This is an archive of past discussions. Do not edit the contents of this page. If you wish to start a new discussion or revive an old one, please do so on the current talk page.

I'd like to request the addition of a trans_quote parameter that would be, at least, a repository for a translation of a quote in a foreign-language source. (I'd be happy enough if the parameter were just ignored for display purposes, as long as the quote translation is available to someone who digs deep enough, though if anybody has better ideas about display, cool.) I've actually used this parameter in a few places in hopes that it might exist someday, but that's become problematic now that unknown parameters are showing an error message. Can we add this? —chaos5023 (talk) 17:30, 4 June 2013 (UTC)

[Not sure whether this belongs here or on one of the related talk pages]

At some point, it's going to be useful to have the dates in citations made machine-readable, so that they can be included in emitted metadata. That could be achieved in a number of ways, for example, having separate parameters for their day, month and year; using YYYY-MM-DD format, or using a subtemplate. Each has advantages and disadvantages. In the first two examples, it's still possible to have the rendered output in the form "17 May 2013" or "May 17th, 2013", according to a setting (see ((Start date)) or ((Birth date)) for examples if this in practice), or even user preference. Anyone got any thoughts? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 15:32, 17 May 2013 (UTC)

Dates used in citations are not always simple dates - you will get things like 13–19 December 2011, or Summer 1975. This would make this sort of automatic generation of metadata difficult.Nigel Ish (talk) 15:47, 17 May 2013 (UTC)

Not at all; we already manage to emit metadata for regular birth/ death dates, while accommodating (and not emitting as metadata) irregular dates such as "fl. 1735". Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:00, 17 May 2013 (UTC)

No thanks. Please don't try to bring back something that was binned two years ago. Machines are perfectly capable of parsing all the date formats that are permitted under our style guidelines, witness the dates that inhabit ((persondata)). Thus extra Date Autoformatting templates that consume extra processing and compilation resources without bringing significant benefits should not be contemplated. -- Ohc ^¡digame!_{¿que pasa?} 15:54, 17 May 2013 (UTC)

I'm not "trying to bring back something that was binned two years ago". I'm proposing something new. Persondata is only read by our own tools; not generic metadata parsers, like the ones that read dates in our infoboxes. We already use date templates such as the ones I mention, elsewhere; with no undue overheads; and with significant benefits. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:01, 17 May 2013 (UTC)

I was merely using ((persondata)) as an example. Our MOSNUM-permitted date formats, of which there are only three, are all capable of being parsed by machine, all that is needed is software configuration that's not rocket science. In the average article, there may be up to 200 references, and thus up to 600 dates. Each set of dates is specific to one citation but most are unimportant. If each set of dates is already sitting in a citation template, we hardly need to nest in another three for each date type. If the dates are not already in citation templates, metadata is next to useless. Without contemplating in detail the computing resources consumed, and the benefit of standardising something that scarcely needs standardising is potentially highly disruptive to the 'pedia. -- Ohc ^¡digame!_{¿que pasa?} 16:37, 17 May 2013 (UTC)

Nigel Ish, above, has already shown that there are more than three ways to enter date information. Using a sub-template was only one of the example methods I gave; there are others. If we want to include machine readable dates to standard metadata formats like COinS or a microformat, then they need to be formatted consistently, not as prose. Your comments about potential for disruption are a straw man. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:59, 17 May 2013 (UTC)

:oppose until there is a way to set the default date on a page or to read the user preference date. --  Gadget850 ^talk 16:10, 17 May 2013 (UTC)

What do you mean by "default date on a page"? User preference for date formatting could be achieved by the same CSS technique that ((Coord)) uses for coordinates. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 16:36, 17 May 2013 (UTC)

That CSS simply places the displayed coordinates. There is no magic word or parser function that will unify all dates on a page, thus citation template can't inherit the date format. The only user preference that I know can be read is gender. There was a 'dateformat' parameter for a brief time after date linking was removed, but there were technical issues an it was removed; you can check the archives. --  Gadget850 ^talk 17:16, 17 May 2013 (UTC)

That (your first sentence) simply isn't the case; see Template:Coord#Per-user display customization. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:37, 17 May 2013 (UTC)

Currently dates in citations are not machine readable because there is no indication what calendar they are stated in, Julian or Gregorian. I suggest Andy go discuss with the various metadata folks about how they should modify their formats to indicate a calendar of Julian, Gregorian, or unknown, and come back when the metadata formats have been suitably modified. Jc3s5h (talk) 16:57, 17 May 2013 (UTC)

This is already resolved for the date templates mentioned above (and which are used on many thousands of articles); see their documentation. It is also resolved in the metadata standards I mentioned. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 17:11, 17 May 2013 (UTC)

Wait. Your original post is confusing, but it looks like this has nothing to do with date formats, and I don't know why you threw that in. You simply want to include the date metadata, which is very possible. We could add an algorithm that would convert the standard date that is displayed to a metadata date that is now shown. For example, May 17, 2013 would be shown and converted to <time datetime="2013-05-17"> which would not be shown. <time> is an HTML metadata element now supported by MediaWiki. --  Gadget850 ^talk 17:27, 17 May 2013 (UTC)

I want to include the date metadata, and pointed out that how we store the values does not need to limit the display format. But yes, we could do as up suggest (providing we don't change the actual date ;-) )Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 18:37, 17 May 2013 (UTC)

We've moved to what is, IMO, a very desirable simplicity in both the display and the syntaxing of chronological items. When you say, "separate parameters for their day, month and year", I start to quail that the whole system will become complex—which is not good for the "anyone can edit" principle. I'm also quite unconvinced that metadata systems cannot be developed without the use of localised syntax. And we already have significant tools at our disposal for searching for years and dates. Tony (talk) 04:37, 18 May 2013 (UTC)

This isn't about "searching for years and dates" within Wikipedia. I'm not clear what you mean by your "localised syntax" comment - please clarify, bearing in mind that we already do this for other dates, mentioned above. Separate parameters are only suggested as one of several options what's your proposed solution? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:07, 18 May 2013 (UTC)

We don't need separate parameters. There is no need to change the template input, thus this would be user transparent. <time> is a standard HTML5 tag. --  Gadget850 ^talk 10:16, 18 May 2013 (UTC)

4.6.10 The time element at W3C --Redrose64 (talk) 11:00, 18 May 2013 (UTC)

How could that be deployed, in this case? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:07, 18 May 2013 (UTC)

Let's say that a given use of ((cite news)) emits <span class="citation news">Lane, Lois (18 May 2013). "Superman prevents another trainwreck". ''Daily Planet''.</span>. All of that is generated by Module:Citation/CS1 using the ((cite news)) parameters as input. We could amend the module so that it tests to see if |date= is a valid date (to cover the situations like 13–19 December 2011, or Summer 1975 mentioned above). If the test succeeds, we put the date through the Lua equivalent of ((#time:Y-m-d)) to generate a valid date string, place that in the datetime= attribute of a <time>...</time> element which is wrapped around the original date, and output it with the rest, something like <span class="citation news">Lane, Lois (<time datetime="2013-05-18">18 May 2013</time>). "Superman prevents another trainwreck". ''Daily Planet''.</span> --Redrose64 (talk) 12:07, 18 May 2013 (UTC) Sorry. I misread the documentation for the time element, so I've rewritten the approprate bits here. --Redrose64 (talk) 16:25, 18 May 2013 (UTC)

OK, so that deals with publication dates. I don't understand to what use the metadata would be put, so please explain to Dummy here how exactly this would be done in the case of access dates, archive dates, and dates within the titles? And to what end? -- Ohc ^¡digame!_{¿que pasa?} 12:42, 18 May 2013 (UTC)

We could use exactly the same technique for access dates and archive dates, in fact these should be easier because they ought to only ever be a single date, and not cases like 13–19 December 2011, or Summer 1975. We shouldn't try to detect dates in titles, these are pretty much free-form, and trying to write code to extract a date sensibly would be a nightmare, and I don't think it's worth the hassle. As regards "to what end"; well, that's up to the harvesters of metadata. --Redrose64 (talk) 16:15, 18 May 2013 (UTC)

Does seem like typical YAGNI request: I am also not convinced the date format even matters, here, because any metadata-processing tools could reformat whatever dates, on their end, and using their processing time. So, beware typical YAGNI options ("You aren't gonna need it"). Instead, wait for the issue to become a top headline in a major tech newspaper:

• "Riot at Wikimedia when users screamed date format more important than fixing Edit-conflicts" (just not likely)

There are probably some huge numbers of feature requests that are just typical YAGNI requests, and so it is a common problem which distracts from fixing the major problems, such as not external-linking the "trans_title=" to the URL address, to allow the trans_title to contain wikilinks, or footnotes, which could better explain the translation. -Wikid77 16:58, 18 May 2013 (UTC)

You seem to present this as an "either/or" choice: a false dichotomy. And that's before we consider the valid use-cases. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 20:15, 18 May 2013 (UTC)

Well, instead of dismissing the opposition to the idea of collecting it, you ought to demonstrate a solid need on our part for ever-increasing quantities of metadata when the data is already in existence and in an exploitable form. Let's have some valid use-cases, then, instead of just saying or implying we need it. If Google (or any other third party) needs it, they can jolly well create it from what we give them – if they are not doing it already. ;-) -- Ohc ^¡digame!_{¿que pasa?} 05:09, 19 May 2013 (UTC)

It's not in a (readily, to machines) exploitable form; I'm proposing that we make it so. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:32, 22 May 2013 (UTC)

• strong support -- i'm not a coder, but i can certainly see the usefulness of this proposal. it would be USEFUL both for us & for external (i.e.: outside of wp/wm/mw) uses. calling it "yagni" it just a fancy way of saying "can't be arsed"; it's a good idea, & it anticipates (a wide range of) future needs & uses. to put it another way: if we don't do it now (machine-readable dates, as per ISO 8601), we WILL be doing it later...

Lx 121 (talk) 20:02, 5 June 2013 (UTC)

as for the question of "extra processing", wouldn't it SAVE processing, if we standardize everything/most-things into one date format? Lx 121 (talk) 20:07, 5 June 2013 (UTC)

Lx 121, ISO 8601 is not suitable for Wikipedia dates. Go read it and let us know if the reason is as obvious to you as it is to me. Jc3s5h (talk) 23:18, 5 June 2013 (UTC)

Pure FUD. You've been banging on about this for years; yet we use ISO 8601 for dates in templates like those listed above, on many thousands of articles. It's trivial, in Lua, to put switches in to suppress metadata emission for pre-Gregorian dates. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 23:51, 5 June 2013 (UTC)

Yea, just like your obsession with all things 'metadata' is well known. You still haven't elaborated how "useful" it is, despite having been asked more than once. -- Ohc ^¡digame!_{¿que pasa?} 02:07, 6 June 2013 (UTC)

I noticed that there are a couple of entries in the International Phonetic Alphabet article, Further reading section that are having trouble with the use of the cite book template, because an editor entered two ISBNs for the |isbn= parameter. The books apparently do have two ISBNs: one each for hard bound and for paperback. It seems that the only way to handle this now is to make two separate citation entries, which seems overkill. So how about two parameters that can produce two ISBNs in the listing (and differentiate hb from pb)? Other solutions are fine with me. I'm just looking for some unclunky way to handle such a situation. Evenssteven (talk) 02:20, 31 May 2013 (UTC)

The idea of two parameters for different ISBNs has been suggested before, several times. Where would we stop? Some books have more than two ISBNs - there are deluxe/limited editions/gift sets, eBooks, abridged versions, large print, etc. My copy of the Concise Oxford Dictionary has two ISBNs on the copyright page - both are for hardback editions, the only difference is whether there's a thumb index or not. The back cover shows just one.

When this topic comes up with ref sources, the usual response is: give only the ISBN for the edition which was actually used. I think that for books given under "Further reading", the same principle applies: you are recommending a particular book, so you must have read it yourself, therefore give the details for the edition which you read, which will have only one ISBN. --Redrose64 (talk) 09:52, 31 May 2013 (UTC)

Ok, that seems a perfectly reasonable rationale that I can use in my own writing. Do you have a suggestion about how I might go about editing the International Phonetic Alphabet article? I don't know which edition the person who put in the ref had read, and I myself have read neither. So let me answer my own query, and you can supply a better one if you know. Since I am in doubt, or don't know something, I am the wrong person to edit the ref. WP is allowed to be imperfect, and if I don't know or expect my change to be an improvement, then it's really not much of a contribution, and so I walk away from that one. Thanks for the excellent advice; practical on two counts. Evenssteven (talk) 12:33, 31 May 2013 (UTC)

There are two possibilities: one is to alter |isbn=0-521-65236-7 (hb); ISBN 0-521-63751-1 (pb))) to |isbn=0-521-65236-7 )) ISBN 0-521-63751-1 or to |isbn=0-521-65236-7 )) - either way, the hardback one is the one that I'd leave inside the ((cite book)). Do the same with the other one as well. --Redrose64 (talk) 13:27, 31 May 2013 (UTC)

"the only way to handle this now" Just want to point out that this never worked properly, as the link to Special:BookSources would contain two ISBNs, which is parsed as a single ISBN, thus is invalid. The templates now do error checking on the ISBN. --  Gadget850 ^talk 13:38, 31 May 2013 (UTC)

I think |isbn=, like any other parameter expecting a single value, should have only a single value; perhaps this should be documented somewhere? In this case the second ISBN should be added following the template. ~ J. Johnson (JJ) (talk) 20:24, 31 May 2013 (UTC)

I thought that's what I put above... nevertheless, I've added a warning to the doc page. --Redrose64 (talk) 23:01, 31 May 2013 (UTC)

I've run into this problem as well. Why not optionally allow a comma-separated list of ISBNs to be given (separated by either "," or ";"). If it is undesirable to link them all, just extract the first one for the link and display the remainder as normal text. Or add support for the comment= parameter I proposed already (Help talk:Citation_Style_1/Archive_2#Proposal_for_comment_parameter), where people could add such extra information in free format. Adding it after the template sounds like a work-around (which sometimes could look rather strange), not a solution. --Matthiaspaul (talk) 09:00, 6 June 2013 (UTC)

The ISBN field is not for identifying all available copies, it is to identify the particular source you read. You can stuff whatever you want in the |id= parameter. If you want 16 ISBNs, then go for it. But you must have read each of them and ensured the page numbers match. WP:SAYWHEREYOUGOTIT. I'm tired of repeating this, and will let it fall out when you try to take the article to FA. --  Gadget850 ^talk 09:28, 6 June 2013 (UTC)

I beg to pardon, but there is not the slightest reason to be aggressive. As has been pointed out by others, there are books, which carry more than one ISBN on a single sample of the book (I've seen up to four). There are also cases, where someone has more than one edition of a book, each with its own ISBN, and has read both of them to be sure that they contain the same information. Of course, we could just pick one of them, but we could also just allow more than one ISBN. The later would be more correct, IMHO. --Matthiaspaul (talk) 09:45, 6 June 2013 (UTC)

And how do you plan to handle different dates? --  Gadget850 ^talk 09:36, 6 June 2013 (UTC)

I haven't seen this in combination with differing dates so far, but just in case, I would propose to handle it in the same way as comma-separated list. Alternatively, the first1 last1 first2 last2 approach could be used here as well, isbn[1], date[1], isbn2, date2, ... --Matthiaspaul (talk) 09:45, 6 June 2013 (UTC)

Here you go:

• Drucker, Sam. The World of ISBNs. ISBN [[Special:BookSources/978-0812695939 (1999) p. 23 (hardback); ISBN 978-0812695935 Parameter error in ((ISBN)): checksum (2001) p. 26 (paperback); ISBN 978-0812691234 Parameter error in ((ISBN)): checksum (2010) loc. 123-987 (Kindle); ISBN 978-0812691234 Parameter error in ((ISBN)): checksum (2013) p.22 (Kindle Fire); ISBN 978-0812345935 Parameter error in ((ISBN)): checksum (2013) p.32 (ePub)|978-0812695939 (1999) p. 23 (hardback); '"`UNIQ--templatestyles-00000009-QINU`"'[[ISBN (identifier)|ISBN]]&nbsp;[[Special:BookSources/978-0812695935 |978-0812695935]]<span class="error" style="font-size:100%">&nbsp;Parameter error in (([[Template:ISBN|ISBN]])):&nbsp;checksum</span> (2001) p. 26 (paperback); '"`UNIQ--templatestyles-0000000B-QINU`"'[[ISBN (identifier)|ISBN]]&nbsp;[[Special:BookSources/978-0812691234 |978-0812691234]]<span class="error" style="font-size:100%">&nbsp;Parameter error in (([[Template:ISBN|ISBN]])):&nbsp;checksum</span> (2010) loc. 123-987 (Kindle); '"`UNIQ--templatestyles-0000000C-QINU`"'[[ISBN (identifier)|ISBN]]&nbsp;[[Special:BookSources/978-0812691234 |978-0812691234]]<span class="error" style="font-size:100%">&nbsp;Parameter error in (([[Template:ISBN|ISBN]])):&nbsp;checksum</span> (2013) p.22 (Kindle Fire); '"`UNIQ--templatestyles-0000000D-QINU`"'[[ISBN (identifier)|ISBN]]&nbsp;[[Special:BookSources/978-0812345935 |978-0812345935]]<span class="error" style="font-size:100%">&nbsp;Parameter error in (([[Template:ISBN|ISBN]])):&nbsp;checksum</span> (2013) p.32 (ePub)]]. ((cite book)): Check |isbn= value: invalid character (help); templatestyles stripmarker in |ISBN= at position 41 (help)

Infinite diversity in infinite combinations. --  Gadget850 ^talk 10:10, 6 June 2013 (UTC)

I used the citation ((cite journal|last=Todd|first=Andrew S|coauthors=R. Mark Sattelberg|title=Actinides in Deer Tissues at the Rocky Flats Environmental Technology Site|journal=Integrated Environmental Assessment and Management|year=2005|month=November|volume=1|issue=4|pages=391–396|pmid=16639905‎|url=http://onlinelibrary.wiley.com/doi/10.1002/ieam.5630010408/pdf|accessdate=9 June 2013)), and the PMID external link in the reference takes me here instead of here. Am a doing something wrong, or is there an error with the template? Thanks! VQuakr (talk) 19:23, 9 June 2013 (UTC)

Here is a version of your citation with most stuff removed. I notice that when PMID is immediately followed by a pipe or vertical bar (|) then the PMID parameter gets corrupted somehow:

((cite journal |title=Actinides in Deer Tissues at the Rocky Flats Environmental Technology Site|pmid=16639905‎|volume=1))

→"Actinides in Deer Tissues at the Rocky Flats Environmental Technology Site". 1. PMID 16639905‎. ((cite journal)): Check |pmid= value (help); Cite journal requires |journal= (help)

If I move |pmid= to the end of the citation then there isn't a problem:

((cite journal |title=Actinides in Deer Tissues at the Rocky Flats Environmental Technology Site‎|volume=1|pmid=16639905))

→"Actinides in Deer Tissues at the Rocky Flats Environmental Technology Site‎". 1. PMID 16639905. ((cite journal)): Cite journal requires |journal= (help)

So, that's the work around until the problem is fixed.

—Trappist the monk (talk) 20:02, 9 June 2013 (UTC)

The problem is that there is an invisible character immediately after the 16639905, specifically a U+200E (Left-to-right mark). It may be removed by placing the cursor somewhere within the PMID value (but before its end), marking text from there to somewhere after the pipe, pressing Delete and retyping:

• Todd, Andrew S (2005). "Actinides in Deer Tissues at the Rocky Flats Environmental Technology Site". Integrated Environmental Assessment and Management. 1 (4): 391–396. PMID 16639905. Retrieved 9 June 2013. ((cite journal)): Unknown parameter |coauthors= ignored (|author= suggested) (help); Unknown parameter |month= ignored (help)

Not directly related, but in templates with more than two or three named parameters, I normally put a space just before each pipe to aid visual separation - trying to spot where one parameter ends and the next begins is more difficult when they're all jammed together. --Redrose64 (talk) 22:48, 9 June 2013 (UTC)

Many thanks to both of you for the rapid feedback! VQuakr (talk) 04:30, 10 June 2013 (UTC)

In ((cite book| editorlink = Robert J. Sternberg| last = Hyman | first = Ray| authorlink = Ray_Hyman|| editor1 = Robert J. Sternberg |editor2 = Henry L. Roediger| editor3 = Diane F. Halpern| title = Critical Thinking in Psychology| url = http://books.google.com/?id=3mA9NPAgWR0C| year = 2007| publisher = Cambridge University Press| isbn = 978-0-521-60834-3| page = 218| chapter = Evaluating Parapsychological Claims ))

Hyman, Ray (2007). "Evaluating Parapsychological Claims". In Robert J. Sternberg; Henry L. Roediger; Diane F. Halpern (eds.). Critical Thinking in Psychology. Cambridge University Press. p. 218. ISBN 978-0-521-60834-3. ((cite book)): Cite has empty unknown parameter: |1= (help); Unknown parameter |editorlink= ignored (|editor-link= suggested) (help)

The url links from the title of the chapter, rather than the title of the book. (This link may still not be accurate, because I modified it from one with two authorlink fields, which was, in tern, modified from one which had first and last (referring to the chapter author) and authors (referring to the book authors, which I called editors). Shouldn't the url link from the book title. — Arthur Rubin (talk) 16:17, 10 June 2013 (UTC)

As can be seen from this comparison, the new version of ((cite book)) links the |chapter= with |url= just as the old ((citation/core)) version did:

I'm inclined to believe that |url= should link |title= and not |chapter= when |chapterurl= is omitted. Doing so complies with the ((cite book)) documentation.

Editor Arthur Rubin can add |chapterurl=http://books.google.com/books?id=3mA9NPAgWR0C&pg=PA216 to the citation as a work-around.

—Trappist the monk (talk) 10:57, 11 June 2013 (UTC)

Even if it's been that way since the old ((citation/core)) days, it still should be fixed.... — Arthur Rubin (talk) 14:02, 11 June 2013 (UTC)

I didn't say that is shouldn't be fixed. The initial goal of the conversion to the Lua module is and should be to be compatible with the existing ((citation/core)) templates. The point of the comparison was to show that the developers did not introduce a bug into Module:Citation/CS1.

—Trappist the monk (talk) 14:44, 11 June 2013 (UTC)

Perhaps someone familiar with this template's intricacies will know what to do: there are template errors showing on the documentation at WP:CS1#Type. I expect they are related to recent reorganization toward a LUA implementation. —EncMstr (talk) 17:34, 14 June 2013 (UTC)

 Fixed --  Gadget850 ^talk 17:38, 14 June 2013 (UTC)

Is there any way to have the 'publisher' break to a new line below the author's name, year and title? As it is, when the cite book items are all filled out, it generates a source listing that is entirely strung together on one line, often listing the publisher on the other side of the screen. The wrap around isn't much help because often times the only items that wrap to the next line are page numbers and isbn numbers, etc. When panning through a lengthy bibliography it is much easier to read the individual sources when the author's name, year and title are on one line, and the publisher, page number, isbn are on the second line, with the publisher always at the beginning of the second line. Since many cite book listings wrap around to a second line anyway, (unless you're using a very wide screen) it would be nice if this was accomplished with a little order and uniformity. Below is an example of how the listing would look. A <br> has been placed in the publisher field to accomplish this but in practice these are sometimes removed by bots. If there is reluctance to alter this cite book template could an alternative cite book template be created? e.g.'cite book2'. -- Gwillhickers (talk) 22:18, 16 June 2013 (UTC)

• Dorwart, Jeffery M.; Wolf, Jean K. (2001). The Philadelphia Navy Yard: From the Birth of the U.S. Navy to the Nuclear Age.
  University of Pennsylvania Press. p.271,. ISBN 0812235754.((cite book)): CS1 maint: extra punctuation (link)

• Dresel, H.G. (1896). United States Naval Institute Proceedings, Volume 22 (Naval warfare, tactics, weapons, etc,).
  United States Naval Institute, Annapolis. p. 866.

• Dull, Jonathan R. (2012). American Naval History, 1607-1865: Overcoming the Colonial Legacy.
  Univ of Nebraska Press. p. 216. ISBN 9780803244719.

That break is going to be stuffed into the COinS output and will be picked up by reference management software. A better solution is to add a class to the 'publisher' field. Then you can add the break in your personal CSS. --  Gadget850 ^talk 00:07, 17 June 2013 (UTC)

Okay, I'm more of a writer than I am a computer software hack. I have no idea what's going on with 'COinS', and I'm not clear on how you add a 'class' to the publisher field or about adding a break to CSS. Any help would be a big help. One more question: Will this only effect my viewing of bibliographies? I was thinking in terms of adding uniformity and readability to bibliographies for the sake of the readers. Many bibliographies are quite long, and when all the source items are just strung together on one line, line after line, it tends to make the bibliography look like a blur. Any help along these lines also would be most useful. -- Gwillhickers (talk) 03:25, 17 June 2013 (UTC)

COinS is invisible metadata that is readable by reference management software. If you stuff markup like a break into a field, then it pollutes the metadata. We have an open feature request to add classes to each field for microdata purposes. This would also allow readers to style fields as desired.

Adding a hard coded break is not the best solution, as it will probably mangle the display for mobile users, and it will create shortened lines for users with large displays. --  Gadget850 ^talk 08:34, 17 June 2013 (UTC)

A question recently came up at the Teahouse where an editor was trying to use cite-book for an eBook citation. The problem arose as for an eBook a page number is not really appropriate. There's a sort of workaround with using a "location (LOC)" within an eBook and the "at" parameter. However it seems the template and/or the documentation needs to be adjusted in order to help users wanting to cite specific locations within eBooks easily. --LukeSurl ^{t c} 21:45, 17 June 2013 (UTC)

Link to archived discussion where this was also discussed --LukeSurl ^{t c} 22:36, 17 June 2013 (UTC)

I'm the original user that had the issue. I've been thinking about this from the standpoint of a data designer. I think I may have been over thinking this. There are already various interpretations for the page number based on the edition, paperback, hardcover, etc. So the loc is just one more example. I think the right answer here might just be use the loc as a regular page number. Mdebellis (talk) 02:40, 18 June 2013 (UTC)

I don't understand why websites are italicised in references when we do not italicise them in any other form on Wikipedia. to me, this seems to be improper formatting, and unless there is a convincing argument in support of it, I propose to have it altered in our system so that websites are no longer italicised in references. an example of this: a reference to an article by the website AllMusic:

True, Chris. "Faith – The Cure: Songs, Reviews, Credits, Awards: AllMusic". AllMusic. AllRovi. Retrieved 15 June 2013.

notice that "AllMusic" is italicised. Lachlan Foley 03:39, 15 June 2013 (UTC)

This is a perennial question, and it comes down to this: the |work= parameter is treated as synonymous with |journal=, |newspaper=, |magazine= (and some others, see Module:Citation/CS1/Configuration and search for the entry beginning "['Periodical'] ="). This is the gross work; the |title= specifies an included work. Gross works are italicised; included works are quoted. --Redrose64 (talk) 06:54, 15 June 2013 (UTC)

In my experience, the "work" parameter isn't usually relevant in "cite web". You have the title of the page (title) and the entity who published it (publisher), which is usually the name of the website. I'm not even sure what "work" means in this context. Whereas in e.g. "cite news", "work" is the name of the publication and rightly appears in italics. In the AllMusic example that Lachlan Foley cites, AllMusic has been included in the title of the page so there is no need to also call it a "work". Although, since AllMusic is a well-known website, I'd have been inclined to call AllMusic the publisher and ignore the fact that the copyright is held by a wider entity called AllRovi. In that case, the inclusion of the word AllMusic in the page title would be redundant. At all events, this seems a rather untypical example. -- Alarics (talk) 07:24, 15 June 2013 (UTC)

there is a |website= parameter (which I have actually only just noticed), and |work= is the alias, meaning that |work= is the correct field to put the title of the website under. |publisher=, it seems to me, is for the entity in charge of or powering the website. the |title= field is for, in this case, the title of the page which the URL is linking to, not the website itself. Lachlan Foley 11:44, 15 June 2013 (UTC)

I have occasionally used |work= for the name of the section of the website or for the type of document this page is, when there are multiple pages of the same type elsewhere on the site.

Consider, in cite web,

|url=http://example.com/safety-directives/left-handed-scredrivers |title=Safety Directives - Safely Using Left-Handed Screwdrivers |publisher=The Safety Society

vs.

|url=http://example.com/safety-directives/left-handed-scredrivers |title=Safely Using Left-Handed Screwdrivers |work=Safety Directives |publisher=The Safety Society

-- 31.54.63.131 (talk) 19:08, 22 June 2013 (UTC)

In your example it is a "work". As you point out it is not the publisher, AllRovi is. It seems quite properly cited. --Bejnar (talk) 07:49, 15 June 2013 (UTC)

• I tend to view these from a practical standpoint – they are the same sides of a coin but used to manage italicisation: |work= italicised, |published= doesn't. The notion of 'publisher' is much more important for books, and the ((citation)) templates instruct us not to cite publishers for periodicals. I would never put Allmusic in the title. It needs to go in |publisher= or |work= depending on whether it is properly italicised or not. -- Ohc ^¡digame!_{¿que pasa?} 11:38, 15 June 2013 (UTC)

in response to Bejnar: it is properly cited, the issue is that "AllMusic" is italicised in the reference, and websites are not italicised anywhere else on Wikipedia, making this a flaw in the citing system. if your message was directed at Alarics, ignore this. Lachlan Foley 11:44, 15 June 2013 (UTC)

Is AllMusic the main work? If it should not be italicized, then should a web page from AllMusic be in quotes as the included work? --  Gadget850 ^talk 08:38, 17 June 2013 (UTC)

It seems that ((Cite book)) does not display the value in the |others= parameter, such as this reference from Johann Heinrich Alsted:

• ((cite book | first = Paolo| last = Rossi | authorlink = | coauthors = | year = 2000 | month = | title = Logic and the Art of Memory | chapter = |others=Translated by Stephen Clucas | others = | edition = | pages = | publisher = University of Chicago Press | location = | id = | url = | isbn = 0-226-72826-9 ))
• Rossi, Paolo (2000). Logic and the Art of Memory. University of Chicago Press. ISBN 0-226-72826-9. ((cite book)): Cite has empty unknown parameters: |coauthors= and |month= (help)

GoingBatty (talk) 01:02, 24 June 2013 (UTC)

It shows the value of the second |others= which is blank. --  Gadget850 ^talk 01:10, 24 June 2013 (UTC)

Facepalm - sorry for not seeing the obvious - thanks! GoingBatty (talk) 01:20, 24 June 2013 (UTC)

Perhaps others here will be interested by what's going on at d:Wikidata:Books_task_force. Seems like there's some commonality of interest. LeadSongDog come howl! 21:04, 28 June 2013 (UTC)

Your link doesn't appear to be correct. It goes to a blank page. Dragons flight (talk) 21:50, 28 June 2013 (UTC)

I've fixed the link. — Mr. Stradivarius ^{♪ talk ♪} 02:20, 29 June 2013 (UTC)

Is there a reason why "edition" is not mentioned in the full parameter set of Template:Cite journal? --82.170.113.123 (talk) 20:37, 23 June 2013 (UTC)

No. I have updated everything but those full parameter sets, as I don't think they are a good idea, so I am leaving someone else to do it. --  Gadget850 ^talk 20:41, 23 June 2013 (UTC)

Edition is rarely useful for journals. Books have editions, since a given book may be revised from time to time, each update gets a new edition but is substantially the same as the previous. Journals (or any other periodicals) do have progressive identifiers, but these are dates, or volume/issue numbers - the change from one issue to the next is close to 100%. To be honest, I've only come across two instances of a single issue of a periodical going through more than one edition - in both cases an article had to be removed for legal reasons. In one case, the "second edition" got a suffixed issue number (i.e. 1234a where the original was 1234); in the other, the only way you could tell that you had a "second edition" was because the page numbering on a particular two-page spread was 34/37. --Redrose64 (talk) 20:55, 23 June 2013 (UTC)

Just for the record, I think these full parameter sets are very useful. I usually start with those, fill in everything I can find and if any of the parameters are unclear I look for information further down these template pages. To give you an example, I just copied the full parameter set of Template:Cite journal, filled in whatever I could find and voilà: [1] Too many templates to know by heart, no time to read through the entire Parameters section, and frequently the most commonly used parameters do not include some additional information I can provide. --82.170.113.123 (talk) 21:01, 23 June 2013 (UTC)

Could you please provide an example of an article that contains a reference with ((cite journal)) where |edition= would be useful? Thanks! GoingBatty (talk) 23:05, 23 June 2013 (UTC)

Me? I have no idea, maybe it's not useful anywhere. I only noticed it wasn't part of the full parameter set, that's all. --82.170.113.123 (talk) 10:09, 24 June 2013 (UTC)

Since cite journal is used for magazines as well as journals, how would we denote that an article was published in a specific national edition of a news magazine, but not in the regular edition? I'm thinking of an article in the US edition of The Economist, which is published in the UK. I used the edition parameter in this footnote for instance for that reason. Imzadi 1979 → 15:16, 24 June 2013 (UTC)

That's why we have |location=. In this instance you would put |location=New York. --Redrose64 (talk) 17:37, 24 June 2013 (UTC)

Except that the news database I consulted specifically stated "U.S. edition" and didn't mention location. Based on the publication's website, how am I to determine based on the information given if it was published in New York or San Francisco? (They have offices in both locations, neither of which is specified as originating the U.S. edition.) How would either U.S. location inform a reader seeking to find a copy of the article, when the databases categorize it as the "U.S. edition"? Yes, there are limited uses for the parameter, but that doesn't mean there are no uses for it. Imzadi 1979 → 17:58, 24 June 2013 (UTC)

In that case, how about putting the title as "Economist US edition" and maintaining the location as London? It is still essentially a London-produced magazine, I think -- or does it have a whole team people in USA producing an entirely different publication from the UK one? -- Alarics (talk) 11:48, 29 June 2013 (UTC)

Another suggestion for formatting references in order to make them easier to read. When cite web references occupy two or more columns, start the text "archived from the original on {date}" on a new line. Additionally, start any quoted text on a new line. These changes would rarely result in the reference occupying more lines in total. - 109.176.243.180 (talk) 18:30, 29 June 2013 (UTC)

I don't think it's possible to set up fixed-position line breaks to apply only when there is a columnar layout. We could set a forced line break at one or another of these positions, but not on a dynamic basis: they would always be present, regardless of the number of columns. That said, when the CSS Multi-column Layout Module gets promoted from Candidate Recommendation to W3C Recommendation, browser writers should then start supporting the Column breaks feature, and it may then be possible to force column breaks to occur between two list items and not inside a list item. --Redrose64 (talk) 18:47, 29 June 2013 (UTC)

((Reflist)) cannot see the content inside <ref>...</ref> tags- it simply applies the column styling to <references /> which is the tag used by the Cite software to generate the reference list. --  Gadget850 ^talk 19:39, 29 June 2013 (UTC)

((Multiple issues)) uses CSS to change the look of the embedded templates. Could the same technique be used here, so that ((reflist)) exposes newlines inside the cite templates that are normally invisible? -- John of Reading (talk) 05:08, 30 June 2013 (UTC)

Is this a question about a single reference that happens to span two columns, or about all references within a multi-column list? I suspect the latter. Starting a new line for archives and quotes would appear to be more readable, however in a single column list the suggestion would always result in many more lines being used. --212.139.98.247 (talk) 20:47, 29 June 2013 (UTC)

@John of Reading: Yes, that is possible. Two things need to be done: first, we need to set up a new class - say reflist-br-condit - and set up the global CSS for that, like this

br.reflist-br-condit {
  display: none;
}
.references-column-count br.reflist-br-condit {
  display: inline;
}

Second, we would modify Module:Citation/CS1 to emit <br class=reflist-br-condit /> at suitable points. Linebreaks would then be visible in reflists, only when the multi-column feature is used. --Redrose64 (talk) 12:58, 30 June 2013 (UTC)

It's very common these days for web-based material have both a date and time of publication. To my knowledge there's no mechanism to include the time information in the cite templates. I suppose the current closet place for it to be put is in the "date" parameter but that's not what that is meant for. Should a new "time" parameter be introduced? Of course, the question is begged: is this level of granularity necessary for the references? I believe it is valuable when included for two reasons. The time would be useful when used in search engines to find a lost source (e.g. current url is dead) and a time-stamp match would produce confidence that one has actually re-located the proper missing source (date alone is insufficient as big news events often have multiple stories released on the same day). It would be especially useful for the ((cite web)) and ((cite news)) templates. Ideas? Jason Quinn (talk) 04:24, 2 July 2013 (UTC)

@Jason Quinn: - Could you please provide an example of a web page where the time is relevant? (i.e. a reliable source that has content posted for less than a day) Thanks!

I would imagine, for a "breaking news" story, where details arrive in the newsroom at random intervals and they then update the web page as and when they have more information. This story, for example, is currently marked 6:23AM BST 03 Jul 2013 - that's 06:23 British Summer Time, which is 05:23 UTC. Check back in a couple of hours. --Redrose64 (talk) 07:43, 3 July 2013 (UTC)

The CS1 templates do support 'time' and 'timecaption':

((cite web |title=Title |url=http://www.example.org |date=July 3, 2013 |time=6:21AM BST |timecaption=​))

'timecaption' defaults to 'Event occurs at ', so I worked around by using a zero width space. If this is going to be used, then we should add an explicit 'none' value to suppress the timecaption, as there is a hard-coded space between 'timecaption' and 'time'. --  Gadget850 ^talk 10:34, 3 July 2013 (UTC)

The time and timecaption are documented at ((Cite video)). Their purpose is to document where in a video the relevant information may be found. This is a different meaning than the time the information was published. There is a potential conflict. Updated videos about a developing news story may be posted by a reliable source. In principle, it could be useful to specify both when the update was published, and at what time within the video the information being cited may be viewed. Although, news stories tend to be short so in most cases the reader would just view the whole story. Jc3s5h (talk) 22:36, 3 July 2013 (UTC)

With the Lua-based templates, all or most parameters are available to all templates. The key is context. And ((Cite video)) is now ((Cite AV media)). --  Gadget850 ^talk 22:55, 3 July 2013 (UTC)

It isn't unusual for templates to be switched from, for example, Cite AV media to Cite web or Cite news when it is determined the source really falls into a different category. It is a bad practice to use the same parameter name for substantially different concepts. Jc3s5h (talk) 23:11, 3 July 2013 (UTC)

We already do it: 'title' in ((cite web)) for a page included in the 'website/work', as compared to 'chapter' for the work included in the 'title'. Which causes problems for some editors. But, time is time— whether is is the time of publication or the time an even occurs. Again, it is a matter of context— this is the first time I am aware that someone wanted to include a time in a source other than audiovisual. Perhaps 'at' would be the better field. --  Gadget850 ^talk 23:23, 3 July 2013 (UTC)

The variable meaning of title is a perfect example of what not to do. As for "at", you mean that videos should use the at parameter rather than the minutes or time parameter to specify where within the video the relevant information is? Not unreasonable.

If we're going to stretch the meaning of an existing parameter to include time, I suggest redocumenting the date and accessdate parameters to state they can include publication or access time when that would be helpful. This would be less of a conceptual difference than the difference between time within a video vs. publication time. Jc3s5h (talk) 00:35, 4 July 2013 (UTC)

'date' is problematic, as it emits COinS metadata. --  Gadget850 ^talk 00:51, 4 July 2013 (UTC)

Consider a more complete example:

((cite web |title=Title |url=http://www.example.org |date=July 3, 2013 |time=6:21AM BST |timecaption=&#8203;| last1 = Doe| first1 = John))

renders as

Doe, John (July 3, 2013). "Title". Event occurs at 6:21AM BST. ((cite web)): Unknown parameter |timecaption= ignored (|time-caption= suggested) (help)CS1 maint: extra punctuation (link)

The concept behind the time parameter shows through; it sticks by the title where it belongs, since it is describing the time within the video. It does not stick by the publication date, which it has nothing to do with. Jc3s5h (talk) 01:03, 4 July 2013 (UTC)

• Put publication time in date parameters: With the common re-updating of news articles, then the time of news update could be inserted into the date, either prepended (14:30, July 3, 2013) or appended after the date (July 3, 2013, 14:30) or as "2:30 pm" format:

{cite web |title=Web Page |url=http://www.example.org |date=July 3, 2013, 14:30 | last1=Smyth| first1 = John |ref=harv}
Smyth, John (July 3, 2013, 14:30). "Web Page". ((cite web)): Check date values in: |date= (help); Invalid |ref=harv (help)

The year is time-extracted from the "date=" value, to set the Harvard referencing, so the time-of-day must be a valid time when using ref=harv. Then the year is extracted to set "<span id="CITEREFSmyth2013">". Otherwise, if the time-of-day contains an invalid hour, then the year will be omitted from the span-tag id (but no error message). -Wikid77 (talk) 11:00, 6 July 2013 (UTC)

How can the time zone be indicated without breaking year extraction for Harvard referencing? The possibility of adding a time should be documented for the date parameters for all types of works where adding a time might make sense. As far as I'm concerned, coders shouldn't bother adding features if there isn't a plan to document them so end users will know how to use them. Jc3s5h (talk) 14:05, 6 July 2013 (UTC)

Append time zone "CST" or "EDT" or such, where "3 May 92, 14:30 CST" still extracts the year as 1992. -Wikid77 11:21, 8 July 2013 (UTC)

If the date/time is a form recognised by ((#time:)) then it should work. For example, ((#time:Y|14:05, 6 July 2013 (UTC))) yields 2013 ... oh, hang on, that's the old way. Now that it uses a Lua module, I don't know - it's very difficult to trace the code through. --Redrose64 (talk) 14:30, 6 July 2013 (UTC)

• Lua uses pcall with lang.formatDate and 'Y': The line of Lua script, to extract the year, is:

good, result = pcall( lang.formatDate, lang, 'Y', str )

where the variable 'good' is a boolean set to true when the year has been extracted, into variable 'result'. A Lua function can return multiple variables, separated by commas (such as "good,result"). We could write a rapid Lua YearExtract function to ignore time validation, and just extract the year-number regardless of invalid hour, but the easy way was to use lang.formatDate to transform the date into just the year portion, inside the Lua Module:Citation/CS1. It's funny how {cite_web} will give error messages for obvious ISBN trouble, but an insideous mismatch of years in Harvard referencing will give no error message at all, just fail to link. I guess I should fix that to auto-correct and find the year else error message when ref=harv, since no one else has yet. We have Module:Citation/CS1/sandbox2 for debugging more complex problems. -Wikid77 11:21, 8 July 2013 (UTC)

I ran some experiments in my sandbox, and found the harv anchor isn't correct if the year is fewer than 3 digits, or if it has a BC after it. Regardless of whether we start putting times into dates, the range of valid years should be documented in the date and year parameters. As for time, I found that EST worked but a random three character group, "BQR", did not. If we are going to put times into dates, the documentation should indicate which time zone symbols are valid for the English Wikipedia. Better still, always use UT, since there will be no facility to convert times to the reader's timezone. Jc3s5h (talk) 12:35, 8 July 2013 (UTC)

Bad harv/sfn links - such as typos in years - may be detected quite easily, by installing this script. They then show up as big red error messages like "Harv error: link from #CITEREFDow1859 doesn't point to any citation", which is how I was alerted to make this fix. --Redrose64 (talk) 20:50, 8 July 2013 (UTC)

This edit request to Template:Cite journal has been answered. Set the |answered= or |ans= parameter to no to reactivate your request.

I understand this template is also used for newsletters but not everybody knows that. I would appreciate it if the word "newsletters" was added to the opening description as follows (bold for highlighting here only): "This Citation Style 1 template is used to create citations for articles in journals, magazines, newsletters and for academic papers." Helen (talk) 06:00, 9 July 2013 (UTC)

The documentation page isn't fully protected, so anyone could have implemented that change. Imzadi 1979 → 06:04, 9 July 2013 (UTC)

Although the wp:VisualEditor (VE) can edit more references than just the wp:CS1 cites, I have created an essay (wp:VECITE or wp:VEREF) which mentions using the CS1 cite templates within the VisualEditor:

• "WP:VisualEditor editing of references" - new essay mentions CS1 cite templates

Currently, the responses about using VE have been horrendous, with many experienced users vowing to no longer or never use it, but it can be a "teachable moment" (or "learnable nightmare") about the difficulty of using point-and-click interfaces to select each parameter from a list, rather than using the power of free-form text technology, followed by using a batch-mode "syntax checker" to proofread for invalid parameters in a macro scripting language (in this case, proofreading as red-error messages from the CS1 templates during edit-preview).

As for the fate of VE, with over 300 reported bugs, the onward mandate is pushed by the mindset that there is no "bad software" but that all software is inherently full of bugs and should be used whenever, regardless of the corruption of text files or psychological scarring of users. Hence, I have created that essay to help users cope with problems when being herded into a channel of VE-centric editing. Currently, all users have the option to use the prior edit-source mode for pages, but the VisualEditor provides a slippery slope of luring editors into a point-and-click interface for revising words, which leads to a tedious, slow, click-click-click-on-parameter mode to insert each separate parameter for citations. The new essay wp:VECITE is just a start to providing more help about that slow process. -Wikid77 (talk) 16:42, 9 July 2013 (UTC)

I think it would be easy to create a Lua-based {Cite_fast}, which would format cites even 4x times faster (over 300 per second) for large articles (lists) which have 400-900 cites. I can understand I might sound like the "mad scientist" in wanting even more speed, more speed. However, what if we discovered some new technique which could make the current cites run perhaps 2x faster, as a result of experimenting with even faster speeds. This is just a suggestion, because I think a Lua {cite_fast} could easily be 4x faster (or more). We could break the sound barrier, then the next step: cites on Mars! (just kidding). Things to ponder. Wikid77 (talk) 14:55, 27 June 2013 (UTC)

Faster is better. But instead of another fork, why not work on improving the CS1 module? What can be done to make it more efficient? --  Gadget850 ^talk 16:32, 27 June 2013 (UTC)

I'm talking about "thinking outside the box" to see what is possible, and then look back inside the box. The assumption of "fork" is the old-style thinking again. -Wikid77 (talk) 10:00, 28 June 2013 (UTC)

Right now a citation takes about 7.7 ms on average to render. Of that, about 2 ms is the parser overhead associated with entering Lua. In other words, we lose 2 ms every time Lua is used even if it does nothing. Given that limitation, it is already clear that no amount of clever Lua code is going to give you a 4x speed boost. You would have to improve Mediawiki itself to cut down on the Lua overhead first. In addition, about 3.5 ms is the average parser time spent converting the wikitext that Lua cites generate into HTML (e.g. [[George Washington|Washington, George]] (1776) ''[http://dreams.info/ My dreams]''. ⇒ Washington, George (1776) My dreams.). Again, that is time that Lua really has very little control over. The parser needs to handle most of that conversion in order to do things like determine whether links are blue or red and update the internal and external links tables. So, of the 7.7 ms spent on the typical Lua citation, about 5.5 ms is pre and post processing overhead that Lua has little control over. As complex as Module:Citation/CS1 might seem, it is only responsible for about 30% of the total run time right now. Maybe there are some clever tricks to make that faster (and I've certainly been paying attention to performance), but even if you could reduce the Lua execution time to near zero you would not even double the total rendering time. At this point one might well get more benefit from finding ways to make the associated Mediawiki parser behavior faster (not to mention that any improvements to the parser would have many advantages in other applications). For example, I think it may be possible to halve the overhead associated with entering Lua. Dragons flight (talk) 16:42, 27 June 2013 (UTC)

Excellent time to re-think the structure: The Lua developers have already quickened the interface, as over 625 #invoke's per second, or 500 per second with 7 parameters, so based on those advancements, then perhaps the new speed would be: 350 cites per second. I did not realize it would be so easy to achieve. -Wikid77 (talk) 10:00, 28 June 2013 (UTC)

Ummm, did you read what I said? I explained in some detail why it is essentially impossible at this point to get a 4x (or even 2x improvement) by making changes within Lua itself. I can't figure out how you could read what I said, and then come to the conclusion that large speed enhancements are easy. Dragons flight (talk) 16:20, 28 June 2013 (UTC)

Another way to see this. If you attempt to preview 310 distinct citations, it takes about 2.9 s to load the page (according to Mediawiki's "served by" clock); however, Mediawiki also reports that only 0.82 s was spent executing Lua code. In other words 70% of the execution time was spent outside of Lua doing things like converting wikicode into HTML, updating links tables, and everything else Mediawiki does during every page render. Of course suggestions for making the citation code faster would be appreciated, but the idea that any change to Lua might somehow go from 130 distinct citations per second to 350 distinct citations per second is simply not realistic. Dragons flight (talk) 16:50, 28 June 2013 (UTC)

As a footnote, let me add that it is important to test using distinct citations if you want to fully understand the real world case. The time required by the parser to resolve blue links and update links tables is dependent on the number of distinct links used on the page. If you simply repeat the same citation 300 times, then the parser doesn't have to work quite as hard and you get an unrealistic estimate of total time. Dragons flight (talk) 17:28, 28 June 2013 (UTC)

• Re-run and average 3 fastest runs then subtract overhead: Beware false premises in the testing. You have been calculating with the overhead of page processing, added as a part of citation speed, but it should be subtracted. Typical short citations can be formatted at 200 per second, with 200 different URL addresses in the linked titles. However, if the testing includes a wikilink to every author name, every publisher, and every location, as well as linking every title, then that is not a realistic test of 300 citations from a typical article or list of footnoted entries. To gain some broader perspectives, run timing tests of pages with no templates (or #invoke's) but hundreds of URLs or wikilinks, and determine the speed of formatting a page with 1,000 such links (to different pages and URLs), versus no links to those 1,000, and compare the runtimes to see the impact of links. -Wikid77 (talk) 21:01, 28 June 2013 (UTC)

I'm using the 310 citations that appeared in Barack Obama as of whenever I copied them. Hence it is a perfectly real world example (though skewed towards "news" citations). Also, you can use Special:ExpandTemplates to fully expand citations into wikicode, and then time the parser on rendering that wikicode. The parser processing of that citation wikicode actually takes longer than the Lua runtime at this point. Dragons flight (talk) 21:49, 28 June 2013 (UTC)

Since you want to argumentative about this. Here are 310 Lua citations [2]. The served by times for page preview in 15 trials are: 3.303, 3.448, 2.727, 2.840, 3.473, 2.953, 4.147, 2.832, 2.907, 2.874, 2.810, 3.550, 2.862, 2.998, 2.830. That gives a mean time of 3.1036 seconds and median time of 2.907 seconds.

Consider the same citations expanded to wikicode by Special:ExpandTemplates [3]. The served by times for page preview in 15 trials are: 1.308, 1.457, 1.129, 1.560, 1.261, 1.586, 1.624, 1.606, 1.302, 1.194, 1.582, 1.549, 1.198, 1.500, 1.170. That gives a mean time of 1.4017 seconds and median time of 1.457 seconds. This helps isolate the post-processing time that citations presently require.

Next, consider the same citation list, but with each Lua call replaced with a call to the Module:DoNothing via ((cite nothing)) which allows us to measure the citation template and Lua overhead [4]. The served by times for page preview in 15 trials are: 1.299, 1.634, 1.384, 1.231, 1.253, 1.157, 1.328, 1.365, 1.163, 1.332, 1.465, 1.401, 1.359, 1.433, 1.545. That gives a mean time of 1.357 seconds and median time of 1.359 seconds. This helps to isolate the overhead associated with Lua citation calls.

Lastly, consider a completely blank page, e.g. [5]. The served by times for page preview in 15 trials are: 0.304, 0.261, 0.289, 0.329, 0.334, 0.337, 0.328, 0.249, 0.264, 0.324, 0.327, 0.285, 0.343, 0.325, 0.313. That gives a mean time of 0.3075 seconds and median time of 0.324 seconds.

Using the median times, the global overhead of Mediawiki for every page preview is about 0.324 seconds. That means the Lua DoNothing overhead on 310 citation invokes is about 1.359 - 0.324 = 1.035 seconds (3.3 ms per citation). The post-processing render time of citation wikicode into HTML on these 310 citations is then 1.457 - 0.324 = 1.133 seconds (3.7 ms per citation). Then we can isolate the Lua citation code execution time as 2.907 - 0.324 - 1.035 - 1.133 = 0.415 seconds (1.3 ms per citation). In other words, of the 2.583 seconds spent rendering citations (= 2.907 - 0.324, 8.3 ms per citation), about 40% goes to the overhead of calling the citation templates and invoking Lua, another roughly 40% goes to parser post-processing of the resulting wikicode into HTML, and only about 20% goes into the actual execution time of the code within Module:Citation/CS1. As I suggested before, that creates a dynamic where most of the execution time is associated with things that Lua Module authors have no control over. I think that further improvements to Mediawiki may alleviate some of those bottlenecks, but no amount of clever Lua code is likely to provide a 2x gain in overall performance at this point. Dragons flight (talk) 23:10, 28 June 2013 (UTC)

'Run more tests without busy servers, then average the 3 fastest times: Do not use the median (middle runtime) of all tests. Subtracting the typical page-format overhead, of 0.324 seconds, from the average of fastest times, will reveal the speed. I think the pre/post Lua processing is delayed by the busy servers; however, even pure Lua execution has been slowed somewhat by busy servers as well. When tests are run on busy servers, then the runtimes treat the other people's server delays as being a major part of citation formatting, and there are some hours when the servers are very busy for hour after hour. Look for a fast runtime of 2.4s when the average tends to be 3.1s. -Wikid77 (talk) 20:39, 2 July 2013 (UTC)

While servers can be busy, it is better to measure real performance of a computing cluster rather than go hunting for the fastest performance. Also, Wikimedia is not a homogeneous server farm, some machines really are 30% faster than others. Regardless, you are free to run your own tests. You won't get a different conclusion. Dragons flight (talk) 20:52, 2 July 2013 (UTC)

I suspect that Cite could be improved. There have been a lot of complaints about the data array. --  Gadget850 ^talk 16:27, 30 June 2013 (UTC)

• Lua fast-cite would run 380/second for max 10,700 cites: The extensive timing tests, for more than 30 runs scattered over several hours, have confirmed the speed would be 2x-3x faster, as over 380 cites per second, or 1,000 cites in nearly 3 seconds. A large article (400 cites) would reformat 2-7 seconds faster. The max capacity would be 10,700 cites, due to the 10-second Lua timeout limit. Otherwise, the parser would allow 23,500 Lua cites per page, but the 60-second parser timeout also limits capacity, below 11,000 cites per page. -Wikid77 (talk) 21:33, 18 July 2013 (UTC)

Hi. I've found a lot of errors like ". Error: |episodelink= requires |number= when using ((cite episode)): Empty citation (help). " This comes up when the program doesn't work with episode numbers. Unlike "seasons" or "series" with distinct numbers of episodes (I think it's 20 something, but you'll have to correct me on that), we have a lot of programs over here which are ongoing - that is they have been weekly for years on end. I'm sure they do have a production number, but as far as I know the episode number is effectively the date. Is there any way to get around the error? Should the template require a number, when a link is present? Flying Buttress (talk) 14:48, 12 July 2013 (UTC)

It is confusing... in fact, it's only shown if all of |season= |seriesno= |number= are blank (or omitted). If there is no suitable value for |number=, do you have something suitable for either |season= or |seriesno=? --Redrose64 (talk) 15:18, 12 July 2013 (UTC)

I have also noticed that this error message is inconsistent with the documentation and possibly erroneous itself. For example, "Error: |episodelink= requires |number= when using ((Cite episode))" appears in Bradbury Fields. There are two problems with this:

1. The citation in question uses a url instead of a wikilink in the episodelink parameter. The current documentation says that episodelink takes a wikilink, so the error should be "Error: |episodelink= requires a wikilink".

2. The documentation for ((cite episode)) does not show number as a prerequisite for episodelink in the table. Jonesey95 (talk) 04:19, 17 July 2013 (UTC)

After further research, it appears that the error message listed above was added to the template's code on May 12, 2013 by User:MSGJ. I will drop that user a note. Jonesey95 (talk) 04:38, 17 July 2013 (UTC)

MSGJ did make that edit to the live template, but the coding was done by 117Avenue (talk · contribs). The background is at Help talk:Citation Style 1/Archive 2#Cite episode deprecated parameters. --Redrose64 (talk) 07:21, 17 July 2013 (UTC)

This is the Bradbury Fields citation:

((cite episode| title = Visually impaired people in TV ads, and charities working together | episodelink = http://www.bbc.co.uk/programmes/b01l0fkf | url = http://www.bbc.co.uk/programmes/b01l0fkf | series = | serieslink = | credits = | network = BBC | station = Radio 4 | city = | airdate = 24th July 2012))

Which gives this:

"Visually impaired people in TV ads, and charities working together". 24th July 2012. BBC. Radio 4. ((cite episode)): Check |episodelink= value (help); Check date values in: |airdate= (help); Cite has empty unknown parameters: |city= and |serieslink= (help); External link in |episodelink= (help); Missing or empty |series= (help); Unknown parameter |episodelink= ignored (|episode-link= suggested) (help)

This citation could and perhaps should be changed to use ((cite AV media)):

((cite AV media |title=Visually impaired people in TV ads, and charities working together |url=http://www.bbc.co.uk/programmes/b01l0fkf |publisher=[[BBC Radio 4]] |date=24 July 2012 |medium=Radio broadcast |work=[[In Touch (BBC Radio 4 programme)|In Touch]]|author=Miles, Ffion (Reporter) |author2=Kumutat, Lee (Producer)))

Which gives this:

Miles, Ffion (Reporter); Kumutat, Lee (Producer) (24 July 2012). Visually impaired people in TV ads, and charities working together. In Touch (Radio broadcast). BBC Radio 4.

—Trappist the monk (talk) 12:27, 17 July 2013 (UTC)

Yes, this particular citation could be changed, but that doesn't fix the problems with the cite episode template errors. And yes, I read through the long discussion that led to this change. I don't see where in the discussion the consensus was formed that |serieslink requires |number, which is also not documented in the template's documentation. Maybe I'm missing something. Jonesey95 (talk) 16:56, 17 July 2013 (UTC)

Well, that was the citation that you chose for an example ...

The "consensus" (if it can be called that) arose from two editors in favor of |serieslink= requiring |number=, one opposed (me), and from the broader community's general apathy.

—Trappist the monk (talk) 13:06, 18 July 2013 (UTC)

The documentation is correct, episodelink is an existing Wikipedia article. The Bradbury Fields example is an incorrect use of the template, and it is being placed into a maintenance category (Category:Articles with incorrect citation syntax), bringing attention to its incorrect use. A perquisite for episodelink, serieslink, and url was brought up in the discussion because we didn't want users to be providing a link that went unused. To activate these links provide the appropriate number, season, series, seriesno, title, or trans_title. If the episode doesn't have these, what is there to link? 117Avenue (talk) 03:00, 19 July 2013 (UTC)

For years we have lived with author initials being used to populate |first1= .. |firstn=, with resultant confusion. For many cited authors this has not been a large problem. If an editor abbreviates "John Paul" to "J.P.", "J. P." or "JP" no one is terrible surprised. But how to handle "Jean-Paul"? Any of the preceding might be used, plus "J-P", "J.-P.", "J.-P" or even "J. -P". Is it time to discuss supporting a distinct parameter |initial1= for such choices? Some article editors seem to prefer full names while others want the most succinct initials possible. Authority control might be used to limit variations, with improved wikidata functionality as a result. See [6] for more. LeadSongDog come howl! 17:49, 15 July 2013 (UTC)

Is it worthwhile to apply a band-aid, or is it more appropriate to provide a set of parameters that truly represents most of the naming practices out there? The first issue is the number of names, and when a name is considered a compound name. For example, for a while, a former first lady of the US used to consider her surname to be Rodham Clinton, which was one compound name. In Republic of China people often have a compound given name. In the English tradition people may have any number of middle names.

Then there is the question of name order. In regular prose, the surname may be first, or the given name may be first. In an alphabetical index, the surname may be first or the given name may be first (the latter is the case in Iceland, I understand). Just adding an initial parameter is an incomplete solution. Jc3s5h (talk) 18:24, 15 July 2013 (UTC)

Name ordering comes under the heading of collation, an extremely complex topic. I don't know how the use of initials enters here, as the terms by which ordering is done are generally not initialized. ~ J. Johnson (JJ) (talk) 22:56, 15 July 2013 (UTC)

If the source being cited does not provide a name, only an initial, there is no choice but to sort the bibliography according to the initial. Jc3s5h (talk) 23:06, 15 July 2013 (UTC)

I'm a little confused by what is being said here. We are discussing the use of initials for first names, right? And by that we mean, in accord with typical practice in English, the author's personal name. Also, we are assuming the primary key for sorting (collating) an author list is the surname ("last" name in Western practice unless inverted, but comes first in Chinese, Japanese, and Hungarian). So the only (?) ambiguity is where, for a given surname, two different "first" (personal) names start with the same letter. E.g., "Jones, Jean-Paul", and "Jones, Jim". Is that the essence of the alleged problem? ~ J. Johnson (JJ) (talk) 21:57, 18 July 2013 (UTC)

The (book) template automatically prepends 'OL' to the Open Library ID, presumably on the assumption that all Open Library IDs start with that. The problem is, they don't, and this prevents linking to those that have Open Library IDs of other formats. For example, the Open Library uses IDs that begin with 'ia:' to refer to works that originate with the Internet Archive. This results in bad links to the Open Library website for such a work. (One such ID is 'ia:publicrecords02conn', which should link to, e.g., http://openlibrary.org/works/ia:publicrecords02conn but instead links to http://openlibrary.org/OLia%3Apublicrecords02conn.) —Gordon P. Hemsley→✉ 13:37, 22 July 2013 (UTC)

The OL is checked to ensure it ends with A, M or W so you should be getting an error:

The public records of the colony of Connecticut from 1636-1776. OL /ia:publicrecords02conn. ((cite book)): Check |ol= value (help)

This error may still be hidden. We need to check to see if the OL begins with ia: and adjust the URL.

Is there an OL document for the various id numbers? --  Gadget850 ^talk 14:10, 22 July 2013 (UTC)

And ((OL)) needs to be updated as well. --  Gadget850 ^talk 15:22, 22 July 2013 (UTC)

Indeed, there is an error when attempting to use an ia: ID. I note that your example uses a preceding slash, which seems to fix the issue of a bad URL, if only in a hackish way. I also want to note that although these two linking methods (parameter and separate template) make a distinction between a "book" and a "work", OL seems to be smart enough to differentiate automatically; at least, it does when using an ia: ID. (I can't speak to "author".) —Gordon P. Hemsley→✉ 01:34, 24 July 2013 (UTC)

The slash is a typo. I notice that your original example of http://openlibrary.org/works/ia:publicrecords02conn is redriected to http://openlibrary.org/books/ia:publicrecords02conn. I have noted this issue at Module talk:Citation/CS1/Archive 7#Open Library: Internet Archive. --  Gadget850 ^talk 10:32, 24 July 2013 (UTC)

I attempted a bold fix to ((OL)), but evidently I got it wrong, as Gadget has reverted me. Guess my command of the template syntax still needs work.LeadSongDog come howl! 13:14, 24 July 2013 (UTC)

The documentation for Cite news fails to explain what it is for. Is it just for citing paper newspapers? Maybe news telecasts? How about Usenet newsgroups? Jc3s5h (talk) 23:14, 3 July 2013 (UTC)

The documentation lead states "news articles in print, video, audio or web". What is missing or vague?

And thanks for reminding me: I have been meaning to add a short version of the table at Help:Citation Style 1#General use to the documentation lead. --  Gadget850 ^talk 23:30, 3 July 2013 (UTC)

Sorry, I missed it. The huge lua notice and the table of contents box dwarfed it and I just didn't see the lead. Jc3s5h (talk) 00:30, 4 July 2013 (UTC)

I added a CS1 template list to ((cite book)). I think it works well with the TOC. I need to polish up the documentation for the Lua templates so we can ditch the notice. --  Gadget850 ^talk 02:21, 4 July 2013 (UTC)

• Shortened Lua-cite notice to 5 phrases: I moved the prior notice into a new essay ("WP:Lua-based cite templates") and shortened the new notice as just 5 phrases, with a link to that essay for more information. That avoids the clutter of the prior 4 paragraphs, and so people can easily see the wording, "news articles in print, video, audio or web". In general, WP has had too much wp:Rulespam, cluttering the top of pages, and recently, the rulespam for a View-source was condensed as a shorter blurb at the top of edit-protected pages, but later restored as a long tedious "shaggy dog story" as to why, oh why, has the page been protected, oh why oh why, yada yada yada, ad infinitum, ad nauseam, and more and more and then some. -Wikid77 (talk) 12:27, 8 July 2013 (UTC)

Back on the original point, it would certainly be good if a way could be found of making clearer to editors that "cite news" is to be used in preference to "cite web" for refs to news articles from bona fide news outlets, whether or not said articles are on line. At present a great many refs put "cite web" when it should be "cite news". (I think the Reflinks tool may be to blame for a fair bit of this.) (On the other hand, one also sometimes finds people using "cite news" when they shouldn't, especially in the case of press releases, which have their own "cite press release" template; the distinction matters because a press release is a piece of news produced by a non-news organisation about itself, and is not therefore objective "news" such as one would hope to find in a reliable news source.) -- Alarics (talk) 14:34, 8 July 2013 (UTC)

There is no semantic difference between the various templates, other than the unused class. The output should also be the same, except when a periodical is defined. The original purpose of the different templates was to provide parameters designed for each type of citation. That dichotomy is not as important now that parameters are pretty much available across the series. --  Gadget850 ^talk 14:53, 8 July 2013 (UTC)

Well, but the output isn't the same. For instance, with "cite news" the location appears in brackets, whereas with "cite web" it doesn't. -- Alarics (talk) 17:34, 8 July 2013 (UTC)

You are right on that one. --  Gadget850 ^talk 19:56, 8 July 2013 (UTC)

And ((Cite press release)) includes "(press release)" in the output. GoingBatty (talk) 03:53, 10 July 2013 (UTC)

Perhaps these templates should be merged, with a |type= parameter added? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:05, 27 July 2013 (UTC)

I propose to rename Cite album-notes to Cite album notes because I believe the hyphen is unnecessary and improper use of punctuation. Lachlan Foley (talk) 11:17, 15 July 2013 (UTC)

I'm about to propose changing it to ((Cite AV media notes)) to match ((Cite AV media)) and expand it to cover all audio and visual sources. --  Gadget850 ^talk 11:31, 15 July 2013 (UTC)

Good call, but I suggest making a redirect at the first proposed alternative name. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 06:55, 19 July 2013 (UTC)

Now at Wikipedia:Templates_for_discussion/Log/2013_July_18#Template:Cite_album-notes. --  Gadget850 ^talk 19:44, 20 July 2013 (UTC)

Now at ((Cite AV media notes)). Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 23:24, 27 July 2013 (UTC)

Could someone please change the use of "..." to “...” for article titles, chapters, etc. in the citation templates? Or is there a style rule requiring "..."? --bender235 (talk) 08:35, 28 July 2013 (UTC)

MOS:QUOTEMARKS

—Trappist the monk (talk) 09:42, 28 July 2013 (UTC)

For those who haven't time or inclination to plough through MOS:QUOTEMARKS: Curly quotes are deprecated. Always use straight quotes. -- Alarics (talk) 11:19, 28 July 2013 (UTC)

Correct. I do have a feature request to move styles to CSS, which would allow users to apply custom styling. --  Gadget850 ^talk 11:42, 28 July 2013 (UTC)

Thanks, I wasn't aware of that. Does anyone have an idea what's the rational behind this? Wikipedia:Manual of Style/Register#Quotation marks lists a boatload of discussions about this, but I'm unable to identify a valid argument against curly quotation marks in this case. --bender235 (talk) 17:27, 28 July 2013 (UTC)

There has been so much discussion on various talk pages- you will just have to do some searching. Regardless, we follow the MOS, so if the consensus were changed there we would follow. --  Gadget850 ^talk 17:52, 28 July 2013 (UTC)

Here's one I made earlier. That was my most recent decline; there have been others. --Redrose64 (talk) 19:41, 28 July 2013 (UTC)

I didn't mean we should ignore MOS. I just wondered whether there was a simple, obvious reason to prohibit curly quotations. --bender235 (talk) 20:08, 28 July 2013 (UTC)

One good reason is because they are "special characters" which don't work on all systems. Also they are a needless complication in life. Keep things simple. There may also be other reasons. -- Alarics (talk) 23:27, 28 July 2013 (UTC)

“ and ” are not valid characters in HTML documents unless they are encoded. See Character Entity Reference Chart at W3. - 79.67.242.207 (talk) 12:08, 30 July 2013 (UTC)

That's a list of possible encodings, it's not mandatory: it contains commonplace characters such as , . which are almost never encoded. Only three characters must always be encoded when they occur in the text part of a HTML doc: & < > - a fourth " should be encoded where its meaning might be ambiguous, such as within an attribute value. All the other encodings are provided for situations where the desired character might not be directly typeable, or where a program which processes the doc might trash a multibyte character. --Redrose64 (talk) 13:55, 30 July 2013 (UTC)

Is there a way to supress the period after the title in ((cite book))?--TonyTheTiger (T/C/BIO/WP:CHICAGO/WP:FOUR) 04:53, 21 July 2013 (UTC)

No. --  Gadget850 ^talk 11:01, 21 July 2013 (UTC)

Why do you want to? An example I could think of would be if the title already contains terminal punctuation such as "!" or "?". Jc3s5h (talk) 12:24, 21 July 2013 (UTC)

Those are probably more common reasons than mine, but I was just wondering if it was possible. I am misusing the template at Whaam! for short citations that result in double periods, but I think the functionality should be there for titles ending in another punctuation mark (most likely !, ? or …).--TonyTheTiger (T/C/BIO/WP:CHICAGO/WP:FOUR) 12:44, 21 July 2013 (UTC)

We can probably add terminal punctuation detection and suppress the period. These templates were never designed for use as shortened citations. We have ((sfn)) and related templates for that. --  Gadget850 ^talk 13:22, 21 July 2013 (UTC)

Wouldn't the programming be easier for a parameter like suppress=yes than a detection algorythm?--TonyTheTiger (T/C/BIO/WP:CHICAGO/WP:FOUR) 13:43, 21 July 2013 (UTC)

So this would only apply to 'title' and not 'trans_title', 'chapter', 'trans_chapter' or any other field? --  Gadget850 ^talk 14:10, 21 July 2013 (UTC)

I would think suppression of a period when other terminal punctuation is present would apply equally to all title-like parameters. On the other hand, if the option to use a comma as a separator is invoked, I would think the suppression should not occur. I would never expect a title of anything, whether an article, book, website, or whatever, to end in a comma. Jc3s5h (talk) 14:42, 21 July 2013 (UTC)

There are occasions when title, publisher, or other fields end with exclamation mark, question mark or other punctuation. It would be useful to suppress the addition of a period on those occasions. -- 212.139.97.146 (talk) 18:48, 21 July 2013 (UTC)

It seemed that there was agreement to do this. I see no changes.--TonyTheTiger (T/C/WP:FOUR/WP:CHICAGO/WP:WAWARD) 05:18, 3 August 2013 (UTC)

Where a reference is formatted using cite web or cite news and the source article is date stamped, that information is conveyed using the |date= parameter. If the source article is not date stamped, the |accessdate= parameter is used instead. Or, at least that's the theory.

A large number of references have both |date= and |accessdate= and in most, if not all, cases there is no need for |accessdate= to be there.

Should the unwanted |accessdate= inclusion be ...

• flagged as a problem using an error message shown in red text (I am fully aware that would flag about 90% of wikipedia pages as having this error), or
• the "Retrieved on [date]" information be surrounded with enboldened red parentheses as a more subtle warning, or
• the |accessdate= data simply be hidden and not displayed when the reference is shown,

or handled some other way?

Your thoughts? -- 86.148.10.139 (talk) 18:28, 20 July 2013 (UTC)

Just hide redundant access dates and add |showaccessdate=yes for those who have a burning desire. --  Gadget850 ^talk 19:50, 20 July 2013 (UTC)

To add: There are some editors who think that access dates aid in recovering from a dead link. Allowing it in the citation fields fulfills this, while suppressing it where there is a publication date fixes the issue of redundant dates. --  Gadget850 ^talk 20:10, 20 July 2013 (UTC)

Where both |date= and |accessdate= exist, I would use red strikeout on the |accessdate= retrieval date. -- 2.31.91.41 (talk) 12:18, 21 July 2013 (UTC)

We have been over all this many times before. It doesn't much matter whether accessdate is shown or not. What matters, for news citations, is that publication date *is* shown. The main error that occurs is when there is an access date but no publication date. In fact, publication date for a news citation is far more important than access/retrieval date. It is a great pity there seems no way of getting this across to those editors (which is the majority, unfortunately) who are not familiar with referencing. -- Alarics (talk) 19:59, 20 July 2013 (UTC)

• Publication date and accessdate are different and serve different purposes.  Even with a publication date, web publishers may post revision dates.  It is also possible for web publishers to make hidden changes in material with a publication date.  It is possible for unintended changes to be published.  In all cases, the accessdate should tell you the date when the material existed that was used on Wikipedia.  This is also why we need to routinely archive every web reference.  Unscintillating (talk) 21:22, 20 July 2013 (UTC)

A revision date is not an access date. If a web document is changed without updating the publication date, the access date tells you nothing. Then there are editors who verify the links on a page and update the access dates. And you can't archive every web page— many commercial sites are out of bounds.

• The premise of this discussion is that the accessdate is duplicative if there is a publication date.  In fact, it is never duplicative because unlike printed material, web pages are transient.  Examples of caches or partial caches are the quote parameter, Google caches, and archived pages.  Unscintillating (talk) 01:10, 21 July 2013 (UTC)
• Regarding the inability to sufficiently archive Wikipedia references, is there a discussion on Wikipedia about this?  Thanks, Unscintillating (talk) 01:10, 21 July 2013 (UTC)

This has been brought up before, but neither Wikipedia:Using WebCite nor Wikipedia:Using the Wayback Machine note the limitations. See http://archive.org/about/faqs.php#14. Many commercial sites such as the The New York Times use robots.txt to prevent offsite archiving. --  Gadget850 ^talk 01:53, 21 July 2013 (UTC)

All this is of quite minor importance: put access dates in if you want, but they are very rarely of any use. Publication date, on the other hand, for news citations (and press release citations) is absolutely essential. That's what we need to focus on getting across to people. -- Alarics (talk) 07:23, 21 July 2013 (UTC)

Articles from the New York Times with URLs of the form www.nytimes.com/yyyy/mm/dd/place-name-or-category/story-name-or-article-title can be archived just fine using the waybackmachine. Where did you hear they could not? I have found many articles from 1985 to 2005 that were archived for the very first time as recently as 2010 or 2011 as well as several articles from recent months that have been archived only in the last few weeks. -- 212.139.97.146 (talk) 18:37, 21 July 2013 (UTC)

• Partially answering my own question, see Wikipedia:Village pump (proposals)#Replacing WebCite citations with archive.is citations.  Unscintillating (talk) 23:04, 3 August 2013 (UTC)

Lets take a look at a citation I just cleaned up:

• From: "NOAC 1998 - A Life of Service". Oa-bsa.org. Retrieved 2012-08-02.
• To: "NOAC 1998 Memories: A Life of Service". Order of the Arrow. August 4, 1998.

The web page had a date at the bottom that was not recorded in the citation, but this was an easy fix.

So- what does the access date tell me? Lets say someone updated the web page in 2002 and did not change the date. How would I know? If I did know the page was changed, what do I do to indicate it using the access date? What if an editor comes along and cleans up the citation and changes it to the date he edited the citation?

I have updated Wikipedia:Using WebCite and Wikipedia:Using the Wayback Machine to indicate the limitations on archiving.

I don't expect much further from this discussion, but I might be surprised. I would hide access dates, but their existence often shows that the originating editor did not dig into the web page enough to find the proper date. Then there are those who add access dates for books and journals. --  Gadget850 ^talk 11:46, 21 July 2013 (UTC)

• In your example at Order of the Arrow, what you call the publication date is a revision date.  The copyright date is 1997, 1998, so part of the article is older than the event that is the main topic of the page.  Do you always use the revision date for the publication date?  As for removing the accessdate, how do you know that the article has not changed since 1998?  Unscintillating (talk) 02:14, 22 July 2013 (UTC)

I routinely delete |accessdate= parameters from web citations that have a proper date. To answer IP editor's original question, if we are to somehow flag citations that contain both |date= and |accessdate= parameters, it should be done in the same manner as existing error messages. All error messages and message style must be consistent. The error message More than one of |param1=, |param2=, and |param3= specified with a bit of editing at Help:CS1_errors to address redundant use of |date= and |accessdate= would seem to fit the bill. All of the date parameters |month=, |year=, and |date= need to be included when determining date / |accessdate= redundancy.

—Trappist the monk (talk) 12:45, 21 July 2013 (UTC)

In computer memory systems, data is transient.  You probably sign online agreements and never think about what happens if the web provider changes the agreement without notice to yourself.  Your legal position is stronger if you save a copy of the agreement at the time at which you signed it.  The publication date as reported by the web provider is of no personal interest, all you can represent is what was there when you signed it.  Unscintillating (talk) 13:12, 21 July 2013 (UTC)

Having both |date= and |accessdate= parameters may be redundant, but it's not an error according to the template documentation. Before adding another red error message visible to readers, I suggest that these steps be taken instead:

• Clean up the articles that have more important red error messages in the citations
• Gain consensus for the change of when to use |accessdate= via RfC
• Update the template documentation
• Submit a bot request to remove |accessdate= from citations where |date= is valid
• Set up a hidden tracking category
• Run the bot against the tracking category

GoingBatty (talk) 23:09, 21 July 2013 (UTC)

• Your premise here is that the two "may be redundant".  Do you have an example of such?  Unscintillating (talk) 02:14, 22 July 2013 (UTC)

@Unscintillating: - It's Trappist the monk premise that there is a "redundant use of |date= and |accessdate=". I was responding to Trappist's post, so I hope Trappist will provide you an example. GoingBatty (talk) 01:04, 23 July 2013 (UTC)

• Not redundant. Hi. I don't think it is redundant at all. Access date is excellent for recovering dead links because it tells me when the link was last active and possibly contained the info I was looking for. Mind you guys, we have several forms of dead: Dead with 404 error, dead with a redirect to irrelevant contents, dead because of being superseded or censored and dead by re-org. Access date helps. Very useful. Best regards, Codename Lisa (talk) 13:31, 3 August 2013 (UTC)
• Its benefit is grossly overstated. If you come across a dead link, you could try to find it again using one of the web crawlers. A page may be crawled many times in its history, so an access date could be used to find the closest version. Often the content is identical irrespective of the date, but when it isn't, it's still relatively simple to go through all the versions and find the one whose content is the closest match to the text to be cited. It also happens that, with very dynamic content – stuff that changes on a daily basis, the crawler doesn't happen to have a snapshot on that exact day, and you're still stuffed even if you have the accessdate. -- Ohc ^¡digame!_{¿que pasa?} 14:08, 3 August 2013 (UTC)

  In other words, you think it is easily replaceable with more efforts, time and sweat on editors' part. I agree with you on that; except your account of how easy to forgo its benefit is a huge understatement. I'd rather have it. Best regards, Codename Lisa (talk) 14:38, 3 August 2013 (UTC)

  What I meant was, most often it's actually not needed because an earlier or later archive version is identical. I've done it before, and often, so I know. It's often no help even if there is an access date. Some web pages just can't be found however complete the metadata is. The only way to make sure you do is to pre-emptively archive, which many editors are too lazy to do but I now do as a matter of course for certain sources. -- Ohc ^¡digame!_{¿que pasa?} 14:50, 3 August 2013 (UTC)

• It seems to me that accessdate is indeed redundant where archivedate is supplied (with appropriate archiveurl). As I've argued before, and along the same lines as Codename, is that it is not entirely redundant to date, because having the accessdate does make finding an appropriate archived version of the page more simple. --Izno (talk) 15:56, 3 August 2013 (UTC)

Hi. You have point there. I'd agree with that. Best regards, Codename Lisa (talk) 19:15, 3 August 2013 (UTC)

We will never get consensus on this. Need to create a perennial discussion page. Meanwhile, I simply hide access dates as useless. --  Gadget850 ^talk 15:58, 3 August 2013 (UTC)

Hi. I am not sure what you are saying. Are you saying you simply circumvent the DR? Policy-wise, I don't see if you can do anything if someone is intent on displaying it. Wikipedia removal policy is blacklist-based: There is a list of banned items; everything else is allowed. Outside the policy, if you forcefully hide the parameter, we will end up with articles that have access dates outside ((cite web)) tags. Best regards, Codename Lisa (talk) 19:15, 3 August 2013 (UTC)

Help:Citation_Style_1/accessdate (there's a link to it at ((cite web))). I venture to guess that Editor Gadget850 is referring to this. No need to tread the policy road.

—Trappist the monk (talk) 20:14, 3 August 2013 (UTC)

• Is this edit your definition of "hide"?  Do you mean that you "hide" the accessdate in the edit history, so that anyone that wants it must look through old versions of the article?  Unscintillating (talk) 20:15, 3 August 2013 (UTC)

Is it possible for the module to detect errors like this one, where multiple journal authors were specified with repeated "first=X last=Y" pairs? -- John of Reading (talk) 08:05, 6 August 2013 (UTC)

I don't think so. The module is handed a list of unique parameters. If there are any duplicates, the last encountered duplicate is the parameter handed to the module. This same issue attends all templates whether they use LUA or not. Compare the outputs of the ((citation/core)) and module versions of your citation:

((cite compare |mode=encyclopedia | last = Fiaud | first = J.C. | last = Kagan | first = H.B. | editor-last = Eliel | editor-first = E.L. | editor-last = Wilen | editor-first = S.H. | year = 1988 | title = Kinetic Resolution | encyclopedia = Topics in Stereochemistry | volume = 18 | publisher = John Wiley and Sons, Inc. | location = New York | pages =249–340))

—Trappist the monk (talk) 10:33, 6 August 2013 (UTC)

OK, if the duplicates are stripped out before LUA can see them there is nothing we can do. -- John of Reading (talk) 10:39, 6 August 2013 (UTC)

It's proposed to use a bot to clean cruft from Google Books URLs, in citation templates. Please comment at Wikipedia:Bot requests#Google Books links on how that should be done. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:00, 27 July 2013 (UTC)

There seems to already be a bot, or maybe it's part of reflinks, because I've seen these links get cleaned by some tool or other... --Lexein (talk) 02:00, 7 August 2013 (UTC)

Need to integrate the capability of {T} Wikipedia citation tool for Google Books to work with Reflinks. Or make a Reflinks-GoogleBooks so whole pages can be done with a tool. -- Moxy (talk) 02:41, 7 August 2013 (UTC)

CitationBot appears to be cleaning these up. --  Gadget850 ^talk 12:08, 7 August 2013 (UTC)

• Title. ((cite book)): Unknown parameter |subscription= ignored (|url-access= suggested) (help)
• Title. ((cite book)): Unknown parameter |registration= ignored (|url-access= suggested) (help)

So, the module includes the functionality of ((subscription required)), but not ((registration required)).

And why do we have the styling on this? --  Gadget850 ^talk 00:41, 12 August 2013 (UTC)

Added |registration= (in a crude sort of way – if only I knew what I were doing) to the sandbox. As you can see, I don't do any error checking and let |subscription= override |registration=. I think the functionality of |registration= matches that of ((registration required)).

—Trappist the monk (talk) 01:56, 12 August 2013 (UTC)

When using ((cite book)), using via= to indicate that the citation came from a Google Books copy, the ((subscription required)) is also dragged into it.

<ref name="Bug">((cite web|url=http://books.google.mu/books?id=SP7Sx5ps2F0C&pg=PA118&dq=muir+velvet+monkeywrench |title=Bug: The Strange Mutations of the World's Most Famous Automobile |first=Phil |last=Patton |publisher=Da Capo Press/Simon & Schuster |year=2002 |via=Google Books |pages=116-119 |isbn=0306813599 |accessdate=August 7, 2013))</ref>^[1]

This doesn't seem right, since many via sources do not require a subscription or other payment. Is this the correct place to bring this up? --Lexein (talk) 01:56, 7 August 2013 (UTC)

You don't need to mention Google Books in the citation. You're not citing Google Books, you're citing the original book and providing a link to an online version for the reader's convenience. So long as there's no reason to doubt that the scanned copy is identical to the original (which there isn't in the case of Google Books), it doesn't matter which version you got the information from. The only purpose of the |via= parameter is to indicate that the online version you're linking to is behind a paywall; usually, ((subscription required)) on its own is sufficient, but if you're linking to a well-known database like Highbeam or JStor, which lots of people have access to, some people prefer to note this in the citation, so that readers can tell at a glance whether they can access the source or not. DoctorKubla (talk) 11:30, 11 August 2013 (UTC)

This answer is incomplete: it ignores our citation policy of citing where you found it (the magazine/book/journal/newspaper, a digital repository of the item (HighBeam (pay), TheFreeLibrary.org (free), scan of an out-of-copyright item held at a University library website (free), or an individual's website (free)) of that archive, and it ignores e-books, out-of-print books where verifiability by others would be extremely difficult. So that restrictive view of via= for paywalls only is too restrictive, in my opinion. I've bolded the text above. I think either via= shouldn't drag in paywall, or that there should be another parameter called via_free= for non-payment vias. --Lexein (talk) 16:46, 11 August 2013 (UTC)

No, WP:SAYWHEREYOUGOTIT does not require you to say which service you got it from. If you read what you believe to be an authentic copy, then you need not specify which library you got the copy from, whether it was digital or hardcopy, etc. We used to explicitly state this, and perhaps we should do so again. WhatamIdoing (talk) 21:20, 14 August 2013 (UTC)

I propose that Template:Cite book be changed to instruct editors that, when citing a book where a place of publication is mentioned, the editor NOT add a wikilink around the name of the place. When a wikilink is added to the city of publication, that city then shows up in the "what links here" of many other articles which have nothing to do with that city. For example, let's say a book about "Vegemite" is published in "Greenville, Mississippi". Then an editor uses that book to update the article about Vegemite, and when citing the book, adds a wikilink around "Greenville, Mississippi" (as a convenience to readers who may not know where Greenville--the place of publication--is). Then, some other wiki user who is reading the Greenville article looks at the "what links here", and sees the Vegemite article (which has nothing to do with Greenville). This leads to unnecessary confusion. Thanks for your discussion about this. Richard Apple (talk) 06:02, 16 August 2013 (UTC)

Same is true for |work=, |publisher=, |authorlink=, |editorlink=, |agency=, |authorlink=, etc. This is a software problem, not a template problem. If this really is an issue, you should file for MediaWiki enhancement that distinguishes direct and transcluded links for whatlinkshere, though I don't know how likely this is to be implemented. —  HELLKNOWZ  ▎TALK 08:10, 16 August 2013 (UTC)

It's already in the documentation. See Template:Cite_book#Publisher.

—Trappist the monk (talk) 11:07, 16 August 2013 (UTC)

I'm trying to reference a featurette on a DVD properly, but can't see how I can do it using the fields on offer. The DVD title is The Pearl of Death, but there is no field in which that can comfortably fit while including the featurette name at the same time...

• Gitt, Robert (2003). Restoring Sherlock Holmes Featurette. MPI Media Group. ((cite AV media)): |access-date= requires |url= (help); Cite has empty unknown parameters: |1= and |2= (help)

Any thoughts? - SchroCat (talk) 21:17, 18 August 2013 (UTC)

Perhaps like this?

• Gitt, Robert (2003). "Restoring Sherlock Holmes Featurette". The Pearl of Death (DVD). MPI Media Group. ((cite AV media)): Cite has empty unknown parameter: |1= (help)

—Trappist the monk (talk) 22:14, 18 August 2013 (UTC)

That's great: many thanks. Perhaps the template documentation could mention that there's a chapter option, as the term doesn't appear anywhere on the page, and I would have thought it would be relatively important? Cheers - SchroCat (talk) 08:02, 20 August 2013 (UTC)

Done. --  Gadget850 ^talk 10:32, 20 August 2013 (UTC)

Great, cheers - SchroCat (talk) 10:37, 20 August 2013 (UTC)

when citing old, public-domain books republished by google books, what should the publisher and publication date be? e.g., philology currently cites the 1880 book Philology, by John Peile, published in 1880 by Macmillan, as ((cite book|url=http://books.google.com/?id=2joVAAAAYAAJ&dq=philology&printsec=frontcover#PPA5,M1 |title=Philology |publisher=Books.google.com |date=2008-02-09 |accessdate=2011-07-16)). it seems misleading to suggest that a book is less than 5 years old when it's really over 130, but OTOH it seems important to keep track the that info came from the scan, and not a physical copy. opinions? did i miss a settled answer to this? Adavies42 (talk) 23:31, 8 August 2013 (UTC)

If you use "Wikipedia citation tool for Google Books" it will find the correct date and publisher for you (and ISBN if it applies) - see example here using your book from above (note how the URL becomes a stable version.) Moxy (talk) 23:38, 8 August 2013 (UTC)

Concur. The cited work is the 1880 book regardless of whether it is the paper version or the digital version. I would cite it like this:

((cite book |last=Peile |first=John |url=http://books.google.com/books?id=2joVAAAAYAAJ&pg=PA3 |title=Philology |edition=4th |location=London |publisher=Macmillan and Co. |date=1880))

Peile, John (1880). Philology (4th ed.). London: Macmillan and Co.

—Trappist the monk (talk) 23:46, 8 August 2013 (UTC)

Except that it is considered good form to cite Google Books as the source using (I imagine) |via=Google Books. Except for the small problem that via= drags in (subscription required). Can I get a response to that question? --Lexein (talk) 09:58, 11 August 2013 (UTC)

Considered good form to cite Google Books as the source using ... |via=Google Books. Where is that stated?

Seems to me that Editor DoctorKubla has answered that question.

—Trappist the monk (talk) 11:53, 11 August 2013 (UTC)

I was invoking "Say where you found it" - a part of our citation policy/guidelines for many years, AFAIK, and a good part of it, in my opinion. --Lexein (talk) 17:01, 11 August 2013 (UTC)

The matter of whether to use the original publication date (and publisher) of a well-known work (e.g.: Darwin, Origin, 1859), or the date of the source actually consulted (e.g., the Norton edition of 1975, which does not readily indicate it is a facsimile of the 1859 edition) is not new to citation. The standard solution is simple: use both. Distinguish the original by putting it in brackets: "1975 [1849]". This seems relevant here. ~ J. Johnson (JJ) (talk) 21:16, 13 August 2013 (UTC)

That is the purpose of the 'origyear' parameter. --  Gadget850 ^talk 21:58, 13 August 2013 (UTC)

Yes — I had forgotten about that, thanks for mentioning it. That would seem useful in the example above. Perhaps the gist of the question turns on whether the Google product is a new edition? ~ J. Johnson (JJ) (talk) 23:12, 13 August 2013 (UTC)

No. It is a faithful copy rendered in an electronic medium presented by a host. --  Gadget850 ^talk 01:43, 14 August 2013 (UTC)

I am rather of the same opinion. But what of faithful copies rendered in an ink medium, such as my Norton edition of Origin? How do we explain to a new editor that one form is a distinct edition, and the other form is not? Or is it incorrect to consider ink facsimiles as distinct editions? ~ J. Johnson (JJ) (talk) 18:30, 15 August 2013 (UTC)

Doesn't this just show the point of the |via= parameter? Surely the distinction is a separate published edition versus an online scan of a previously published edition, i.e. Google did not create a new edition, they scanned somebody else's. A bot task to set |via= for Google books would be easy enough. It should also be easy enough to retrieve the actual publisher of the edition digitized from the Google books meta data. Rjwilmsi 19:01, 15 August 2013 (UTC)

@Editor J. Johnson (JJ) re: ink media editions. I'm inclined to think that a paper and ink version of a book not produced by its original publisher is a distinct work and should be cited as such. I think that the publishing industry would agree – if not, why issue different ISBNs for the same work from different publishers? When an author makes changes to a work then |edition= applies, regardless of publisher. If a publisher simply conveys the the right to print and distribute a work to another entity, either by sale or by license, then the work is new and distinct.

The work when it is available at google books or archive.org is a facsimile of a particular printed work:

Darwin, Charles (1859). On the Origin of Species by Means of Natural Selection (facsimile). London: John Murray.

Darwin, Charles (1909). Eliot, Charles W (ed.). The Origin of Species (facsimile). Harvard Classics. New York: P F Collier and Son. ((cite book)): Unknown parameter |editorlink= ignored (|editor-link= suggested) (help)

—Trappist the monk (talk) 20:27, 15 August 2013 (UTC)

1. If I scan a book, then print it, bound in a new cover, is that a distinct work? Does it make any difference if I print the scanned image provided by Google?
2. The first example misleadingly suggests that the 1859 John Murray edition is a facsimile. Surely what is meant is that editor consulted a facsimile of that edition?
3. The second example is clearer on this, but Darwin did not publish in 1909. Use of origdate here would clarify which edition of Darwin's is being replicated (faxed?). ~ J. Johnson (JJ) (talk) 21:03, 15 August 2013 (UTC)

1. Yes and yes. And your point is?

2. Understood. How about this:

Darwin, Charles (1859). On the Origin of Species by Means of Natural Selection (facsimile). London: John Murray. ((cite book)): External link in |type= (help)

Darwin, Charles (1909). Eliot, Charles W (ed.). The Origin of Species (facsimile). Harvard Classics. New York: P F Collier and Son. ((cite book)): External link in |type= (help); Unknown parameter |editorlink= ignored (|editor-link= suggested) (help)

3. No. The work being cited is the 1909 publication. The Citation Style 1 templates aren't here to provide a publication history of a work; they are here to support a fact stated in the Wikipedia article by clearly identifying what source the editor consulted for that fact.

And then there is this type:

Darwin, Charles (22 January 2013) [1859]. "On the Origin of Species". Project Gutenberg.

which is not a facsimile but where |origyear= is appropriate.

—Trappist the monk (talk) 22:18, 15 August 2013 (UTC)

Interesting. I see two different concepts here: I took "facsimile" as characterizing the work described (and you'll agree the Murray edition certainly is not a facsimile!), while you are taking it as "here is a link to a facsimile of the work". Perhaps a clarifying convention could be devised here, such as reduced type and an arrow: "→facsimile". ~ J. Johnson (JJ) (talk) 19:07, 17 August 2013 (UTC)

I moved your comment out of my post so that it doesn't cause confusion.

Since external links are provided with an external link icon, why the extra arrow? How does reduced font size make the facsilimle external link more understandable to the reader?

—Trappist the monk (talk) 01:13, 18 August 2013 (UTC)

The arrow explicitly suggests a link, though as you say this may be "extra" to the usual highlighting of a link. (I am ambivalent about the arrow.) The reduced font size, by making the term less obtrusive, reduces the association with the work itself than the link. Compare with the opposite effect produced by "... Origin of Species, FACSIMILE". ~ J. Johnson (JJ) (talk) 20:39, 18 August 2013 (UTC)

@Trappist, the use of distinct ISBNs often represent as small a difference as that between distribution territories in which otherwise-identical printings are sold, between pulp vs gloss paper, or between paperback vs boards. An ISBN is essentially just a stock control number for the book distribution business and by itself tells us little about whether the "realization" is of the same or distinct "editions". Small publishers have even been known to improperly re-use ISBNs for completely different works as a cost-cutting measure. Our interest should be in correctly identifying the work and edition cited, not the ISBN of its realization. LeadSongDog come howl! 03:33, 16 August 2013 (UTC)

I don't think that I've written anything contrary to what you just wrote. Have I?

—Trappist the monk (talk) 12:25, 16 August 2013 (UTC)

You wrote "I'm inclined to think that a paper and ink version of a book not produced by its original publisher is a distinct work and should be cited as such. I think that the publishing industry would agree – if not, why issue different ISBNs for the same work from different publishers?" My contention is that it is not a distinct work unless edited to become so. The 1909 work, for instance, includes an "Introductory Note" (by Elliot?) which makes it distinct from earlier texts. In other respects it is a facsimile, as annotated in its LoC cataloguing. LeadSongDog come howl! 16:22, 16 August 2013 (UTC)

The 1909 work is a reproduction but not a facsimile of 1859 – at least as I understand facsimile. If it were a facsimile then I expect the pagination, type-font, and even the title to be exactly the same. They are not. The words may be the same, but human hands have recomposed 1909 so that it is obviously similar but noticeably different. If an editor consults 1909 for an article then that is the work that should be cited, none other.

—Trappist the monk (talk) 00:35, 17 August 2013 (UTC)

(inserted out of time-sequence) I reflected the LoC usage of "facsimile" in the cataloguing data, though it would not be my preferred choice of terminology. Bartleby's equivalent record does not call it that. This is though, getting rather off-topic, so I won't belabour the point. If it really matters, we might want to "ask a librarian" at the LoC to clarify their meaning. LeadSongDog come howl! 17:05, 19 August 2013 (UTC)

On my point #3 ("Darwin did not publish in 1909") you say that the purpose is not to provide a publication history of a work, but to identify what source an editor consulted. I agree with this, but point out there are two levels of identification. First, there is which edition of Darwin's original text, and it makes a big difference between quoting (say) the first edition versus the fifth edition. Second is the actual representation (most likely a republication, unless one is a rare-book dealer) consulted. Whether I consult a facsimile edition (which I take to be a photographic image, which does not allow editorial improvements), or (say) the Modern Library edition (some correction of errors), is important to know, as in the event of discrepancies we want to check the exact text. But it is also important to know which original edition is used. And unless one is extensively familiar with the publishing history of different editions the dates 1909, 1936, 1975, 2001, etc. mean nothing. Therefore we need to cite both publication dates (such as "1909 [1859]"). ~ J. Johnson (JJ) (talk) 20:24, 17 August 2013 (UTC)

If you are going to consult both the 1909 and 1859 works, then it seems to me that your citation should be one or the other when both are in agreement. When they don't agree, then the article text can note the difference with citations to both works. As I've noted elsewhere in this discussion, 1859 and 1909 have different pagination, type-font, and titles. A single citation doesn't work well with those conditions.

—Trappist the monk (talk) 01:13, 18 August 2013 (UTC)

It is not a matter of consulting "both" editions (and the chance of actually consulting the original 1859 edition is quite unlikely, as there are so few copies). Even given a truly facsimile (intrinsically the same text, pagination, etc.), we do not (and should not) expect an editor that consults a derivative (i.e., corrected and modernized) edition to note the differences. Nor is there any reason why a single citation "doesn't work well", because all that is necessary is to add the date of the original edition. Look at it like this: all of the subsequent derivations of the first (1859) Murray edition of Origin constitute a branch of the original document, and all of the subsequent derivations of the fifth edition (1869) a different branch. When an editor cites some edition of Origin it is crucial to know not only the specific edition consulted, but also which branch (original edition) that edition derives from. ~ J. Johnson (JJ) (talk) 20:46, 18 August 2013 (UTC)

When I wrote that a single citation doesn't work well with those conditions, I meant that a single citation doesn't allow an editor to cite the same point of information from two sources. For example if an editor is citing Laws governing the Sterility of first Crosses and of Hybrids that citation, for 1859, is:

Darwin, Charles (1859). On the Origin of Species by Means of Natural Selection (facsimile). London: John Murray. p. 245. ((cite book)): |website= ignored (help); External link in |type= (help)

and for 1909:

Darwin, Charles (1909). Eliot, Charles W (ed.). The Origin of Species (facsimile). Harvard Classics. New York: P F Collier and Son. p. 305. ((cite book)): |website= ignored (help); External link in |type= (help); Unknown parameter |editorlink= ignored (|editor-link= suggested) (help)

These two citations cannot be combined into a single citation referencing both works.

Why is it crucial to know ... the specific edition consulted, [and] which branch (original edition) that edition derives from? (emphasis mine) Why is it not sufficient to simply cite the work that is at hand? An editor is not obliged to know and record the lineage of the source. Such a requirement would be a rather burdensome. This, it seems to me, follows rather naturally from your statement that we do not (and should not) expect an editor that consults a derivative ... edition to note the differences.

—Trappist the monk (talk) 14:44, 19 August 2013 (UTC)

You misunderstand. I am not suggesting that citations be combined, or that an individual citation should have to give the location of a fact (or quote) in multiple editions. The page number (and other details) should be of the edition the editor consulted. But if you were to (say) quote from your 1909 edition (which I do not have) the second sentence from the chapter on Hybridism as: "This view certainly seems at first probable...", and upon consulting my Modern Library edition I find that it reads "highly probable", I might wonder if you had left a word out. But if in your citation you had included the year of the original edition (i.e., the "branch") from which your version is taken (e.g.: "1909 [1959]") then I would instead consult my 1964 facsimile edition of the first edition, and all would be clear. When I don't have your particular edition at hand (and who could have them all?), knowing the original source of your edition is crucial to determining that you have not been careless.

Note this is not just being picky about direct quotes, but of such basic things as chapter numbering. I reiterate what I said before, that we don't expect editors to note all these differences. But provided they add the single clue of the original "branch". Determining this is generally not burdensome, and certainly less so than trying to verify something that cites an unavailable edition of an unknown branch.

  ~ J. Johnson (JJ) (talk) 21:48, 20 August 2013 (UTC)

So if I don't include the ... year of the original edition ... from which [my] version is taken (e.g.: "1909 [1959 [sic]]") you won't consult another source? That's nonsense. If you note a discrepancy, surely you will seek resolution regardless of |origyear= inclusion in my citation.

You're right, clearly there is something I don't understand about the importance of providing a publishing heritage in a citation. If the cited source supports the facts as stated in the Wikipedia article, that is sufficient.

—Trappist the monk (talk) 10:28, 22 August 2013 (UTC)

Bullshit. That's not what I am saying at all. I am saying that if you would be so kind as to indicate which original edition (such as Darwin himself wrote) from which your obscure and likely unavailable edition is based then we can avoid wasting time trying to find some edition that verifies your source. It is a courtesy to other editors and readers, and I don't understand why you are so adamant against that. You seem to think that would be a major piece of scholarship ("providing a publishing heritage"), but it is only a date, a mere six characters (counting the brackets). You seem to have forgotten that the essence of WP:V is not merely asserting that some evidence exists somewhere, but pointing to it so it can be found. That is the whole point of citation: to enable finding such evidence. ~ J. Johnson (JJ) (talk) 20:32, 22 August 2013 (UTC)

This has gone on far too long. Unwatching. Whatever conclusion is reached, I shall ignore it (mainly because my sources burn at 451°F). --Redrose64 (talk) 21:14, 22 August 2013 (UTC)

We might want to compare our approach to two printed style guides. First, here is APA style p. 203 example 19, "Electronic version of a print book" (first two entries) and example 21, "Electronic version of a republished book" (last entry):

Chicago Manual of Style p. 710–711 says to site the version that agrees with the page numbers in the citation. In the case of reprints and electronic versions, the date of the orignial may be included if relevant. A description of how a reprint differes from the original, such as the addition of new material, may be included in the citation if needed.

Chicago also says the citaion should indicate the citation writer consulted an electronic version if both a paper and electronic version are available, due to the potential for differences (p. 726). The citation if the citation writer colsulted the electronic version is shown first below; the second citation below is if the citation writer consulted the paper version (note the different publication date).

According to Chicago (p. 727) if it were relevant to include the original publication date, it would be done as follows (citing a different electronic version form above):

Jc3s5h (talk) 19:26, 18 August 2013 (UTC)

I would appreciate some guidance, and I would suggest it be added to the Template:Cite journal/doc page, on quarterly periodicals. If the periodical does not have a volume and number, should, for example, "Fall 2012" be in the issue parameter, the date parameter, or both? --JFH (talk) 18:35, 23 August 2013 (UTC)

|date=Fall 2013. If the citation will be referenced by ((sfn)) or one of the ((harv)) family of short form notation templates, include |year=2012 as well.

—Trappist the monk (talk) 19:05, 23 August 2013 (UTC)

Hmm, I had never thought of putting "Fall" (etc.) into |issue=. I kind of like it. But doing it as TM describes is quite adequate. ~ J. Johnson (JJ) (talk) 20:47, 23 August 2013 (UTC)

Using anything but a date in 'date' will break the anchor:

((cite book |title=Title |date=Fall 2013 |ref=harv))

--  Gadget850 ^talk 15:56, 24 August 2013 (UTC)

Which is why I remarked on the necessity to include |year=. This reference link^[1] is ((sfn||2013)). The full citation is listed below.

• ((cite book |title=Title |date=Fall 2013 |year=2013 |ref=harv)). Fall 2013. ((cite book)): Invalid |ref=harv (help)CS1 maint: date and year (link)

—Trappist the monk (talk) 16:23, 24 August 2013 (UTC)

Take a look at:

"Welcome Center Plan Worth Seeing". The Times Herald (Editorial). Port Huron, MI. August 19, 2013. Retrieved August 19, 2013.

for an example. The article is an editorial from the paper, so I want to attribute that fact using |type=. The paper's name doesn't include the city, so I want to include that in |location=. However, the type indication should really fall after the title of the article, and not in between the newspaper name and location. The result is that the location is disconnected from the title, and the type is disconnected from the headline. We also get the less than ideal line up of consecutive items listed in separate parentheses. (Note that if I used |format= to indicate it was an editorial, that would fall after the title, but that indicator is supposed to be used for the format of the resulting link, like a PDF or Excel spreadsheet file.) Imzadi 1979 → 22:43, 19 August 2013 (UTC)

'department': Regular department within the periodical. Displays after title and is in plain text.

"Welcome Center Plan Worth Seeing". Editorials. The Times Herald. Port Huron, MI. August 19, 2013. Retrieved August 19, 2013.

--  Gadget850 ^talk 00:44, 20 August 2013 (UTC)

Several Michigan papers use "Our Views" as the name of the "department" of the paper, but that doesn't necessarily make it explicit that the content is an editorial, I guess. Imzadi 1979 → 01:46, 20 August 2013 (UTC)

How about using "author= Editorial", like this:

Editorial (August 19, 2013). "Welcome Center Plan Worth Seeing". Our Views. The Times Herald. Port Huron, MI. Retrieved August 19, 2013.

~ J. Johnson (JJ) (talk) 17:55, 22 August 2013 (UTC)

I abhor putting false values into computer parameters to make the result look right. That practice always comes back to bite you in the ass when the metadata is used for a purpose other than just creating a pleasing appearance. If a piece of data is essential for a citation, and Citation Style 1 does not have a template parameter in which the information can be properly entered, there are two options. First, leave the information out of the template, but put it in plain text immediately after the template. Or second, don't use a template, even if the rest of the article uses templates. The absence of a suitable template for a particular source justifies writing the template in regular wiki source code; WP:CITEVAR was never intended to justify suppression of a source just because the source information doesn't fit neatly into an existing template. Jc3s5h (talk) 18:41, 22 August 2013 (UTC)

  Should we take that as a "no"? :-)

  I am in general agreement with you regarding misuse of parameters. (Though I disagree on mixing templated and non-templated usages. Surely your first option is sufficient?) But consider this: as "editorial" adequately (and pragmatically) indicates authorship — which is to say, a work of the editorial staff that represents the views of the publisher — how is it in any sense "false"? ~ J. Johnson (JJ) (talk) 20:02, 22 August 2013 (UTC)

If we think of the metadata being used in a different format than a citation, using "editorial" for the author's name produces strange results. It might be interesting to know how many times John le Carré or NASA were cited in Wikipedia, but what would it mean if "editorial" we knew how many times "editorial" was cited? Jc3s5h (talk) 18:51, 23 August 2013 (UTC)

"Editorial" would be meaningful in search of citations constrained to a single publication. But I get your point: we usually expect |author= to be filled with the name of a person or entity, not a description. Perhaps "|author=Editorial" should be understood as a special case that requires reference to "publisher" to be fully resolved. Or perhaps we just accept that the metadata isn't perfect, and we are not going to fret about. Actually, I think we're going about this sideways. We should first consider just how such a citation might be formatted, without considering how to do it in a template. ~ J. Johnson (JJ) (talk) 21:08, 23 August 2013 (UTC)

Using the author field to state Editorial is a bad idea, partly as many editorial pieces will have named authors.. If it is reasonable or necessary to distinguish between editorial and non-editorial articles without named authors, then the same arguments should apply to articles with named authors.Nigel Ish (talk) 22:46, 23 August 2013 (UTC)

I think it is more a matter of distinguishing editorials — signed or unsigned — from other content. (Which I think is reasonable.) But you raise an excellent point: if an editorial is signed then 'author' (but preferably 'first' and 'last') are needed for that, and "editorial" would have to go somewhere else. That's sounding like a good argument. Dang, you've about convinced me to go back revise some references. ~ J. Johnson (JJ) (talk) 20:59, 25 August 2013 (UTC)

There is not a satisfactory explanation for credits whatsoever. It only is aligned with aliases, which doesn't make any sense. This should be elaborated on and an example given. Since these are television episodes, I don't see how an alias will ever fit into the episode mold. MagnoliaSouth ^(talk) 19:16, 24 August 2013 (UTC)

I'm going to presume you are discussing ((cite episode)). In this context alias indicates duplicate parameters for the same variable. Thus, 'last' and 'last1' are simply different names for the same variable. When ((cite episode)) was originally created, all of the creators were stuffed into the one 'credits' field. When it was updated, the preference is to use the individual author parameters 'first' and 'last'. 'credits' was kept for backwards compatibility. --  Gadget850 ^talk 13:27, 25 August 2013 (UTC)

Most, if not all, of the other Citation Style 1 templates display the date in parentheses immediately after the authors' names (or if there is no author, after the title). But the ((cite web)) template places the date near the end of the citation. When was this change made and why? Jc3s5h (talk) 12:22, 22 May 2013 (UTC)

AFAIK ((cite book)), ((cite journal)) and ((cite web)) behave similarly: if there is an author, the date goes in parenthesis after that, otherwise it goes after the publisher. Please give examples of where you see different behaviour. --Redrose64 (talk) 15:18, 22 May 2013 (UTC)

The date formats in the citation templates are often inconsistent. With ((cite news)), if there's an author it's in parentheses, and if there's no author it's at the end. So the article ends up inconsistent, and the only current solution (that I know of) is to remove the templates, or not use them in the first place.

Can the templates be changed to allow editors to place the dates where they choose, so that they can be consistent within each article? SlimVirgin ^(talk) 17:06, 22 May 2013 (UTC)

The article where I observed the inconsistent behavior was International Atomic Time. This is the citation

• "Time". International Bureau of Weights and Measures. n.d. Retrieved 22 May 2013.

Apparently the inconsistency is caused not by the type of template (web), but because there is no author. I see no justification for changing the date location on the basis of having an author or not. Jc3s5h (talk) 17:26, 22 May 2013 (UTC)

Looking at the documentation for the |author= parameter, I find this paragraph:

The main function of |author= being used for organizational citation is when the cited source, such as a committee report, specifically names an official body or a sub-unit (of the publisher as the collective author of the work, e.g. |author=Commission on Headphone Safety or |author=Rules Sub-committee. Using this parameter to assert what you think was probably the collective author, when the source itself does not specify that body as the collective author, is original research and falsification of source verifiability and reliability.

So it appears that the Citation Style 1 requires that when the author and the publisher are the same, the author parameter is to contain a comment and not be displayed; some printed style manuals would do the opposite, put the publisher in the author position and leave the publisher position empty. Jc3s5h (talk) 17:37, 22 May 2013 (UTC)

If you don't know the date, simply omit |date= (or leave it blank). There is no need to set |date=n.d.. --Redrose64 (talk) 17:50, 22 May 2013 (UTC)

The way the documentation for ((sfn)) is written, it's hard to guess it would work if the date is left out, but it does, as long as ((Sfnref)) is also used. But in the case of articles that don't use the sfn family of templates, and just leave it to the reader to look up the full reference from the text in the footnote, n.d. is helpful in showing there is no date, as opposed to the editor not bothering to write the date. It also helps in distinguishing among several works by the same author or publisher, when some of them have dates and some don't. Also, the cite templates look a lot like APA style, and that style calls for the use of n.d.

From a technical point of view, it is not hard to change. One would need to be specific about what you want the change to look like (before and after examples are helpful). As others have noted the present behavior is to show a "(DATE)" after the authors / editors if any are given, and to use "DATE." much later in the citation if neither an author nor an editor is present. The citation templates have worked this way since at least 2007. I'm not sure the history / logic behind this choice of behavior, but given that it has been maintained this way for many years, it might be good to announce any proposed changes to this through the village pump (or other popular forum) to make sure this is something the community agrees with changing. Dragons flight (talk) 17:54, 22 May 2013 (UTC)

The easiest thing would be always to have it at the end, except for page numbers:

Smith, John. Book title, Name of Publisher, 2013, p. 1.

Smith, John. "Article title," Name of newspaper, 22 May 2013 (page number optional).

"Article title," BBC News, 22 May 2013 (page number not available).

SlimVirgin ^(talk) 18:11, 22 May 2013 (UTC)

Not all articles that use short footnotes use templates to link the short note to the full citation; instead it is up to the reader to look in the alphabetical reference list to find the source. Most commonly, works by the same author are distinguished by year of publication, so the short footnote might read "Miller 2005, p. 23." (Example taken from WP:SFN.) So the date should immediately follow the author name (or whatever is the first element of the displayed citaiton) so the reader can easily see if he has found the right citation in the list. Jc3s5h (talk) 18:19, 22 May 2013 (UTC)

Per the template documentation: "date: Full date of source being referenced in the same format as other publication dates in the citations. Do not wikilink. Displays after the authors and enclosed in parentheses. If there is no author, then displays after publisher"

In the example, 'date' is set to "n.d.", there is no 'publisher', which would go after 'title'. Thus, the placement is as described, and is the same for all templates.

As to why this date location is author dependent: The original creators designed it that way. This question has come up before, and I have looked at some really old archives without finding any discussion.

A significant portion of the discussion above is not related to the core issue, and diffuses the discussion. If there are other issues, please create a new discussion.

If the desire is to make the date location static, then make a proposal. --  Gadget850 ^talk 18:36, 22 May 2013 (UTC)

I am mainly aware of this problem in "cite news". Many newspaper reports have an author; many do not. In a typical WP article one finds both kinds cited under "cite news" and the final result looks inconsistent on the WP page. This has been raised many times in the past, including by me, but nobody ever does anything about it. I would suggest the date should come near the end of the reference, after the name of the newspaper (and city of publication, where given). -- Alarics (talk) 11:02, 23 May 2013 (UTC)

That would make it more or less consistent with the Chicago Manual of Style. I find this a strange choice on the part of Chicago; they advocate author-date in-text references, but put the date near the end of the bibliography entry, so when there are several works by the same author, the reader must look near the end of the entry to see if it's the right one. The APA Style put the two elements in the in-text reference, the author and date, right up front where the reader can find them. Both styles put the title first when there is no author, and both styles put a shortened version of the title in the in-text reference when there is no author.

MLA puts the author and the short title in the in-text reference, but that is seldom used on Wikipedia. Sensibly, MLA normally uses the title as the second element of their bibliography entries, so the two facts in the in-text reference are the first two facts in the bibliography entry. (I understand MLA did influence the original design of the templates.) Jc3s5h (talk) 11:42, 23 May 2013 (UTC)

So we have an undocumented 'via' parameter. What do we use it for? --  Gadget850 ^talk 14:10, 11 August 2013 (UTC)

• I think it should be used for a third-party deliverer of the content. In the case of books, this might be an online publisher, as discussed in above sections. It could be a mirror of a web page, a video of an event, a news article with online copy, a magazine scan, or a radio show transcription, etc. I think it is important (though not required) to specify the intermediary deliverer of the content, for verification and future editors, especially if the accessed content does not have the same medium or presentation. I also don't think this should be limited to subscription-based services, in fact we could use with a |subscription=yes or similar for such cases if one wants an inline tag. I agree with DoctorKubla that we don't need to mention this, as the source of the citation is what matters. But I think it makes sense to make it clear to readers and editors how the content was actually accessed (not limited to paywalls and such). —  HELLKNOWZ  ▎TALK 14:29, 11 August 2013 (UTC)
• How does this help identify the source? Does knowing that it is hosted at Google, Scribd, YouTube or my local library add anything? Is Google really a publisher? --  Gadget850 ^talk 14:49, 11 August 2013 (UTC)
  • No, it's not a publisher or related to the source in any way. And it doesn't have to identify the source. This is about the citation itself and how the material was accessed. In this case, specifically if the source was located or presented other than in its original form (so library doesn't count). What it adds is extra verification and access for future editors, I mean, why are we adding |url= to books in the first place? —  HELLKNOWZ  ▎TALK 15:51, 11 August 2013 (UTC)
    • Hellknowz gets it. Url= is a silent or at best implicit way of acknowledging the "where I found it", but urls can become 404s at any time. Via= is a way of making that path to the citation explicit (which is why I don't think via= should necessarily always drag in paywall). --Lexein (talk) 16:53, 11 August 2013 (UTC)
• Documentation:

via: Name of the content deliverer (not publisher) who presents the source in a format other than the original; for example: Google Books, Scribd, Project Gutenberg, YouTube.

--  Gadget850 ^talk 17:49, 11 August 2013 (UTC)

via(expanded): Name of the content deliverer (not publisher) who presents the source in a format other than the original; for example: Google Books, Scribd, Project Gutenberg, YouTube, HighBeam, TheFreeLibrary, EBSCO, etc.

• Support adding this documentation text, if Gadget850 is actually proposing it. It's currently not in ((cite)), ((cite book)), ((cite news)), ((cite journal)), ((cite web)), Help:Citation style 1, etc. --Lexein (talk) 20:08, 11 August 2013 (UTC)

Since it follows 'publisher', I'm tending to put the doc chunk in the Publisher section. --  Gadget850 ^talk 23:33, 11 August 2013 (UTC)

I've tweaked Module:Citation/CS1/sandbox so that |via= does not add (subscription required) to the citation. If this change is to be kept, then those citations that are out there in the wild that rely on |via= to provide the link note will need to be updated so that they use |subscription=yes.

—Trappist the monk (talk) 20:39, 11 August 2013 (UTC)

Good, but this was reported above. --  Gadget850 ^talk 23:38, 11 August 2013 (UTC)

That was in response to Editor Lexein's ... I don't think via= should necessarily always drag in paywall comment. If we are going to keep |via=, then it should be usable for citations that aren't hidden behind a pay wall as well as for those that are.

For me, the only reason to keep |via= is as a mechanism to show that the cited work is located in a place different from where the reader would expect to find it. As an example, a ((cite journal)) citation identifies an article by Caspar Milquetoast complete with |volume=, |number=, |doi=, etc. and a link to the article. When the link points to Dr Milquetoast's web page instead of the journal's web page there is a minor cognitive dissonance for the reader. If the citation also included |via=CasparMilquetoastCitizenScientist.com then that dissonance is mitigated. |website= won't work in this case because it is an alias of |journal=. When the article is available from the author's web site, (subscription required) is not appropriate.

—Trappist the monk (talk) 13:44, 12 August 2013 (UTC)

• Hold on. We seem to be making drastic changes to our citation policy based on a misunderstanding of WP:SAYWHEREYOUREADIT. This guideline doesn't demand that we note which content deliverer holds the specific copy of the source that we accessed. It simply says: "Don't cite a source unless you've seen it for yourself." If I've read Peile's Philology through Google Books, then I have read it – the medium is irrelevant. How do the words "via Google Books" make it easier for the reader to locate the source? If there's a link, they'll click it; if not, they can click the ISBN and find out for themselves that it's available through Google (and if there isn't a link, they'd have to do this anyway, regardless of whether the citation says "via Google Books"). See also WP:SOURCELINKS: "For a source available in hardcopy, microform, and/or online, omit, in most cases, which one you read." Granted, this guideline seems to be talking mainly about journal articles etc., but the principle is the same. DoctorKubla (talk) 06:28, 12 August 2013 (UTC)

1. I don't think we're changing policy, we're trying to implement it, and discussing that. |via= is two things. It's a tool which formalizes "say where you found it" at the editor's discretion within the syntax of CS1, without requiring external templates like ((HighBeam)). It is also a means by which citation requirements imposed by content providers other than the original publishers can be satisfied (such as WebCitation, HighBeam, EBSCO and others which all have terms of use). Parenthetically, for completeness, we could state via=Archive.org, instead of the current triggered text "from archive" (lowercase) whenever |archiveurl= is used. AFAICT "from archive" has become generic, due to being lowercase, with no other formal mechanism to indicate explicitly which archive provider has been used (I'd rather not use provider-dependent citation mechanisms or templates, just CS1).
2. I do understand the viewpoint that "only the original source matters" - but the extreme of that is to never include URLs, and never mention EBSCO or HighBeam, even though we never could have found the source without them. See WP:V - we do want our sources to be verifiable wherever possible, though some are difficult; this is not a game of hide-the-treasure. If we use a provider to source a claim, there's no good reason not to just mention the path via which we found it.
3. As an example of why |via= is useful: at The Bridge (2006 drama film) User:Cirt found a lot of sources to support keeping the article, while it was at AfD. He apparently used a research provider, but didn't include URLs or "by" or "via" comments. So it was a hard slog for me to verify the sources he dug up. Why did I bother? Because I wanted our readers to be able to more easily verify the sources for claims, and because that was a Scientology-related, and therefore potentially highly controversial article, subject to an extreme level of scrutiny due to ARBCOM decisions. I'm a big fan of easy verifiability, wherever possible.
4. Here's a specific use case for |via= and not |url=. The research provider EBSCO's URLs don't link purely to EBSCO domain+assets, they link through libraries/institutions, and so are accessible only to "local" patrons/customers. So I say |via=EBSCO, so the reader knows they can verify the source via EBSCO just like I did, but I leave out the |url= because the url wouldn't work unless they are patrons of the same library, with the same account number as me, and there's no way to generalize the URL.
5. I'm finding a lot of my work lately involves verifying that a source really stated what is written in a Wikipedia article. Bare URLs now dead, partial citations missing title or publisher, paraphrased titles in citations, and no |via= all combine to make WP less verifiable, and, personally, annoying. I guess I'm just saying every little bit helps.

Sorry for the length, --Lexein (talk) 19:12, 12 August 2013 (UTC)

Uh, hello? I did not intend to stop discussion. --Lexein (talk) 06:20, 14 August 2013 (UTC)

I'm very neutral about all of this, however, if we're going to so this, the via text should be inserted into square brackets after the publisher, not appended to the end of the citation, disconnected by intervening page numbers or ISBNs/OCLC numbers. Take the example given above with a page number provided:

Essentially, whichever entry listed in the |via= parameter is a republisher, and it shouldn't be disconnected from the original publisher. As for the "subscription required" or "registration required" indications, those are good at the end of the citation. Imzadi 1979 → 23:07, 14 August 2013 (UTC)

Perhaps some clarity of terms is lacking. An "edition" is distinguished from another by the fact of having had editorial input. A "printing", "facsimile", "reprint" or "reissue" without editorial input constitutes neither a new edition nor a "republication". The real utility that I see for |via= comes into play when there is reasonable doubt either as to the fidelity of the copy or the specific edition upon which a (possibly incomplete) copy was based. LeadSongDog come howl! 19:40, 15 August 2013 (UTC)

Another use for "via" would be if the URL of the website does not make it apparent what organization is hosting the website. If the link goes dead, knowing the name of the hosting organization provides a clue about where to look for a replacement hosting URL. Jc3s5h (talk) 19:49, 15 August 2013 (UTC)

Yes, this is more to the point than mypoint #4 above. --Lexein (talk) 18:19, 16 August 2013 (UTC)

@LeadSongDog: True! Thanks for reminding me, I have a real life example of this: HighBeam has a full copy of a newspaper article, while EBSCO's copy (attributed to HighBeam!) is missing the last two sentences, which were uncontroversial, and had no obvious reason to be omitted. I reported it to EBSCO, and linked and attributed HighBeam. --Lexein (talk) 18:19, 16 August 2013 (UTC)

Resolution?

I have added documentation for |via= under Help:Citation Style 1#Work and publisher for all to review. Hopefully it is not seen as any sort of policy change, merely documenting an important two-function parameter, and its justified uses: a) saywhereyoufoundit and/or b) politely attributing the content deliverer, as may be requested/required by their terms of use. So I would like sandbox to be moved live, so the |subscription= parameter can be activated, and then documented. --Lexein (talk) 16:27, 22 August 2013 (UTC)

I agree with the idea of using it to convey any information that the website requires to be conveyed as a condition of linking, if any such requirements are valid. No one has explained to my satisfaction how a web site can impose linking restrictions, so I make no statement about whether such purported terms of use are actually binding upon Wikipedia.

Lexein's edit has made the decision that Google Books and Project Gutenberg and the like are not publishers. I think the excerpts from printed style manuals indicate otherwise. I also think no consensus has formed in the discussion to define, for purposes of Citation Style 1, Google Books and Project Gutenberg and the like as non-publishers.

Even if we decided to classify information providers who provide a copy of another work as non-publishers, what is the threshold for changes before we consider them to be publishers? Project Gutenberg re-typesets the work in a format that does not have page numbers (at least in the case of The Adventures of Sherlock Holmes. Doesn't that make them a publisher? Jc3s5h (talk) 16:44, 22 August 2013 (UTC)

Huh. That's a rather harsh, non-constructive screed after days of saying nothing, and doing nothing. You seemed interested in having |via= documented. I'm not talking about "linking restrictions", I'm talking about citation attribution such as presented in examples by HighBeam and EBSCO and others. I have further not "made a decision", I'm documenting a feature, and providing use cases where it makes sense. Everybody stopped talking, so I just f'ing documented the damned thing. I fully know that Google Books both simply delivers (where there is a prior published/printed work) and also publishes (where there is no prior published/printed work). |via= is about those deliverers who are not acting as publishers. Jeez. There's only one publisher field. Think it through. --Lexein (talk) 23:08, 22 August 2013 (UTC)

Thanks for creating some documentation, and modifying it so as not to imply a consensus on points that have not been settled in this discussion. I do not take it as a given that the current parameters provide a way to describe, for example, a Google e-book that combines a facsimile with a text based on optical character recognition, in the way we would prefer. I do not take the attitude that the templates are wonderful and can do anything, and we should infer the meaning of the parameters from that point of view. On the contrary, I think we should decide what should be in the citation and how it should look, and if need be, modify the templates accordingly. If we want to cite the google e-book as a recent publication by the publisher, Google, and note the original publisher when relevant, we should find and/or create parameters to accommodate that. Jc3s5h (talk) 23:30, 22 August 2013 (UTC)

My main focus is improving verifiability, and not hiding the treasure. Here are my real-world use cases so far: If I found it using HighBeam and couldn't find it anywhere else, I'm indicating via=HighBeam and having done with it. If it's a citation of Ebony, and I found it using Google Books (real example, by the way, they seem to have many volumes of the magazine), I'm putting in the via. If it's an ebook sold by Google Books, they're getting the publisher credit, and not via. EBSCO gets a via, mainly because EBSCO URLs are untransferable, and therefore useless. With my most common use cases crying out for via and subscription=yes/no, yes I'm advocating publish-the-docs, and unsandbox the fixed code. --Lexein (talk) 23:47, 22 August 2013 (UTC)

From the current |via= documentation at Help:Citation Style 1#Work and publisher:

when the deliverer requests attribution (HighBeam, EBSCO, etc). ... |via= also satisfies citation attribution requests by content providers (such as HighBeam, EBSCO and others which have terms of use).

Citations needed. Where do these archivers require or request attributions from linkers (Wikipedia, in this case)? Specific urls please. A quick scan of the WebCitation terms of use^[1] didn't seem to me to require or request attribution from third party linkers like Wikipedia. Similarly, neither EBSCOhost^[2] nor HighBeam Research seem to have these requirements, though HighBeam does state that all links to the 'service' must be to the home page.^[3] If that HighBeam requirement applies to Wikipedia, then there are a lot of citations in article space that are not in compliance.

—Trappist the monk (talk) 12:33, 24 August 2013 (UTC)

One possible example is Grace's Guide which hosts a large scanned archive of The Engineer - "You may copy and use any of the content of this site provided you make a clear link on your web site or printed matter to Grace's Guide as the source of that information." In cases like that, it would at least be polite to provide some sort of attribution.Nigel Ish (talk) 15:58, 24 August 2013 (UTC)

Perhaps. But, "Copy and use" is different from "link to", isn't it? If I add a fact to an article that is supported by a page from The Engineer delivered by Grace's Guide, I cite it:

((cite web)): Empty citation (help)

I haven't "copied and used" anything; |via= simply identifies the provider – more for the reader than as a response to a perceived request or requirement imposed by Grace's Guide.

—Trappist the monk (talk) 17:54, 24 August 2013 (UTC)

After some time away from this topic, I see where I made an assumption. I assumed (and still believe) that the WP:The Wikipedia Library agreements User:Ocaasi reached with providers like Credo, HighBeam, Questia, JSTOR, and Cochrane came with a string attached: to identify the provider in citations. I got this impression from every single one of the example citations for Credo, HighBeam, Questia which always identify the provider. Since Ocaasi and User:Steven Walling have both been involved with those agreements, I'll ask them about this detail. I have altered the via documentation to reduce emphasis on attribution. --Lexein (talk) 14:27, 30 August 2013 (UTC)

Hey folks, I appreciate all of the careful discussion here. My original thinking with citations from donated accounts was that WP:SAYWHEREYOUGOTIT applied for identifying the actual site where research was conducted and the version which was actually read. That guideline may leave us in a bit of a gray area as it excludes search engines and library catalogues which led to research... I'm not exactly sure what its intended status on archives/databases that host the research is though. So, Linking to the HighBeam article, for example, was expected if an editor read the article there; actually attributing HighBeam in the reference was not explicitly required--though the citation example did encouraged it. I do think there's also a decent argument that subscription required, is more useful if you know subscription to what. For example, if there's a HighBeam citation on a page, I'd think editors/readers would benefit from knowing if that's a source they have access to. And you know thereby that you're reading the same version the editor was. This is in the context of always providing the original citation information so anyone can look it up in whatever location they have easiest access to. There may, realistically, also be benefit in attracting future donations of this kind if the via parameter is used, but I wouldn't put that motive first in our considerations. In short, there is no contractual agreement with the donors to attribute the source (there's no contract of any kind in fact), but there were usage expectations for editors for linking to the source and citation examples that encouraged giving in-reference attribution. Happy to discuss all this... Ocaasi ^{t | c} 15:11, 30 August 2013 (UTC)

Thanks for clarifying. I happen to like attributing the provider, others may not like doing it. So perhaps this belongs in the "leave it as the citing editor did it" class of things to generally be left alone, and not enforced either way. --Lexein (talk) 13:51, 2 September 2013 (UTC)

The documentation for the publisher parameter claims "Most professional and academic citation standards (and thus everyone familiar with any of them) do expect the publisher to be explicitly included, even where this may seem redundant. Adding it doesn't hurt anything, and eliminates the possibility that later editors will assume it was left out by mistake and waste time looking up the missing information."

That statement is true for books, but false for periodicals, especially academic journals. Jc3s5h (talk) 16:54, 22 August 2013 (UTC)

Hello. I do not endorse WP:WEASEL or WP:OR about claims but that also applies to you. That statement goes, yes, and thanks for notifying us; but it is not replaced by a rule that you prefer to enforce.

Previously, editors have talked and talked about including or excluding |publisher= but there has never been a clear consensus. In the last discussion, vote count was in favor of not excluding it in some cases but the reason was more or less "I just don't like it" or "others do it", which is responded by "Wikipedia is full of things you might not like that are going to stay put" and "we do a lot of things others don't do".

Best regards,
Codename Lisa (talk) 07:11, 24 August 2013 (UTC)

Someone who favours universally inserting publishers even when there is no manifest (only claimed) benefit should be more circumspect when making such a claim. I'm sure jc3 will happily demonstrate what he stated. -- Ohc ^¡digame!_{¿que pasa?} 07:26, 24 August 2013 (UTC)

It is certainly true, in my experience (primarily involving computer science and mathematics), that publishers of journals are typically not included in academic citations, exactly as Jc3s5h states. I don't see why you (Lisa) object to removing a falsehood from the documentation; it is not the same thing as stating a consensus for how we do things here. —David Eppstein (talk) 07:32, 24 August 2013 (UTC)

And when did I object to removal of falsehood? (In fact I said "That statement goes, yes, and thanks for notifying us" and my edit did not restore it.) I object to addition of rules not supported by consensus. If you wish to merely add a fact about what others do, add it in the article about the that citation. If on the other hand, implies we should do the same; I afraid that's when I objection. Best regards, Codename Lisa (talk) 08:00, 24 August 2013 (UTC)

i'm glad to read your view. to me, it's quite objectionable that Codename Lisa has stated her view without substantiating, and launched what is tantamount to a personal attack, alleging "I just don't like it". -- Ohc ^¡digame!_{¿que pasa?} 07:42, 24 August 2013 (UTC)

I do not understand why you would like to antagonize me, but no personal attack was intended. People value their own opinion; that's all. Only that value does not executive force in Wikipedia. Best regards, Codename Lisa (talk) 08:00, 24 August 2013 (UTC)

I'm not deliberately antagonising Codename Lisa, but it seems that she is taking this personally. JC3 made an edit that I don't necessarily agree or disagree with, but seems anecdotally supported. Codename Lisa's reverts seem quite unreasonable and doctrinaire. My attempt at reinserting this was summarily reverted and seen as "antagonism". Humph! — Preceding unsigned comment added by Ohconfucius (talk • contribs)

Oh, My, God! You are talking in third person and have forgotten to sign you comment. Put point-blank revert without an edit summary (treating me like a vandal) and ignoring WP:BRD together... I have not seen so much aggression and hate since I joined Wikipedia. I swear I did not murder your entire family.

Officially panicking,
Codename Lisa (talk) 08:33, 24 August 2013 (UTC)

APologies for omitting my signature. I have an unstable internet connection, and it's causing much discomfort in these parts. -- Ohc ^¡digame!_{¿que pasa?} 14:52, 24 August 2013 (UTC)

Codename Lisa, please don't feed the troll. And don't worry: He who treats disagreeing parties like vandals only makes himself look stupid. Now, although someone else has broken WP:BRD and some people here have stepped over the line (and got so far from it that they no longer see the line) your extra revert made out of panic has turned the situation into "specific text Lisa keeps removing and Jc3 keeps restoring" instead of "specific text that Jc3 and Ohconfucius keep restoring". Please be aware that the piece of text that Jc3 added makes no mandate for your or anyone else. It is not even an imperative sentence and even if it was, it would have created a mandate on books only. You are still at full discretion to add publisher whenever you wish. In fact, it is a matter of optional style and if you write or develop an article that uniformly uses publisher, per ArbCom discretionary sanction, they cannot remove the publisher.

And if you feel like murdering someone's entire family, I know a couple of good video games... 08:06, 25 August 2013 (UTC) — Preceding unsigned comment added by 188.245.108.156 (talk)

• Oppose Jc3's recent edit which deprecates the use of |publisher=. I think, and the prior consensus-based text supported, by implication, that adding |publisher= improves verifiability, especially for some use cases: lesser-known, or similarly-named publications. I'm a big, big, fan of improved verifiability, and not hiding the treasure, as I've mentioned elsewhere. The common sense of the editor creating the citation should be engaged, not discouraged. --Lexein (talk) 12:46, 24 August 2013 (UTC)
  • Jc3's edit doesn't deprecate anything, it just reports an observation of common practice. It just states exactly how certain publications don't cite the access date. If you can find advice to the contrary for those scientific citations with an access date publisher, please cite. -- Ohc ^¡digame!_{¿que pasa?} 14:52, 24 August 2013 (UTC)

Wait a minute, how did this conversation shift from |publisher= to |accessdate=?

I stand corrected. my bad. -- Ohc ^¡digame!_{¿que pasa?} 06:03, 25 August 2013 (UTC)

—Trappist the monk (talk) 15:46, 24 August 2013 (UTC)

I believe my edit is very much in line with WP:CITE#What information to include . The idea of treating Citation Style 1 as a distinct citation style, rather than a template to be used in any old way, is a rather new concept, and the documentation here is really not up to the job of specifying a citation style. I am not aware of any discussion supporting a departure from the universal academic practice of not citing publishers of journals and a number of other kinds of works. Certainly there is no navigation aid in the project page to find such a discussion. (Of course, additional information can always be added to any citation when normal citation practices will result in ambiguity.)

As for citations to support the ommission of the the publisher for many kinds of works, see the following:

• Sec. 5.4, "Citing periodical print publications" in MLA Handbook for Writers of Research Papers, 7th ed. (New York: Modern Language Association, 2009), p. 138.

• "Periodicals" in Chicago Manual of Style 16th ed. (University of Chicago, 2010), in section 14.170, indicates "the word periodical is used here to indicate scholarly and professional journals, popular magazines, and newspapers." Section 14.171 listes the items to include in citations of periodicals; publisher is not among the items, even for electronic periodicals.

The APA style manual has equivalent advice, but I don't have time to find the page just now. Jc3s5h (talk) 13:13, 24 August 2013 (UTC)

If we cut through the smoke, I think the core issue is whether the help page should state that, in regards of journals, "publisher" should, or should, not be included. Perhaps we are also in accord that, in fact, "academic" (scholarly, scientific, etc.) citations do omit the publisher of journals. (Right?) So I ask: is the problem that we do not have consensus here on whether to follow that practice? Or is the problem that the specific text Lisa keeps removing and Jc3 keeps restoring cites only WP:Citing sources, without referencing any notable MOS? ~ J. Johnson (JJ) (talk) 21:40, 24 August 2013 (UTC)

Looking at the history of the help page, I see the novel rule about always including the publisher was made in a series of edits by User:SMcCandlish on 16 July 2011. The talk page associated with the help page was not created until two days later. When the talk page was created, SMcCandlish's addition was not discussed. But I don't believe Citation Style 1 was viewed as a style in its own right at that time, so no one would have looked at the help page for advise on what to include and what not to include in a citation; just on what parameter to use if you decided to include a bit of information. Decisions on what to include in a citation would be made based on WP:CITE or normal academic practice.

The only expression of Wikipedia consensus I can find on when to include the publisher in a cite is on WP:CITE. I don't include this page because no one would have thought to look here to form such a consensus. The printed style guides agree with WP:CITE. Can someone point to a consensus, formed after a properly-advertised discussion, that Citation Style 1 should depart from WP:CITE and printed style guides, and require publishers be cited for periodicals?

Another approach to judging consensus is to see what is done in the better articles. Maybe we should look at some featured articles that use Citation Style 1 and see whether they cite publishers for periodicals. Jc3s5h (talk) 23:53, 24 August 2013 (UTC)

Good info. Thanks for digging it out. ~ J. Johnson (JJ) (talk) 20:36, 25 August 2013 (UTC)

Here's my opinion on the matter: most of the time, I've followed the usual academic writing practices from my years in high school and college, which is to omit the publisher on most periodicals. However, there are cases where including it for lesser-known publications is beneficial. Each editor in each situation is going to need to make a value judgement on whether or not including a publisher is necessary in those cases. I think this is a case where the documentation is going to need to describe the established practices already followed, rather than establishing the practices to follow. Imzadi 1979 → 00:21, 25 August 2013 (UTC)

I agree with Imzadi1979 but would say that for lesser-known publications, ISSN should be preferred, falling back to publisher if that is not available. In general an abstracting identifier like PMID, MR, JSTOR, ... should be preferred, maybe DOI too. RDBrown (talk) 01:01, 25 August 2013 (UTC)

Actually, I'm of the opinion that we should include an ISBN, ISSN, OCLC number, DOI or whatever ID number we can whenever possible. When citing a journal or newspaper article, I always attempt to include either the ISSN or OCLC number, and then I look for a DOI as well. This is regardless of any publisher I might indicate for lesser-known works, and I include these even with well-known newspapers like The New York Times. Imzadi 1979 → 11:54, 25 August 2013 (UTC)

I concur with all of this. In general we should (and do) include all useful bibliographic detail, including ISBN, ISSN, DOI, etc. However, in regard of journals and other periodical "publisher" is generally not useful, and not included, though editors have discretion on this. This is standard practice (at WP and generally), and adequately described at WP:CITEHOW. ~ J. Johnson (JJ) (talk) 20:40, 25 August 2013 (UTC)

Agree publisher for common periodicals isn't needed. It's no disaster if an article chooses to always have periodical AND publisher as a 'citation style' variation, but it's unnecessary and not the normal practice. There are tens of publisher-level parameters that could be included in each citation (publisher name, location, print ISSN, online ISSN, subscription model, open source/licence, editors, number of peer reviews, founded, website, office address, number of publications, parent company etc.), if you want to reference/detail this publisher info it seems to me it would be better to wikilink the journal name to refer readers to the journal infobox on its own article. Rjwilmsi 21:00, 25 August 2013 (UTC)

Yes, quite often the publication itself is more notable than the publisher. ~ J. Johnson (JJ) (talk) 21:17, 25 August 2013 (UTC)

Hi. I am afraid I have to disagree with your "quite often"; it is the opposite: Microsoft is one publisher which is more famous than its publications. And its publications are numerous. But that's not just it: I edit in an area of Wikipedia in which the main medium is the web, the second is paper books. Some may dispute me by naming several journals, only to realize "medium is the web" remains uncontested. Fact is, I am yet to see a source that has DOI or ISSN. Best regards,
Codename Lisa (talk) 23:29, 25 August 2013 (UTC)

This discussion is about whether a publisher should be generally included for periodicals, not web pages. As for citing a periodical published by Microsoft, if you feel it is necessary, important, or even just useful to mention that, fine, there is no prohibition. But, for periodicals, that is the exception, and I am hard put to think of any major periodicals whose notability is eclipsed by that of the publisher. As to web pages, sure, as you say, but that is not in contention here. ~ J. Johnson (JJ) (talk) 22:06, 26 August 2013 (UTC)

I think in the case of widely known periodicals and highly reputed journals publisher is not needed. However as mentioned in WP:SCHOLARSHIP there are journals "that exist mainly to promote a particular point of view" and "that appear to be reliable journals, are instead of very low quality". Giving the publisher can help identify if a journal or periodical is from a source recognized by the academic community. I am of the opinion that the more information about a source the better, balanced my keeping references as concise as possible. Joining in the off topic discussion, WP:ISSN states, " Normally an ISSN is not part of a citation to a particular article. An ISSN is used to identify the serial, i.e., the journal or periodical. Only use an ISSN in a citation for a particular article if the journal is relatively unknown and the ISSN is provided for verification of the existence of the journal." I think use of DOI is important as it can lead to an item that has moved. - - MrBill3 (talk) 18:52, 21 September 2013 (UTC)

I'm not really editing here any more, I just logged in because someone e-mailed me that I've been mentioned by name recently in an blame-casting way, and upon seeing it, I feel like responding, because it was unnecessarily personalizing. I do not propose that anyone be given a finger-wagging "talking to" for not adding the publisher to a periodical citation, nor that the Citation Style 1 templates spit out huge red error messages when a publisher is not given in one. It is useful (often very, very useful) information to include, however, especially in politically-charged topic areas or when citing obscure sources. Wikipedia is not an academic journal. I don't know why people cannot seem to get this through their heads: WP is emphatically not bound to cite its sources the way your field or industry happens to do so. WP does not care what other publication do for their own internal purposes. Wikipedia does what is best for its own purposes, which generally means whatever is best for its readers, first and foremost. Providing useful information clearly does what is best for WP's readers, rather than discarding the info for no reason other than to conform to some external expectation that is really of zero relevance here. Another thing people are not "getting" is that periodical citation templates are not used only for academic journals, but for all types of periodicals; the fact that professional academics in a field generally already know who the publishers of famous academic journals in their field are, is totally irrelevant to Wikipedia's interests, goals, procedures, and audiences. Get your head out of your own little fiefdom and think about other people, with different needs, knowledge and expectations for a change. Sometimes the number of oh-so-credentialled academics editing here, unable to see past their own noses, are a curse as much as the boon they obviously are in adding a professional level of expertise on academic topics. <sigh> Now, one could say that when the publisher is visually redundant to add, as in |journal=Proceedings of the Botswanan Society of Omphaloskeptics , that the utility-of-the-data argument is potentially inapplicable. However, fans of metadata and the semantic Web would argue that it would actually be better to add a |hide-pub=y parameter and keep the publisher info without displaying it, since the data can still be found and operated on by third-party tools, e.g. one that wanted to compare the number of times that journals by one university were cited on WP vs. those of another, or whatever. Oh, and per WP:CONSENSUS, the fact that "my" wording, making the publisher of the standard dataset of a Style 1 citation (incl. for periodicals), has been around for a long time without controversy actually indicates that it does in fact have consensus; trying to attack it on the basis that it was written by one person instead of arrived at through a long committee process – especially a committee that included you and pandered to your personal pet peeves – is total WP:BOLLOCKS, and clearly forget that WP:BOLD is a matter of policy. No one needs your permission to institute a best practice here, community acceptance of one without a process you personally favor doesn't mean the acceptance is magically invalid, and that acceptance means that the onus is on you to demonstrate why the idea should be revoked. I'm logging back out and going away again, so feel free to rant and rave at whatever strawman you're going to erect in my absence, since its so important for some of you to personalize the issue and point "SMcCandlish is a bad guy, and we can undo everything he ever did" fingers now that I'm generally not around any more. Have fun with that. I have a real life to get back to, thanks. — SMcCandlish  Talk⇒ ɖ^⊝כ^⊙þ Contrib. 16:44, 3 October 2013 (UTC)

Last. Title. ((cite book)): Unknown parameter |subscription= ignored (|url-access= suggested) (help)

Last. Title. ((cite book)): Unknown parameter |registration= ignored (|url-access= suggested) (help)

Seems to me that if one is linked, the other should be linked as well. --  Gadget850 ^talk 22:30, 26 August 2013 (UTC)

Have these been added recently? What an excellent idea - much better than separate templates. Can we get a bot to convert existing instances? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 22:40, 26 August 2013 (UTC)

At the moment they both reflect their template heritage. I would suggest that if either or both link to something it should be to a page for a reader rather than a page for the editor – |registration= links to WP:V, a Wikipedia policy which has as its target audience Wikipedia editors, not readers.

It seems to me that if we link either of these parameters, the implication is that by clicking that link, the reader can register or subscribe. Perhaps better would be to add a small help link:

(Subscription required (Help)) (Help:CS1 errors is probably the wrong place to link but serves the purpose in this example)

—Trappist the monk (talk) 23:13, 26 August 2013 (UTC)

Help:Source accessibility? --  Gadget850 ^talk 23:41, 26 August 2013 (UTC)

Like this?

—Trappist the monk (talk) 23:56, 26 August 2013 (UTC)

Once there's consensus for the links and they've been changed, could someone please update the template documentation? (e.g. Template:Cite web/doc and the others, Template:Subscription required/doc, Template:Registration required/doc) Thanks! GoingBatty (talk) 16:44, 29 August 2013 (UTC)

I'm with Trappist. I also don't think 'registration required' should be linked to any WP page. -- Ohc ^¡digame!_{¿que pasa?} 22:36, 29 August 2013 (UTC)

So the above ((cite compare)) examples are not currently showing what my "Like this?" question was about. That's because I found a bug in the current live version of Module:Citation/CS1 so I copied live to Module:Citation/CS1/sandbox, fixed it there where it awaits movement to live version – I'm not special enough to do those kinds of things.

As I have given this topic some more thought, I'm not sure that a help link in these link notes is really necessary. As plain text, they forewarn the reader that access to the cited source requires either a subscription or registration. How much more do we need to tell the reader? Surely, when they click the associated citation link they will figure out that the source material has access restrictions.

I have also had second thoughts about the current choice of a page name if we do decide to provide a help link. Accessibility as a term comes with a bit of baggage. It is very closely connected to making Wikipedia available to those with disability whom we want to be reading our work. Unfortunately, I'm not sure I know of a better term but maybe I don't have to because as it sits right now, I'm content to not have links as part of the subscription and registration link notes.

—Trappist the monk (talk) 23:10, 29 August 2013 (UTC)

An alternative to having a regular link to off-page help text might be something like this:

(subscription required (Help))

This uses <span title=""></span> for the text and <abbr></abbr> for the text decoration. I think that this violates the proper use of the <abbr></abbr> tags. Perhaps there is another way to accomplish the same thing without the violation.

—Trappist the monk (talk) 12:53, 30 August 2013 (UTC)

And there is a way. Style the span with the same css as is used for the <abbr></abbr> tags:

(subscription required (Help))

—Trappist the monk (talk) 13:21, 30 August 2013 (UTC)

With the bug fix now applied to the live version of Module:Citation/CS1, the ((cite compare)) above now reflects my latest thinking regarding the Subscription / Registration link note help text.

—Trappist the monk (talk) 23:12, 1 September 2013 (UTC)

I took a shot at updating the documentation at Template:Cite web/doc. Any improvements would be appreciated. Thanks! GoingBatty (talk) 14:56, 14 September 2013 (UTC)

Yeah, good. I changed the style a bit so that it matches the rest of the parameter definitions. Ultimately, the text needs to move into its own template so that it can be transcluded in the other CS1 documentation pages. We can wait on that until we're sure that the documentation at ((cite web)) is acceptable.

In the template data is this: When set to <code>yes</code>, ... Does the visual editor display that correctly (hides the<code></code> tags and displays yes in the proper font)?

—Trappist the monk (talk) 15:49, 14 September 2013 (UTC)

What on Earth happened to the Cite book template? Wherever I used it, there is now an error saying: |accessdate= requires |url= (help). See, for example, Margaret of Burgundy, Dauphine of France#References and Royal touch#References. How can this be fixed? Surtsicna (talk) 10:49, 27 August 2013 (UTC)

• If there's no url, the accessdate is redundant, by definition because |accessdate= refers to the date upon which the on-line citation is accessed by the editor citing it. If you are referring to a printed work, it's pretty obvious that it's the edition of the book (or its ISBN) that matters to the person wanting to look up the physical work. The date on which the person who input the citation read it is utterly unimportant. In the case where there is no url, the existence of the warning suggests that the best option is to remove the access date. -- Ohc ^¡digame!_{¿que pasa?} 10:55, 27 August 2013 (UTC)
  • • URL can be anywhere. See below: "#People sick of wrong cite messages". -Wikid77 04:29, 6 September 2013 (UTC)
  • Thanks! It's still a bit odd, though, but at least I know how to sort it out. Surtsicna (talk) 12:44, 27 August 2013 (UTC)
    • It's odd because it's not true, as noted months ago, a URL can be in a page number, so accessdate applies. See below: "#People sick of wrong cite messages". -Wikid77 04:29, 6 September 2013 (UTC)
• How did the the help page not make this clear? --  Gadget850 ^talk 11:49, 27 August 2013 (UTC)
  • I did not know about the help page, sorry. I came here via Template talk:Cite book redirect. Surtsicna (talk) 12:44, 27 August 2013 (UTC)
• And these errors were enabled yesterday. I have tweaked the help page a bit. --  Gadget850 ^talk 14:06, 27 August 2013 (UTC)
  • Well I wish someone had told me about this from the start. I will be wasting valuable editing time removing redundant access dates. Would be much easier is someone could just reverse the change that caused the error to show up. Where is the discussion? Just want to have a nosey. :)Rain the 1 19:42, 27 August 2013 (UTC)

• What is frequently updated without changing the publication date, but doesn't have a URL? I don't have a good answer for that, so I'm not in a position to request this change be reversed. Jc3s5h (talk) 19:49, 27 August 2013 (UTC)

In the old version of the templates, 'accessdate' without 'url' did not show. That change was made years ago and the discussion is buried in archives linked at the top of the page. We often got questions about why 'accessdate' did not show. Now we give an error telling you that it is either redundant or something is wrong in the citation markup. --  Gadget850 ^talk 14:41, 30 August 2013 (UTC)

People sick of wrong cite messages

Someone might wonder, "Why are only "10" people complaining about the excessive cite messages?" Well, people have talked and talked, and explained and explained, for months and months, but some just refuse to listen:

• "A URL can be in a page number, volume or issue, and accessdate applies" (hello?)

For example, "page=[http://www.x.com/p8 8] |accessdate=7 May 2011" where the URL is coded inside the "page=" or likewise, inside a "volume=" or inside an "issue=" (etc.), but instead we get these dim, wrong error messages:

• ((cite_web|title=Report Z|page=[http://www.x.com 8]|accessdate=7 May 2011))
  → "Report Z". p. 8. ((cite web)): |access-date= requires |url= (help); Missing or empty |url= (help)

So, we have 2 wrong error messages, which completely ignore the URL in the page number and insist on scarring the cite as being doubly-invalid, with an accessdate but supposedly no URL (hello, see the URL in the page number?). Unfortunately, many people do not have the patience, the time, the tolerance, nor the self-control, to keep explaining (over & over) how the restrictions about accessdate and "url=" are wrong, wrong, wrong, and what's the word? ...oh yes, wrong. Consequently, many people are utterly, totally, completely, and thoroughly fed-up, sick, tired, and over-it about the continual forcing of these narrow, severe, trivial messages which are wrong, wrong, WRONG, as explained for months and months. The cite templates should never flag the use of accessdate, nor insist that "url=" must be set, when a URL can be passed in other cite parameters. At most, set a link to a warning category. In fact, for many web news stories, a source webpage (by title/date or author) can be found at many URLs which published the same story (such as by AP feed), and there is no need to restrict to one URL of several matches. -Wikid77 (talk) 04:29/04:51, 6 September 2013 (UTC)

State the source of your quoted text: A URL can be in a page number, volume or issue, and accessdate applies.

Because Citation Style 1 produces COinS metadata, the inclusion of a url in |page=, |pages=, or |at= corrupts the metadata. The metadata for your example citation is:

<span title="ctx_ver=Z39.88-2004&rfr_id=info%3Asid%2Fen.wikipedia.org%3AHelp_talk%3ACitation+Style+1&rft.btitle=Report+Z&rft.genre=book&rft.pages=%5Bhttp%3A%2F%2Fwww.x.com+8%5D&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Abook" class="Z3988">

and specifically this bit: &rft.pages=%5Bhttp%3A%2F%2Fwww.x.com+8%5D, which should be just: &rft.pages=8.

There is a proposal for a new |pageurl= parameter at Module talk:Citation/CS1/Feature requests which will allow CS1 citations to present a linked page number without also corrupting the COinS metadata.

—Trappist the monk (talk) 11:54, 6 September 2013 (UTC)

Turn off error messages

A bot request has sensibly been made to fix the errors discussed in the preceding section. I suggest that, in the short term, we turn off the error warning (while retaining the category), specify the bot and let it do its job, then turn on the error warning to catch future issues. The same should apply to any other error states that might be waiting to be flagged up. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 19:56, 27 August 2013 (UTC)

Can we do this as a priority, please? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 11:07, 2 September 2013 (UTC)

@Andy Mabbett, I have explained below how to easily re-hide the thousands of annoying CS1 error messages, in subthread below, "#Fix Module:Citation/CS1/Configuration to reset each hidden=true". -Wikid77 (talk) 21:34, 3 September 2013 (UTC)

Fix Module:Citation/CS1/Configuration to reset each hidden=true

The numerous excessive red-error messages, in the wp:CS1 cites, can be re-hidden by fixing Module:Citation/CS1/Configuration to reset each newly activated message, back to "hidden=true" (set "false" on 26 August 2013 by User:Gadget850, see: dif410). For example, the message named by anchor = 'cite_web_url' (which displays "Missing or empty |url=") can again be hidden, as during the past 5 months, by setting the associated variable as "hidden=true". Do a similar reset to hide other messages which are flooding major articles with numerous annoying messages about fixing dozens of URL parameters or other tedious busywork. -Wikid77 (talk) 21:34, 3 September 2013 (UTC)

Your fix for Module:Citation/CS1 presumes that it is broken. It is not. It is functioning as it should. If you don't want to see the error messages, see Help:CS1 errors#Controlling error message display.

—Trappist the monk (talk) 21:46, 3 September 2013 (UTC)

@Trappist the monk: Please see my comments immediately above. While we have a bot in prep to fix these errors, there is no need to alarm many editors and readers with them. The warning should be temporarily removed, the bot allowed to complete its task, and only then should the warnings then be re-enabled, to catch new instances as they arise. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 12:17, 4 September 2013 (UTC)

Yep, read that. We unveiled CS1 errors in three groups: 13 April, 19 April, and 26 August 2013. In the time since the first unveiling, I have seen only a smattering of confusion, interest, concern, some hostility; but, overall, I see no groundswell of opposition or alarm from editors or readers to the presence of CS1 error messages, just apathy, lots of apathy. If there is evidence to the contrary, please show me where it is.

—Trappist the monk (talk) 13:18, 4 September 2013 (UTC)

Well, WP:VPT#What's messing with Citation templates? for one; and this latest round seems to be more problematic than those before it. However, you have this the wrong way round, If a bot remedy is in in hand, what is the justification for displaying red error messages in the meantime? Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 14:26, 4 September 2013 (UTC)

Eleven editors in that discussion. Doesn't seem like much of a rebellion against the CS1 error messages.

—Trappist the monk (talk) 12:24, 5 September 2013 (UTC)

Numbers are not the point; but it doesn't demonstrate a wider consensus to display these error messages so extensively. You appear to have overlooked my question. Andy Mabbett (Pigsonthewing); Talk to Andy; Andy's edits 13:34, 5 September 2013 (UTC)

I don't think that I made any claims about consensus one way or another. I merely stated that I haven't seen the masses rising up to denounce the unveiling of the last group of error messages. If your bot is in hand, is it not already repairing the citations so what need to hide the error messages?

—Trappist the monk (talk) 15:23, 5 September 2013 (UTC)

RfC