism genetics: on the threshold of a new neurobiology |author= Abrahams BS, Geschwind DH |doi=10.1038/nrg2346 |pmid=18414403}}
• Abrahams BS, Geschwind DH (2008). "Advances in autism genetics: on the threshold of a new neurobiology". Nat Rev Genet. 9 (5): 341–55.
  PMID 18414403
  .

• Abrahams BS, Geschwind DH (2008). "Advances in autism genetics: on the threshold of a new neurobiology". Nat Rev Genet 9 (5): 341–55. doi:10.1038/nrg2346. PMID 18414403.

so that there is a little box-with-an-arrow after the DOI. Eubulides (talk) 05:51, 5 July 2008 (UTC)

OK, I brushed up on the HTML and now have a specific proposed patch, namely, change "plainlinks neverexpand" to "neverexpand" in this template, as in this diff. Any objections? Eubulides (talk) 05:13, 8 July 2008 (UTC)

I'm not too fond of proposals to add clutter to citations, unless it's necessary. Would any reader honestly expect the DOI, or PMID for that matter, link to lead to a wikipedia article on "10.1038/nrg2346"? Martin (Smith609 – Talk) 09:30, 8 July 2008 (UTC)

I suspect any reader who doesn't know what a DOI is--ie the vast majority--wouldn't know that. The box is a helpful visual cue. ASHill (talk | contribs) 11:26, 8 July 2008 (UTC)

Good point. Fair enough then - you may as well make the change! Martin (Smith609 – Talk) 12:04, 8 July 2008 (UTC)

{{

Done. Please check. --Rifleman 82 (talk) 05:29, 9 July 2008 (UTC)

A suggestion comparing this to the other cite XXX templates:

• {{cite web}} & {{cite news}} place after the title "(format) (in language)", whilst {{cite journal}} & {{citation}} order these two as "(in language) (format)" - which is better ? David Ruben ^Talk 03:42, 8 July 2008 (UTC)

support given for changing this template to "(format) (in language)" - see

Template talk:Citation#Standardising with other templates. David Ruben ^Talk

02:35, 9 July 2008 (UTC)

and

Template talk:Cite web#Standardising with other templates David Ruben ^Talk

01:29, 13 July 2008 (UTC)

Why is the entire output of this template italicized, unlike the other cite xxx templates where only the title is in italics? It looks bad when several different type inline cites are footnoted to have this one stick out from the rest. Besides, why should the author's name, page number, etc., be in italics along with the journal's title? JGHowes ^talk - 03:07, 19 July 2008 (UTC)

It's not entirely italicized; see the examples in #Formatted DOIs missing the box-with-an-arrow above. Perhaps you put the template within an italicized section or something? Do you have a specific example of the problem? Eubulides (talk) 03:38, 19 July 2008 (UTC)

It is not all italicised, only the journal name! Have you altered your skin or font settings ? This cite_journal follows cite book where the "book" is in italics, and "chapter" subsection is not:

author (2008). "title". journal name. {{cite journal}}: |author= has generic name (help)

author (2008). "chapter". book title. publisher. {{cite book}}: |author= has generic name (help)

David Ruben ^Talk 03:45, 19 July 2008 (UTC)

{{Citation}} now produces the same format as Template:Cite journal. To keep things this way, this template should be modified to use the same core as Template:Citation.

In order to implement this I've set up a "Test template" at User:Smith609/Cite journal. To test it out, use the syntax

{{:User:Smith609/Cite journal | title= whatever |author = etc}}

I'm pretty sure it should work, but would like to give the community the opportunity to make sure before I propose an edit! If no errors are reported this code should be ready to go. Thanks! Martin (Smith609 – Talk) 15:07, 25 July 2008 (UTC)

I put some examples in my sandbox. If I kept everything straight, the method of separating elements has been inverted in the Smith609 versions of the template. For the official templates, Citation separates elements with commas and Cite journal separates them with full stops. For the Smith609 versions, the reverse is true. In any case, I would like to know the rational for whichever separator is used. --Gerry Ashton (talk) 17:02, 25 July 2008 (UTC)

The Harvard citation format uses a full stop. No official format uses commas, and it makes more sense to use an established reference format than to invent one. Martin (Smith609 – Talk) 19:02, 25 July 2008 (UTC)

The Chicago Manual of Style uses commas for footnotes, and uses full stops for the Harvard format reference list. Also, although I've never used the MLA style, apparently it also can use footnotes with commas (see http://www.aresearchguide.com/7footnot.html ). --Gerry Ashton (talk) 20:45, 25 July 2008 (UTC)

While at it perhaps you can fix the problem with article titles ending in question marks being followed by an additional stop after the quote. See Ant#cite_note-55. Shyamal (talk) 17:22, 25 July 2008 (UTC)

This can be easily done if it's really necessary. Martin (Smith609 – Talk) 19:02, 25 July 2008 (UTC)

I like the suggestion too. If it's done, please treat exclamation point and period the same way you treat question mark. '?".' and '!".' and '.".' are all equally awkward. Thanks. Eubulides (talk) 20:17, 25 July 2008 (UTC)

Due to the limitations of the available functions, there would have to be a manually (or

bot) specified "no comma after title" parameter - but I take your point. Martin (Smith609 – Talk

) 20:38, 25 July 2008 (UTC)

Looking through the source, I do not see how it handles coauthors. « D. Trebbien (talk) 23:05, 25 July 2008 (UTC)

Ooh, good spot. Thank you!  Fixed Martin (Smith609 – Talk) 09:05, 26 July 2008 (UTC)

I have made a new tool to generate a cite journal template from a PMID. It has a slightly different formatting than the Diberri tool. The new tool is available from http://hendrix.imm.dtu.dk/cgi-bin/brede_bib_pmid2wpcitejournal.py . — fnielsen (talk) 09:46, 31 July 2008 (UTC)

How about writing the author's name in capital letters, by using {{

I'm trying to re-point attention on this again. ––Bender235 (talk) 18:36, 6 August 2008 (UTC)

What paper style manual or major publication does that? I think there are very few, if any. Why should we depart from customary methods of citation? --Gerry Ashton (talk) 18:41, 6 August 2008 (UTC)

Note: I'm not pro or con, I'm just trying to find out whether there is any consensus on this, because some article use capital letters (e.g.

Olmec), others don't. ––Bender235 (talk

) 19:22, 6 August 2008 (UTC)

Sure, there's consensus.

Wikipedia:Manual of Style (capital letters)#All caps says "Avoid writing in all capitals: ... However, write acronyms and initialisms in all capitals.". RossPatterson (talk

) 02:55, 7 August 2008 (UTC)

Okay, thanks. ––Bender235 (talk) 11:31, 7 August 2008 (UTC)

I don't think that the particular MOS guideline given above is talking about this context (ie bibliography/referencing styles), and so I don't think it applies in this case. Also, out in the Real World quite a few well-known and respectable publishing houses do elect to use small- (or all-) caps for authors' names in bibliographies/references sections, including Oxford University Press, University of Texas Press, Dumbarton Oaks/Trustess for Harvard University, Elsevier, and Akadamie Verlag. It's far from an uncommon presentation.

However, while an optional smallcaps formatting feature could readily and technically be incorporated into {cite journal} or other citation templates, the same functionality can just as easily be achieved by using {{

aut}}. I don't see the need to amend these citation templates to provide this facility. Just so long as the citation/referencing style within an article is consistent, I don't see a problem if {aut} is used in a references/biblio section. --cjllw ʘ TALK

05:18, 11 August 2008 (UTC)

I'd like to weigh in and note smallcaps looks like crap on a macintosh using Firefox 2. Smallcaps are terrible. Its also not a common style in academic publishing despite individual books having used it. Rsheptak (talk) 23:27, 17 August 2008 (UTC)

Update. In Firefox 3, Macintosh, small cap rendering is fixed and they look fine. Rsheptak (talk) 18:05, 1 September 2008 (UTC)

{{

Questions: sometimes a book is published that consists of articles collected from various journals; all the articles relate to a certain topic. In such a case, would it be useful to use work to refer to the book and journal to refer to the journal in which it originally appeared? Also, if any change is made to the template, it should be documented what will happen if non-equal values are specified for work and journal. --Gerry Ashton (talk) 16:11, 16 August 2008 (UTC)

 Done: the behaviour is that |work= is evaluated only if |journal= is undefined (not simply empty). Please ensure that the documentation is updated. Happy‑melon 15:53, 17 August 2008 (UTC)

Can the templates be fixed so that they allow the form "Retrieved on January 1, 2008" as an alternative to "Retrieved on 2008-01-01"? If I indicate the article date as June 1, 2007, for consistency, I prefer to use the same date style for the accessdate: "Retrieved on January 1, 2008" as opposed to "Retrieved on 2008-01-01". Also, I would prefer it if wikilinking the accessdates was optional (requiring the use of brackets). Thanks. --Phenylalanine (talk) 06:07, 26 August 2008 (UTC)

Can someone fix the problem here http://en.wikipedia.org/wiki/Ant#cite_note-0 ? You may need to click the clade expand (show) to expand the location where the cite occurs. The ref tag appears to have the url added as expected. Shyamal (talk) 11:34, 29 August 2008 (UTC)

Fixed. The problem was there was a new-line partway through the title (after "ants"). While individual parameters can be on individual lines, items with links need to be on one line. +mt 14:33, 29 August 2008 (UTC)

Thanks a lot. Shyamal (talk) 03:35, 30 August 2008 (UTC)

Co-authors' names aren't displayed if no author (via author= or last= |first=) is given. Can someone please fix that bug?

Code: {{cite journal |last= |first= |authorlink= |coauthors=Howitt, P. |year=1989 |month= |title=Gradual Reforms of Capital Income Taxation |journal=American Economic Review |volume=79 |issue=1 |pages=106–124 |doi= |url= |accessdate= |quote= }}

Result: "Gradual Reforms of Capital Income Taxation" (1989). American Economic Review 79 (1): 106–124.

But should be: "Gradual Reforms of Capital Income Taxation" with Howitt, P. (1989). American Economic Review 79 (1): 106–124.

––Bender235 (talk) 09:44, 3 September 2008 (UTC)

In what cases would one want to list the "coauthors," while omitting the primary author? --Karnesky (talk) 13:16, 3 September 2008 (UTC)

In just about every list of publications by a single author (bibliography), like Hans-Werner_Sinn#Selected publications. ––Bender235 (talk) 16:41, 3 September 2008 (UTC)

That seems to be a poor style. For a small amount of brevity, we give up a great deal of clarity. It can be inferred that the subject is an author on all papers, but the order of the author list can not be determined. Sinn is the second author on the Am. Econ. R. article. The "issue" of only listing coauthors when the primary author is given is mirrored with the cite book template. See 'Jumpstart' on that page. Perhaps "poor" use of templates is better than not citing (in which case we should implement the change), but I certainly think that pages like that should be fixed. --Karnesky (talk) 19:14, 3 September 2008 (UTC)

I agree that that is a poor style, but if there really is a good reason to use a highly unusual format like that, why not use the author= parameter?

Code:

{{cite journal |author= with Howitt, P. |year=1989 |month= |title=Gradual Reforms of Capital Income Taxation |journal=American Economic Review |volume=79 |issue=1 |pages=106–124 |doi= |url= |accessdate= |quote= }}

which produces:

with Howitt, P. (1989). "Gradual Reforms of Capital Income Taxation". American Economic Review. 79 (1): 106–124. {{cite journal}}: Cite has empty unknown parameter: |month= (help)

This need seems to me to be unusual and specialized enough that the template probably shouldn't be designed with it in mind; listing coauthors without a first author is more likely to be a mistake than a deliberate choice. —Alex (ASHill | talk | contribs) 19:29, 3 September 2008 (UTC)

No, that won't do it, because it makes one particular citation inconsistent with the rest of the bibliography. Where's the problem with adding another function to this template? ––Bender235 (talk) 21:34, 3 September 2008 (UTC)

(unindenting) Please also note that this template embeds

And how do we solve this problem? Does that mean that on Hans-Werner Sinn's bibliography (for example), his name needs to appear on every single entry? That solution doesn't sound convincing to me. ––Bender235 (talk) 16:34, 4 September 2008 (UTC)

Cite templates are only suitable for reasonably commonplace situations. If you have an unusual situation, write the citation by hand. --Gerry Ashton (talk) 16:54, 4 September 2008 (UTC)

No, that's not a good option, because we use citation templates not only for consistent style, but also for future microformats implementation. Therefore, writing the citation by hand is not a good choice at all. ––Bender235 (talk) 20:49, 6 September 2008 (UTC)

Bender235, please tell me who will write a cite template improvement within 60 seconds whenever I need to cite something not covered by the existing templates. Right after you tell me that, I will begin to consider your concerns abotu microformats. (Why 60 seconds; because that's about the maximum I'm willing to wait when I'm in the middle of fixing a problem in an article.) --Gerry Ashton (talk) 21:17, 6 September 2008 (UTC)

I don't think it's a big deal anyway. It's a simple IF ... THEN ... function, and it's not like it's the first one in this template:

If you enter first= and last=, {{cite journal}} will produce:

Sinn, Hans-Werner (1989). "Gradual Reforms of Capital Income Taxation". American Economic Review 79 (1): 106–124.

If you don't enter this parameters, it will produce:

"Gradual Reforms of Capital Income Taxation" (1989). American Economic Review 79 (1): 106–124.

Obviously there already is an IF ... THEN ... function that decides whether year should appear ahead or behind title. It might take 60 seconds to implement another IF ... THEN ... function that decides: IF no author THEN coauthor should appear behind title. Don't tell me that's too difficult. ––Bender235 (talk) 08:54, 7 September 2008 (UTC)

Bender235, if your latest post is directed at me, then you misunderstood my point. Editors are not programmers. Even if they even choose to use cite templates at all, they are always free to instantly abandon them and hand-code a reference if the available parameters do not adequately represent a source they are citing. If you think they are going to stop in the middle of an article and learn to become programmers just to support microcoding (which isn't even mentioned in the MOS, so how should they even know about it) then you need to completely revise your mental image of an editor. --Gerry Ashton (talk) 20:01, 7 September 2008 (UTC)

• Bender, I don't think that Gerry is familiar with the Wikipedia parser function syntax. Perhaps you would be willing to make modifications to this template to create a working sandbox to accommodate the change that Gerry has requested. If you really can do that in under 60 seconds, you're a much better coder than me! You may also wish to consider how the microformat will deal with a publication that has no author specified.
• Gerry, as an alternative for now, you might consider using the style adopted by many journals of using an em dash (&mdash; = —) in place of the author field. It might be simpler just to repeat the author's name; this will allow you to be consistent with multiple author publications of which the subject is not the first author.
• Perhaps the best solution is to create a "hide_first_author" parameter, which either does not display the first author's name, or replaces it with a dash. This will allow the microformat to recognise the omitted author without clogging up the page.
• Martin (Smith609 – Talk) 20:54, 7 September 2008 (UTC)

I'm not asking for help. I am objecting to the fact that Bender advised an editor that it was a bad practice to hand-code citations, even in situations where the tempate was not up to the job. --Gerry Ashton (talk) 21:21, 7 September 2008 (UTC)

Fact is that this template does not only provide a consistent citation style, it also implements COinS and might implement a citations microformat (hCite or whatever the name might be) in near future. Which means in this case, the use of templates is strongly recommend. ––Bender235 (talk) 23:38, 7 September 2008 (UTC)

I'd like to see this "hide-first-author" function implemented. ––Bender235 (talk) 23:38, 7 September 2008 (UTC)

Is somebody working on this as of now? ––Bender235 (talk) 18:56, 15 September 2008 (UTC)

Whoever does implement this should wait until the edit requested below is completed to avoid having to make it twice! Martin (Smith609 – Talk) 19:22, 15 September 2008 (UTC)

{{

Quite a large change. Is the code well tested? Can you move it to the sandbox? And why is it using markup like {{Template:Citation.. when "Template:" is clearly not needed? Disabling until these questions are answered. Feel free to re-enable if / when appropriate. --MZMcBride (talk) 03:30, 1 October 2008 (UTC)

The code was thoroughly tested when I requested the edit mentioned above, and no complaints were left unresolved. It's now in the Sandbox, and I're re-enabled the request. Thanks, Martin (Smith609 – Talk) 15:40, 1 October 2008 (UTC)

Done. Cheers. --MZMcBride (talk) 19:37, 1 October 2008 (UTC)

Great, thank you. Martin (Smith609 – Talk) 19:48, 1 October 2008 (UTC)

Smith609, I hope you plan to edit the documentation to reflect the changes. --Gerry Ashton (talk) 19:50, 1 October 2008 (UTC)

Loss of functionality

PMC display

Citation core not quite as subtle in its handling of PubMed central as cite journal previously was. 'pmc' correctly used to link title where no 'url' value, however if 'url' is provided then the PMC value should be shown after the PMID:

Hence: {{cite journal |title=Title |author=Blogs |pmc=12345 |pmid=1234 |url=http://www.example.org}}

Gives: Blogs. "Title".

PMID 1234. {{cite journal}}: Cite journal requires |journal= (help) - links to url, but PMC not shown. David Ruben ^Talk

21:50, 1 October 2008 (UTC)

doi_brokendate parameter

{{#if:{{{doi|}}}
  |. [[Digital object identifier|doi]]:{{#if: {{{doi_brokendate|}}}
      | {{#tag:nowiki|{{{doi}}}}} (inactive [[{{{doi_brokendate|}}}]]) {{#ifeq: {{NAMESPACE}} | {{ns:0}} | [[Category:Pages with DOIs broken since {{#time: Y | {{{doi_brokendate|}}} }}]] }}
      | <span class="neverexpand">[http://dx.doi.org/{{urlencode:{{{doi}}}}} {{#tag:nowiki|{{{doi}}}}}]</span>
}}

{{ 
<!--============ DOI ============-->
  #if: {{{DOI|}}}
  |{{
     #if: {{{URL|}}}{{{IncludedWorkURL|}}}
     |, {{doi| {{{DOI}}} }}
     |<span class="printonly">, doi {{{DOI}}}</span>
   }}
}}

David Ruben ^Talk 22:23, 1 October 2008 (UTC)

Hi, I'm currently developing an EndNote plugin for Wikipedia, and something that's causing me problems is that the software requires the first author to be separate. This is also helpful for many library and university plugins that use the citation data to provide users with links to electronic versions of articles at their libraries. The use of the "author" parameter is, from this perspective, very undesirable. The use of the "first" and "last" parameters makes the COinS metadata much more useful. Another benefit of this is that it produces a consistent formatting between citations, and it allows the use of harvard style linking to citationswith the {{

I use Zotero and LibX, so appreciate detailed COinS entries. That being said: any referencing is better than no referencing & we should be careful to encourage citation in general. "Consistent formatting" means different things for different articles. Order of given/family names (or initials) do differ & so any template style will be subverted to display things a different way on some pages to match a pre-existing citation format. The more "free-form" author parameter may be useful for these cases. For the purposes of plugins that try to process the data, the coauthors field may prove just as problematic.

If you have issues dealing with the "author" parameter in your plugin, I don't know why you think bot transformation of templates is a good idea. Any bot should be manually run & I'd expect that the initial rate should be slow enough to double-check that the bot operator was not screwing things up; a lot of pages use this template and they use it in different ways to achieve the best human-readable results for a particular article.

I'd have no problems encouraging the use of first/last more (rewriting examples, etc.), but don't think that 'author' should be undocumented or deprecated at this time. --Karnesky (talk) 05:19, 27 September 2008 (UTC)

For biomedical topics, Dibberi's tool for generating markup from

WP:MED) will continue to insert references with the markup it generates (i.e. author=). Have as an equally valid alternative fine, but don't "discourage" 'cos that'll just get ignored. Of course if Diberri is persuaded to alter his coding to use author list (PubMed XML output shows that would would be possible) fine. But use of multiple first/last parameters seems would be harder on occasions I need to hand code the template. David Ruben ^Talk

22:03, 1 October 2008 (UTC)

I strongly disagree with the original suggestion. Coding additional mark-up parameters is a burden on editors. If you want to replace 1 param with an indefinite number, you need to produce a form-driven tool to make the job easier, so that editors can just paste in the values. Such a form would need a "scratchpad" area so that editors can first paste the entire citation there and then continue without flipping between web pages. Such a form would be about 1 page of Javascript. I know Martin has provided a link to a citation-formatter for Google Scholar, but that only works for Google Scholar and therefore only for "academic" topics. The change to {{cite journal}} should also not be retrospective, as that would invalidate citations in thousands of articles. -- Philcha (talk) 07:31, 2 October 2008 (UTC)

PS I think I could write such a JS if the documentation for the templates is accurate and up-to-date. I'd also be inclined to make the JS for "cite journal" a wrapper for a JS that does the same for "citation", so the "cite journal" JS simply validates and then, if OK, calls the "citation" JS. That way we can easily extend it to other citation formats. -- Philcha (talk) 07:47, 2 October 2008 (UTC)

I don't believe that tools should impact this decision, but you could use zotero or another reference manager to generate your citation templates for you. --Karnesky (talk) 07:57, 2 October 2008 (UTC)

A decision that makes life more difficult for editors will discourage new editors, and Wikipedia will stagnate. I think that's more serious than not making optimal use of the COinS metadata, whatever that is. BTW what is it and how does it help Wikipedia readers?

If Wikipedia can get its act together, providing the tools should not be difficult.

I didn't realise there was a Firefox extension to generate citation templates. But I'd prefer not rely on FF extensions: a lot of editors don't use FF (e.g. K-meleon provides all the benefits of the Gecko engine); new FF releases often invalidate extensions, at least for a while and often forever - I used t have a few, but concluded it's not worth the hassle.

PS if, as Martin suggests, a bot could automate the first-last format, why are we discussing the "author" param rather than the implementation of the bot? BTW the same bot should also handle the "editor" param - Martin and I have cited a quite few collections in paleo articles. -- Philcha (talk) 08:33, 2 October 2008 (UTC)

I have also constructed a PubMed to Wikipedia cite journal template markup, see, e.g., [5]. In the tool I have made wikilinks for the authors and use firstname-lastname as the ordering. I suppose that it may not be appropriate to add formatting (i.e., wikilinks) in the template field values, and that some regard the many redlinks the tool generates as bad. I can change the script to use last, first and coauthors fields, but I like the firstname-lastname: In Wikipedia people are not generally referred to as, e.g.,

Andersen, Hans Christian. Also the links to authors provide some context for the reference I think. — fnielsen (talk

) 12:55, 2 October 2008 (UTC)

Does your tool check to see whether the author has a page under a differently formatted name (e.g. JRR Tolkien exists, J.R.R. Tolkien doesn't), or whether the page it links to is relevant, and not about an unrelated person of the same name or a disambiguation page? Martin (Smith609 – Talk) 20:33, 2 October 2008 (UTC)

Re reference construction, my Universal reference formatter will generate a reference from pretty much any information you care to throw at it. Re placing burden on editors; there is no need for editors to change their habits, and we should never pass a rule saying "Editors MUST do this or their referencing will be removed". Having a consensus on the "best way" to format a reference allows a bot to go round and format references that way, creating a better reader experience with no added burden on editors. But I don't want to set a bot loose making things appear "my way" if there are alternatives with different merits. Martin (Smith609 – Talk) 18:51, 2 October 2008 (UTC)

Hi, Martin, we gotta stop meeting like this.

Your Universal reference formatter isn't universal, as it only works for journals, not for e.g. books or web pages. fnielsen's formatter is even more specialised, as it accepts only one input, PMID. Both are also on separate sites, are not advertised in the Edit page layout and need to open in a separate tab / window. I'd find the last point a nuisance as I usually have 6-9 tabs open while editing. All of these are reasons why I suggested JS, but a prerequisite would be a commitment from "the management" that a functionally satisfactory JS would be loaded automatically and represented by a button in the Edit toolbar.

Now if the JS also remembered all ref names assigned in the article and offered a menu of already-defined refs, that would really be getting somewhere. -- Philcha (talk) 19:14, 2 October 2008 (UTC)

Try popping an ISBN into the Univ Ref Formatter. It produces a full citation for a book. It's not as reliable the other way round but there've not been any requests for that feature; if there were any, I could implement it quite easily. You can pop a website's title, URL and author in to create a suitably parsed references, if web pages count as references these days. I agree that integrating them with the WP layout would be advantageous, and would be happy to put together some JS to do so - but I've no idea how this could get integrated with the user interface. Martin (Smith609 – Talk) 20:30, 2 October 2008 (UTC)

Re "how this could get integrated with the user interface", I've posted a query at Wikipedia:Village_pump_(technical)#Reference_formatter. -- Philcha (talk) 05:48, 6 October 2008 (UTC)

Answer to Martin: No, my tool (http://hendrix.imm.dtu.dk/cgi-bin/brede_bib_pmid2wpcitejournal.py) does not check to see if a page exists or if there is a collision. — fnielsen (talk) 07:12, 3 October 2008 (UTC)

Opposed to this idea to the extent it will affect bio/med articles that use Diberri's PMID filling template. SandyGeorgia (Talk) 04:30, 6 October 2008 (UTC)

Hi, SandyGeorgia. As a computer consultant I've seen many situations where an unsatisfactory system is propped up by a series of mostly-incompatible user-created bolt-ons. The only answer is to create a single system that does the job properly - otherwise more bolt-ons are added, the disincentive to change increases and the result is stagnation at the centre and chaos round the fringes. Solving the problems properly will require significant work, first to add a formatter that's easier to use than the alternatives and then to convert the old data, but it's the only way out of the mess. -- Philcha (talk) 05:55, 6 October 2008 (UTC)

A possible way out of the mess

Forcing the biomedical articles to first/last inputs will be painful if it has to be done manually, but if automatic there might be less objection. As a first step, is there any reason that cite can't be made to turn arguments of the form: {{cite journal |last1=Viollet |first1=B |last2=Andreelli |first2=F |last3=Jørgensen |first3=SB |last4= ''et al'' |title=The AMP-activated protein kinase alpha2 catalytic subunit controls whole-body insulin sensitivity |journal=J. Clin. Invest. |volume=111 |issue=1 |pages=91–8 |year=2003 |month=January |pmid=12511592 |pmc=151837 |doi=10.1172/JCI16567 |url=}} into an NLM/URM-style display that looks like the Viollet B, Andreelli F, Jørgensen SB; et al. (2003). "The AMP-activated protein kinase alpha2 catalytic subunit controls whole-body insulin sensitivity". J. Clin. Invest. 111 (1): 91–8.

I'd be in favor of this & I think that this template used to behave. (And it is still how the template is documented right now!) --Karnesky (talk) 23:06, 6 October 2008 (UTC)

I'm flattered that the perhaps thick-headed way the template filler does its job has any influence on the discussion here. I wholeheartedly agree with Philcha's concerns above. If there's a problem with the way we're generating citation templates, then let's fix it. I'm happy to update my template filler if it will contribute to the solution. --David Iberri (talk) 02:55, 7 October 2008 (UTC) PS: Thanks LeadSongDog for the heads-up about this conversation.

Any change to David's template filler would be a later step anyhow, although it would be helpful to know if he forsees any major complications in implementing it. First off, we need to decide whether we really want template:cite journal to render URM/NLM style for the reader. For now,

WP:MEDMOS#Citing medical sources is only suggestive on that, not even making a direct recommendation. It seems to me to boil down to a choice of APA or URM style in the rendered text. I would hope that that choice could be made independent of the choice of parameters passed to the template, but at the moment it isn't. Perhaps the presence of a pmid= parameter could be used as a surrogate for style=URM (although I'm loathe to encourage side effects and modal behaviours).LeadSongDog (talk

) 15:47, 7 October 2008 (UTC)

Please remember that any change you make will not be confined to medical sources. Please explain your suggestions in terms that non-medical people can understand. For example, wtf is URM or NLM? --Gerry Ashton (talk) 19:11, 7 October 2008 (UTC)

Sorry, Gerry, I forgot where I was posting. At

Pubmed facilities used by Diberri's and other tools here.LeadSongDog (talk

) 19:43, 7 October 2008 (UTC)

May I just point out that there are swathes of people in the world who do not have surnames, only

patronyms, in which case the "author" parameter remains relevant. The parameter is also useful for corporate authors. — Cheers, JackLee ^–talk

– 04:00, 27 November 2008 (UTC)

Am I going nuts? I use the various citeX templates regularly, and it seems that I'm getting different outputs at different pages (I have purged cache multiple times). The reference at Tastavinsaurus, which I just added, has an ampersand between the first and second authors, commas instead of periods between elements, no space between volume and issue, and no period at the end of the reference. Meanwhile, a months old ref using the same template at Calsoyasuchus has no ampersand, periods instead of commas, a space between volume and issue, and a period at the end. Is this a template change that has yet to universally take? (Personally, the only change I like is the removed space; I think it looked more professional without the changes, and the ampersand looks silly when you end up with a citation like "Howard & Fine; and Howard"). J. Spencer (talk) 22:44, 1 October 2008 (UTC)

This template (but perhaps not all the cite xxx templates) has just been changed to produce the same output as the Citation template. I know that template does separate elements with commas rather than full stops. This has been discussed but never resolved. --Gerry Ashton (talk) 22:56, 1 October 2008 (UTC)

However it comes out, it needs to be tweaked to better represent references having more than two authors. J. Spencer (talk) 04:02, 2 October 2008 (UTC)

As the template is used by many pages, it takes a while for edits to it to propogate. Martin (Smith609 – Talk) 18:53, 2 October 2008 (UTC)

Proposal to change it back the way it was before

The recent change to make {{cite journal}}'s output mirror (actually use the code of) {{citation}} was ill-conceived, and I think we need to set it back the way it was before. This change is messing up who knows how many pre-existing uses of cite journal—probably thousands, I wouldn't be surprised—as others note above, and there are a few other reasons why the change is sub-optimal. Firstly, if I'd wanted to achieve a citation output looking like {{citation}}'s, I'd be using that template and not {cite XXX} templates. There are good reasons why some folks decide to use cite XXX and not citation, and vice versa; the latter's output does not suit all comers and disciplines. Secondly, the template's output now diverges considerably from the other main templates in the "cite XXX family", like {{

• Campbell, Lyle & and Una Canger (1978), "Chicomuceltec's last throes", International Journal of American Linguistics 44: pp.228–230, doi:10.1086/465548, ISSN 0020-7071

or with 3 or more authors:

• Whorf, Benjamin Lee & Frances Karttunen and Lyle Campbell (1993), "Pitch Tone and the "Saltillo" in Modern and Ancient Nahuatl", International Journal of American Linguistics (Chicago: University of Chicago Press) 59(2): pp.165–223, doi:10.1086/466194, ISSN 0020-7071, OCLC 1753556

Even if the new {cite journal} is filled in knowing the linking ampersand is gonna be supplied by the template whether you like it or not, and its (now undocumented) fields are populated as intended, to my mind the output can still be confusing and ambiguous. For eg, are there 2 or 3 authors to this paper:

• Joyce, Rosemary & Kent Flannery (1996), "Zapotec Civilization: How Urban Society Evolved in Mexico's Oaxaca Valley", Mesoamerica 23(3): pp.100–122

(I realise this is using coauthors field instead of last2, first2, etc. But the issue with using last1, last2, last3 etc is that the output imposes a "lastname, firstname" order for all of the contributors, whereas in much of the humanities referencing style the preferred ordering for multiple authors goes "Lastname1, Firstname 1; Firstname2 Lastname2, Firstname3 Lastname3, and Firstname4 Lastname4". Don't see why we must all be using Last, First). I must have coded multiple hundreds of {cite journal} calls, many with more than 1 author; I doubt I'm alone in this and I really don't feel inclined to (a) find 'em all, and (b)re-edit them to take out the and's and do other changes so they don't look so silly. It's much the same situation as with the many many instances where "p." or "pp." has been manually prepended to page numbers in templates that did not supply them— if the template is changed to auto-insert these p's and pp's there'd be too many cases of doubling-up to fix. Fourthly, although someone above asked for it to be done, it seems the template documentation here and elsewhere has not been updated to reflect the current changes. Perhaps just as well, since I propose undoing the recent change to this template. Anyone in disagreement? --cjllw ʘ TALK 01:05, 3 October 2008 (UTC)

The issues you raise are important. Taking them one at a time:

• • Different templates. Having two different templates to fulfil the same purpose was fraught with errors. Because each did things in subtly different ways, editors had to learn twice the number of parameters if they wanted to interact with both templates. More majorly, when an update was required - and they are more frequent than you might think - it was almost always the case that one template was updated but not the other. This problem is now fixed - an update need only be written once. This brings me nicely to:
  • Consistency. A consensus was reached that the Cite Journal and Citation templates should produce the same output, so that they could be used in the same article as one another. I agree that it makes sense to make it the same as "Cite book" and friends; no-one threw up any major differences as the consultation stage, but if you can outline the differences here, these can be readily fixed in the code. That way it will also be possible to add a "citation" template to an article without checking the entire source to see if there's already a "cite book" tucked away somewhere.
  • Unusual "and" occurrence. There is no reason for the word "and" to appear in an author field; indeed this should be discouraged as it will break browser plugins that use COinS metadata. It would be easy for the Citation bot to go around and remove all unwanted occurrences of "and" if you can give me a list of cases where it might appear. Can you give examples of pages where this has been a problem?
  • Documentation. I'll make a start.

Although there may be a couple of niggles to smooth out, having a single format for all citations, and one place for the code to be updated, is a vastly superior situation to two independent templates. There's been a lot of discussion establishing this point in the past which reached as good as a unanimous decision to move to the point we're at now; please let's work forwards from here instead of going backwards. Martin (Smith609 – Talk) 01:31, 3 October 2008 (UTC)

I completely agree that having consistency is a progressive move. The problems noted may need to be fixed, but rolling back to the old template is a bad idea. The old template was not even producing formatting that was accepted as covering

WP:MOS on the FACs. The visibility of problems is now high and that is a good incentive for work on this neglected template. Shyamal (talk

) 01:56, 3 October 2008 (UTC)

I thought that the only style issues with the original version of this template is that it couldn't be mixed-and-matched with the citation template? I think the current template has WORSE style problems & would encourage a temporary roll-back unless I'm missing some major deficiency in the old template. I don't think the Cite XXX templates should be converted one-by-one. I think that the Citation backend should first be improved so that any of the cite XXX templates will look fine & then ALL cite XXX templates should be converted at around the same time. --Karnesky (talk) 02:51, 3 October 2008 (UTC)

It would be helpful if you could list the "WORSE style issues" so they can be addressed. Nobody seemed to notice them when the changes were proposed. Martin (Smith609 – Talk) 03:04, 3 October 2008 (UTC)

The biggest problem is that Cite XXX templates have been freely mixed in articles, while Cite XXX and Citation have been rarely used together. To only adjust one Cite XXX template makes a lot of pages have mixed reference styles, which is exactly what we wanted to avoid. --Karnesky (talk) 03:39, 3 October 2008 (UTC)

While I would agree that there was consensus to make the Cite * templates look more like one another, I don't recall there being any comments that the way to do this would be to change ONE cite template at a time & to put up with breakage. What is wrong with reverting the change for now? We can make the change when we're closer to being ready (documentation, support for all cite xxx templates, fixing all the bugs listed here, etc.)? The main namespace should not be used for testing. --Karnesky (talk) 03:49, 3 October 2008 (UTC)

The design was available for testing for well over a month. All problems and comments raised by people who took the time to test it were promptly addressed. Let's get a list of problems now; it should take a matter of minutes to fix them once they've been identified. Going round fussing about how terrible the whole thing is is not very productive, and I feel is a waste of everyone's time unless specific, fixable points are brought to the table. So far there are no bugs with the template; documentation is in place; and a bot cannot very well go round removing "and" from templates unless editors can see why the change was made. So there doesn't seem to be a strong case for reverting. Martin (Smith609 – Talk) 04:27, 3 October 2008 (UTC)

Oppose Making {{tl:cite journal}} based on {{citation}} was right - they should never have been allowed to diverge in the first place. {{cite journal}} and other variants should be just convenience wrappers,,plus checking that required params for these variants are present and correct. Then {{citation}} should do the actual formatting - it's the only long-term way to ensure that formatting is consistent - especially e.g. if we decide that DOIs should make the title a link unless there's also a URL.

Once the module structure of the templates is sorted out, there are enough citation bots around that can be adapted to clean up any inconsistencies.

Martin, next time you need help with testing, give me a call directly. I don't normally watch this type of page, but I'm a retired computer consultant and I know a bit about testing. It's only fair to warn you that the cost might be a lot of queries and requests to improve documentation :-) Philcha (talk) 07:49, 3 October 2008 (UTC)

I first noticed the problems with this template when I was adding a reference and couldn't get the month to show up. Now I see there are multiple other technical problems as well as inconsistencies, such as journal references not ending with a full stop when other references do. This was obviously not ready to go live. I think the best solution would be to revert to the stable version, and not implement this new version until the bugs have been resolved in the sandbox and there is consensus that it is ready for implementation. Pagrashtak 13:53, 3 October 2008 (UTC)

Disagree. We already reached consensus that it was ready to go live. Clearly not enough editors had been involved in its testing. But proposing the changes on a talk page which is not really read until something goes wrong doesn't seem to get us anywhere; it seems better to get the bugs fixed once and for all, and all at once, than to fix one set of bugs, go live again, have a different set of editors find another set of bugs, go back to the old version, go live again, and so forth. The bugs that have been listed below are minor and easily fixed, and I'll sort them out as soon as I get the time if no-one else is capable. Martin (Smith609 – Talk) 14:07, 3 October 2008 (UTC)

Could you point me to that discussion? I think I'm looking in the wrong place. Sorry, but I don't consider the complete removal of an in-use parameter a "minor" bug. Pagrashtak 14:51, 3 October 2008 (UTC)

I support reversion for a different reason than stated by other editors. I believe the documentation and the code should be considered a unit. When code that changes function is put into production, the new documentation that describes the new function should be already written; the new code and new documentation should go into production within a few seconds of each other. --Gerry Ashton (talk) 13:59, 3 October 2008 (UTC)

I thought I'd addressed the documentation issue. Once again, I'll repeat my plea to make any criticism specific so I can actually do something about it. What is undocumented? What is documented incorrectly? Thanks, Martin (Smith609 – Talk) 14:09, 3 October 2008 (UTC)

Examples of some problems

1) backwards compatability issues. Until about two days ago, {cite journal} had only the |coauthors= field to cater for instances when there was more than one author. I doubt that I've been alone in manually inserting an "and" into the coauthors field (not the author field) at the appropriate juncture, to make it scan better since the template didn't do it, thus:

{{cite journal |author=Brown, Cecil H. |coauthors=and Søren Wichmann |year=...

which used to display as:

Brown, Cecil H.; and Søren Wichmann (1986). "Name of paper". Some Journal. etc...

but now shows as:

Brown, Cecil H. & and Søren Wichmann (1986), "Name of paper", Some Journal, etc...

I have no idea how many instances of this there are that now need fixing, or how they might be located. Other than to remark, prob. enough to be a painful exercise, the template's been used this way for a few years now. It's not just me, someone else gave another example above. I also don't think it would be as simple as to have some bot trawl through to take out the extra ands. For more than one coauthor, the field could be populated as

{{cite journal |author=Campbell, Lyle |coauthors=Terrence Kaufman, J. Richard Andrews, and Thomas C. Smith-Stark |year=...

but now displays as

Campbell, Lyle & Terrence Kaufman, J. Richard Andrews, and Thomas C. Smith-Stark (1986), "Name of paper", Some Journal, etc...

In fact, unless you sunset the coauthors field altogether, it seems that this new cite journal version will not properly cope with there being more than one coauthor in that field (and remember the coauthor field was the only one {cite journal} had to contain them). Even if we didn't have the manually added "and" in there, the template now produces:

Campbell, Lyle & Terrence Kaufman, J. Richard Andrews, Thomas C. Smith-Stark (1986), "Name of paper", Some Journal

with the ampersand misplaced btw the 1st and 2nd authors. Either the & added in by the template needs to be suppressed if the coauthors field is present, or it should be removed altogether. 2) inconsistency with other cite XXX templates. If the objective was to impose greater uniformity in formatting among the various citation templates, then by changing {cite journal} but not {cite book}, {cite web}, {cite conference}, etc as well, it's having the opposite effect. With {cite journal} now a {citation}-clone, if you compare it now with say {cite book} there's a noticeable divergence:

cite journal:

• Campbell, Lyle; Canger, Una (1978). "Chicomuceltec's last throes" (
  ISSN 0020-7071. Retrieved 2008-09-30. {{cite journal}}: |pages= has extra text (help
  )

cite book:

• Campbell, Lyle (1978). "Chicomuceltec's last throes". International Journal of American Linguistics. American linguistics series. Vol. 44. Chicago, IL: University of Chicago Press. pp. pp.228–230.
  ISBN 1234567890. {{cite book}}: |access-date= requires |url= (help); |format= requires |url= (help); |pages= has extra text (help); Check |isbn= value: checksum (help); External link in |chapterurl= (help); Unknown parameter |chapterurl= ignored (|chapter-url= suggested) (help); Unknown parameter |coauthors= ignored (|author= suggested) (help
  )

cite journal (new version) using appropriate parameters (Martin (Smith609 – Talk) 19:15, 12 October 2008 (UTC)):

As noted, differences in element separators (period vs comma); location:publisher is bracketed in the new cite journal but not the others, some corresponding elements like format are placed differently. Before the changes to cite journal, the two outputs above would've looked pretty much identical, IIRC.

I've

put forwards a fix for the misplaced format parameter; it appears where it does in "cite book" because it is designed to give the format for the url linking the book's title, not the chapter title. The other difference concerns the positioning of the publisher's name. I don't think there's a case for displaying the publisher's name at all, in fact. I've never seen a bibliography include the publisher of a journal, and besides, journals often switch publishers. I can't see how it would ever help to locate a journal. In books, the publisher affects the edition being cited, hence reference to pages; this isn't the case with journals, which are not reprinted. Can anyone present a case for citing the publisher or location of a journal article? Martin (Smith609 – Talk

) 19:07, 12 October 2008 (UTC)

I think it's desirable/useful/does no harm to have location, publisher fields for cite journal. Some reasons:

• It's been requested a few times previously, such as here, here and here.
• Comparison with how printed journals/books do things does not really relate. Printed works generally need to minimise the info/space taken up by references to save on the # of pages, this is not a concern on wiki. Nor does wikipedia necessarily follow any given "real world" bibliographical formatting style.
• Bibliographies/references given in journals or specialist books usually presume the reader is already familiar with the cited journal & don't need to spell out its particulars. On wikipedia we can't presume this knowledge of a general readership. Wikipedia has a different purpose, scope and readership.
• To distinguish between periodicals with the same or similar name. Some examples: compare Geowissenschaften, Weinheim:Verlag Chemie [a Deutsche Forschungsgemeinschaft publication], with Geowissenschaften, Berlin:Ernst & Sohn [a publication of the Alfred-Wegener-Stiftung]. Or Géologie [a journal issued by the Université de Lyon], and Geologie [formerly published in East Germany by Akademie Verlag]. Or compare Mesoamérica, South Woodstock, VT:Plumsock Mesoamerican Studies and CIRMA, with Mesoamérica, San Pedro, Costa Rica:Institute for Central American Studies (the first is a journal on precolumbian research, the second more of a modern politics periodical).
• To provide a consistent level of information about the cited work, when (as commonly done) {cite journal}s are intermingled with {cite book}s in an article's references or bibliography section.
• Yes, often a journal can go through a number of publishers in its lifetime, and associate or disassociate with some sponsoring body. This can sometimes make a difference in where it can be tracked down, how it is indexed, or even the content/scope. For eg in the post-WWII period a number of German scientific institutions & societies were split or replicated between East and West. For eg the BRD and DDR each had a similarly named national geosciences society, each issuing similarly named geosciences zeitschriften. Following reunification the societies eventually remerged but for a time the journals coexisted, until the publisher's new owner hived off the former East German title and it's now issued by another entity altogether (but with the same name). Having the fields allows somewhere to distinguish between a title's various incarnations and sponsors.
• Publisher, or at least location, info can help to refine a catalogue search, if other unique identifiers are not known or supplied.
• The template itself may be used for other periodical-type publications (reports, proceedings, etc), not just journals. Location:publisher can be useful.
• The practice of providing location, publisher for journals as well as books may be infrequent but not unknown. I've seen it done in various works, eg here.
• The fields are optional anyway, so anyone who doesn't care to use them can safely ignore, no harm done.--cjllw ʘ TALK 04:45, 15 October 2008 (UTC)

Well argued. So, where in the reference should the Location:Publisher be rendered? Martin (Smith609 – Talk) 13:37, 15 October 2008 (UTC)

I would suggest, pretty much where it does now in the live version of {cite journal}, ie:

• Author, A. (2008). "A paper", Some Journal 42 (3):pp.123–125. Berlin: Akademie Verlag. doi/issn/oclc/etc..

This deviates slightly from the current {cite book} presentation— in {cite journal} loc:publ follows the pages field, while in {cite book} it precedes it. But I don't see that as being much of an issue, and it would make sense not to separate volume (issue) from pages. Possibly {cite book} could be amended so that pages precedes loc:publ (in a bibliographical listing pages wld usually refer to the chapter being cited, ie it tells you where to find the contribution being cited within a larger body, analogous to a paper in a journal). --cjllw ʘ TALK 23:13, 15 October 2008 (UTC)

I agree that this seems to be the best solution. Does anyone else feel that the loss of consistency between the two types of resource is more important than a common-sense layout as CJLLW proposed? If not, I'll ensure it's enacted. Martin (Smith609 – Talk) 01:50, 16 October 2008 (UTC)

3) Handling of blank date parameter. IIRC previous version of cite journal could handle the presence of a blank |date= parameter, if |year= was used instead (say, when someone copies the template parm list from the doco pg and fills it out, leaving date empty):

{{cite journal |last=Campbell |first=Lyle |year=1986 |date= |title=A paper |journal=A Journal}}

instead of getting

Campbell, Lyle (1986), "A paper", A Journal

now the year is gone altogether:

• Campbell, Lyle (1986). "A paper". A Journal.

Have found a few in this situation with the 'year' now disappeared because of this effect, & since I'd spose lots of folks copy the template's parms from the doco and neglect to trim out those not used, there's potential for many more out there. Also, looks like the |month= parameter is now non-functional. As for the "WORSE style issues", as mentioned there are perfectly valid reasons why some incl. myself have preferred cite XXX over citation. Such as, the awkward/unnecessary "Last, First" reversal for all of the authors' names in a multiauthored work, instead of the first only. While this may be usual for some disciplines in 'hard' sciences or mathematics, for others it ain't. But won't get into those stylistic issues ATM, before that wld like to see what's proposed to address those noted. --cjllw ʘ TALK 09:52, 3 October 2008 (UTC)

At last, something concrete to address! Thanks, CJLLW.

It seems the best way to fix (1) is to replace the & with a ;. Are there any disadvantages to that?

(3) ought to be easy to fix; hopefully someone will be willing to take a glance at the code and sort it (otherwise I will do so at the weekend)

(2) has two obvious solutions: (a) change "Cite journal" (and thus "citation") to resemble "Cite book"; (b) change "Cite book" to resemble "Cite journal". If Cite book should use the same citation format as "cite journal", then long term the only way to achieve this is for them to use the same central code to create their output. Shall I look into doing that and post up a sandbox version for testing? Martin (Smith609 – Talk) 14:02, 3 October 2008 (UTC)

Would be nice if it could also support a choice of styles [6] [7]. This perceived breakage of the code is kicking more interest than the deathly wait that this template has had with Waterfall model of having perfect requirements, code and documents ! Good luck. Shyamal (talk) 14:11, 3 October 2008 (UTC)

I'd use a comma for (1) -- that is what is STILL documented as an author separation parameter. (2) should implement solution (a), as there are more Cite XXX templates than cite book and cite journal & they used to all match fairly well.

Consensus can change. Is there anyone besides Martin, Shyamal, and Philcha who think that we should put up with the new/broken system (rather than reverting, fixing the new system, and trying the change again)? There seems to be at least as many people who think that the change was premature. I still fail to see the downside in this. --Karnesky (talk) 15:05, 3 October 2008 (UTC)

Now that we've got quite a few editors interested in this, we've got the support to test the new code if we revert in the mean time. Pagrashtak 15:17, 3 October 2008 (UTC)

Karnesky, please don't misunderstand me. I'm not in favour of putting up with any broken system, I'm concerned that ad hoc changes will just leave us wandering from one mess to the next. I'd rather sort it out properly, from the foundations up.

Shyamal, while the Waterfall model has its frustrations it may be necessary in this case, where we're dealing with a lot of inter-related sub-systems and a lot of stake-holders. IMO the problem is that the Waterfall model has not started yet, because there's no commitment to produce a draft spec that covers all variations. -- Philcha (talk)

My suggestions- for issue (1), the unwanted ampersand, replacing it with a semicolon or comma may cause a further issue. I gather the code for {citation} is designed to use individual lastn, firstn fields for each of the authors' names, and it's able to correctly work out where to put the & (between the penultimate and ultimate names):

• Campbell, Lyle; Kaufman, Terrence; Andrews, J. Richard; Smith-Stark, Thomas C. (2000), "a paper", a journal

Otherwise it uses a semicolon to separate authors' names, and commas in btw last, first since it always reverses these. The placement of the & works OK if the individual last, first fields are used. However, as noted most of the {cite XXX} templates have/had only a single coauthors field for multiple authors, the placement of the ampersand doesn't work for these. Since all of the multi-author cite journals, cite books, etc entered to date use coauthors—which the {citation} template code treats as a single field/author—then to make it work for both the last-first and coauthors systems you'd need to include a conditional test so that if coauthors is used don't insert the &, while if last, first is used then you can.

For (2), I have no problem with there being a central code generating the various outputs. However, I would strongly recommend the output to be formatted as per the original cite journal, which corresponds with cite book, cite conference, cite web and (to a lesser extent) cite news. This would cause the minimal disruption.

For (3), I presume there'd be a ready enough fix to handle blank date parameters, and also to bring back the month parameter as it originally functioned. --cjllw ʘ TALK 10:00, 6 October 2008 (UTC)

Anther issue I noticed (and which is why I reverted to the pre-change version) is that the formatting would be inconsistent on the same article, depending on whether a page number was called on a previous template. The change to the template was ill-advised as there was no broad solicitation for input; citation templates and {{citation}} operate in different spheres and should continue to do so; otherwise, seeing as the cite xx templates are much more widely used, citation should be changed to match them. Der Wohltemperierte Fuchs ^(talk) 23:30, 6 October 2008 (UTC)