Template talk:Sfn/Archive 3

Archive 1

Archive 2

Archive 3

Archive 4

Archive 5

No author name

During my FAC of the 2012 Tour de France there has been a comment made by Ealdgyth of the consistency when using an acronym (ASO) in the template but the full name (Amaury Sport Organisation) in the citation. I've used the guide at Template:Sfn#No author name in citation template, so I'm not really sure what to do about it. BaldBoris 14:50, 7 January 2017 (UTC)

For me, I would use the source title for the name simply because the title is the first thing that appears in the full citation. Readers may expect when they click on ASO 2012 that they will be taken to something that is obviously ASO. But, that citation is not obviously ASO. So, minding the options listed at Template:Sfn#No author name in citation template, I would write {{sfn|''Race regulations''|2012|p=35}} and |ref={{harvid|''Race regulations''|2012}} so that clicking on Race regulations 2012 and landing at Race regulations meets the readers' expectations.

—Trappist the monk (talk) 15:22, 7 January 2017 (UTC)

(Edit conflict.) Sometimes people may want to find the full citation that corresponds to a short citation manually, rather than clicking a link. This might occur if the article has been printed on paper, the device being used to read the article does not follow links gracefully, or if the reader is using a large screen, and has one window for the running text and a different window for the reference list. In these cases, the field that is used instead of an author name in the short cite should be the same that is used to alphabetize the reference list. Since "Race regulations" is used to alphabetize the reference list, that's what I would use in the short citation. I believe Template:Sfn#No author name in citation template ought to be revised to reflect the desirability of the short citation helping the reader to find the correct reference in an alphabetical list. When a title must be shortened, I think the alphabetical aspect should be considered when choosing the short title. Jc3s5h (talk) 15:25, 7 January 2017 (UTC)

I would keep formatting - such as italics - out of the link. --Redrose64 🌹 (talk) 15:43, 7 January 2017 (UTC)

Why?

—Trappist the monk (talk) 15:44, 7 January 2017 (UTC)

The parameters of {{sfn}} are intended for clean values. Anyway, I have working examples at NBR 224 and 420 Classes#Notes - the Gradient Profiles 2003 ref and all the SLS 1970 refs. Or try Reston railway station#Notes - there are at least six authorless refs there. --Redrose64 🌹 (talk) 16:06, 7 January 2017 (UTC)

...intended for clean values. What does that mean, and where is it documented? Interestingly, the second example in the no-author documentation uses: {{sfn |''Popular Magazine''|1996|p=29}}

—Trappist the monk (talk) 16:14, 7 January 2017 (UTC)

@Redrose64: Do you have a response to Trappist the monk's question? I only ask because the FAC is the oldest listed so there's relative urgency. Another question Trappist, should I be using Augendre 2016 instead of Historical guide 2016 at 1962 Tour de France (and the 100 odd other Tours)? BaldBoris 16:53, 8 January 2017 (UTC)

Using the author's name is, I think, the preferred usage in author-date referencing, so yes, Augendre 2016.

—Trappist the monk (talk) 17:23, 8 January 2017 (UTC)

Sorry, I forgot to that I used "Historical guide" because the PDF (4 MB) has Textes de Jacques Augendre on the front page, and because I've only referenced the data I didn't use Augendre 2016. Should I have him as the author? Also, if Redrose doesn't reply, shall I go ahead with your suggestion? BaldBoris 17:38, 8 January 2017 (UTC)

It is not at all clear to me which 'textes' are authored by Augendre. The "Préface" is not signed as those kinds of thing often are. Does 'textes' apply only to the "Préface" or does it apply to all 'textes' in the individual year pages as well (the La petite histoire items, for example)?

To the original question, you know my opinion and the opinions of the other editors here. The decision is yours; we cannot make it for you.

—Trappist the monk (talk) 19:08, 8 January 2017 (UTC)

I'm sure that there were problems in putting markup into templates that use their input params to make links. It's certainly nothing that I've seen out in the field - conclusion, we just don't do it. --Redrose64 🌹 (talk) 19:11, 8 January 2017 (UTC)

This insource: search finds about 600 pages with {{harvid}} templates that have as their assigned values italic wiki markup. Another search finds about 20 pages with {{sfnref}} templates and italic markup. Not very common, but they do exist.

There once was an issue in the pre-Lua and possibly early-Lua days having to do with anchor encoding. After {{sfn}} was converted to use the Lua Module:Footnotes, this edit made it so both Module:Footnotes and Module:Citation/CS1 used the Lua library function mw.uri.anchorEncode to encode the <cite>...</cite> tag id= attribute (cs1) and the URI #fragment (Footnotes). What appears to have once been a problem no longer appears to be a problem.

—Trappist the monk (talk) 20:05, 8 January 2017 (UTC)

Considering he's 91, I'm pretty sure it's not the most recent Tour pages. All I know is that he is/was Tour de France's archivist. It would most likely be all you said (possibly copied from this book). I'll just leave him as the author to be on the safe side. Sorry for going off topic, just thought I'd ask while I had you. Thanks for the help. BaldBoris 20:33, 8 January 2017 (UTC)

Thinking about it, wouldn't |others=Text by [[Jacques Augendre]] be on the safe side? Then use the same as your suggestion for my original query {{sfn|''Guide historique''|2016|p=29}}, changing it to the French title. BaldBoris 19:41, 9 January 2017 (UTC)

You can do that if you believe it to be correct. |others=, to me, implies or infers that there is a primary author. In future, after you have abandoned Wikipedia for greener pastures elsewhere, editors who remain here may be perplexed when they see the template with |others= but without a primary author.

—Trappist the monk (talk) 23:20, 9 January 2017 (UTC)

Cheers for the responses guys. I had actually used Race regulations previously, until I read the doc here and changed it back in September 2015. I'll wait or the above convo to be resolved before I implement it. BaldBoris 16:51, 7 January 2017 (UTC)

I've had discussions about similar in the past, and I've a slight suspicion that there may be a transatlantic divide in practice. My feelings are that the author field should reflect "who is responsible for writing it", even if that is a corporate author. In this case that is clearly ASO. Not being familiar with sport cycling I'm making a guess here, but if the ASO is a recognised body known by those initials, why not use ASO as the author and "Amaury Sport Organisation" as the publisher. You can then use {{sfn|ASO|2012}} and both the linkages and manual observation work. I think it would be hard to justify "Race regulations" as an author! See Frindsbury where you'll see both the "Chatham News" and the "Ordnance Survey" listed as authors. Martin of Sheffield (talk) 17:10, 7 January 2017 (UTC)

The citation templates have a bunch of parameters, but the rules for how to use the are not fully specified. Some external style manuals encourage using corporate authors, others discourage this. The citation templates have no guidance one way or the other. I have never heard of any style manual that suggests listing the same entity as both the author and the publisher (except sometimes when a natural person is both the author and the publisher, in which case "by the author" is sometimes put in the publisher position). Also, it is common practice in short citations to use a shortened title instead of an author when no author is known. External style guides need to accommodate sources that would be unacceptable in Wikipedia, so those guides have to provide for anonymous works where neither the publisher nor the author is known. Jc3s5h (talk) 17:55, 7 January 2017 (UTC)

There are problems with citations like this:

{{Citation
  | author = Chatham News
  | date = 1 December 1967
  | title = Probe proves area is riddled with tunnels
  | url = http://cityark.medway.gov.uk/query/results/?Mode=ShowImg&Img=/cityark/Scans/Unofficial_or_Privately_Originated_Collections/DE0402_Couchman_ephemera_and_MSS_/DE0402_13.html/DE402_13_55.jpg
  | postscript = .  Cutting held by Medway Archives and Local Studies Centre.
}}

Chatham News (1 December 1967), Probe proves area is riddled with tunnels. Cutting held by Medway Archives and Local Studies Centre.{{citation}}: CS1 maint: postscript (link)

Newspaper article titles should not be rendered in italics. The title is rendered in italics in this example because {{citation}} defaults to 'book' style in the absence of a |work= (or alias) parameter. Chatham News is/was a newspaper and should be identified as such. Newspaper, magazine, and other periodical titles are, per MOS:MAJORWORK, to be rendered in italics. cs1|2 at Help:Citation Style 1#Authors suggests that when there isn't an identified author, editors should use |author=<!--Not stated-->.

Writing a citation like this example produces corrupted metadata: Chatham News is not an author; "Probe proves area is riddled with tunnels" is not a book.

The {{sfn}} and {{harv}} templates do not produce metadata that can be corrupted. If you wish to use Chatham News as the 'name' in the short reference, that's ok because you can write |ref={{harvid|''Chatham News''|1967}} in the cs1|2 template which maintains the link between the {{sfn}} and the {{citation}} templates and protects the citation's metadata.

—Trappist the monk (talk) 18:02, 7 January 2017 (UTC)

Right. Also right that Chatham News is not an author. That produces a seeming disconnect. The missing author in the citation and use of the publication name as author in the sfn can be accommodated by something like the following (see wikitext for details):

• some text
• some text elsewhere
• "Probe proves area is riddled with tunnels", Chatham News example, 1 December 1967. Cutting held by Medway Archives and Local Studies Centre.{{citation}}: CS1 maint: postscript (link)

Wtmitchell (talk) (earlier Boracay Bill) 23:27, 7 January 2017 (UTC)

In the absence of an author, I don't think using the name of a newspaper as the name that links the short citation to the full citation is a great idea, because it would not be surprising cite have several different articles from a single issue of a newspaper. It's much less likely that the same issue of a newspaper would have two or more articles with the same title. Jc3s5h (talk) 00:07, 8 January 2017 (UTC)

Referring to newspaper articles by name of the paper where the author is (as common) unidentified is usually more significant to a reader than the title of the article. Note also the standard practice of listing all newspaper articles under the name of the paper, and by date of publication, regardless of author or title (as seen here). Where there is more than one article on a given date just add a suffix letter to the date. ~ J. Johnson (JJ) (talk) 00:30, 10 January 2017 (UTC)

Looking for advice

Is there a way to create a short ref with {{sfn}} for two authors so that it says "Smith and Jones", not "Smith & Jones", and without p.? I'd like it to create "Smith and Jones 2017, 1". SarahSV ^(talk) 04:52, 5 March 2017 (UTC)

@SlimVirgin: I'm not aware of any good way to do what you're asking for here. {{sfn}} doesn't have any way to change the & to " and " (it's hardcoded in Module:Footnotes). You could try something like {{harvs|txt=yes|nb=yes|last1=Smith|last2=Jones|year=2017|loc=1}}, which produces:

Smith and Jones 2017, 1

(the year is the link to the full citation, unlike {{sfn}} which links author and date).

For {{sfn}} I think the best you can get is using |loc=1 instead of |p=1 to avoid the addition of a "p. " prefix. {{sfn|Smith|Jones|2017|loc=1}}

Smith & Jones 2017, 1

It looks like adding the ability to configure the separator in Module:Footnotes could be doable within reasonable parameters of complexity, but I'm not sure the need for that feature is big enough to be worth the effort, or that the feature is necessarily desirable (for several reasons). --Xover (talk) 07:52, 5 March 2017 (UTC)

Xover, thanks for letting me know about |loc=. SarahSV ^(talk) 21:08, 5 March 2017 (UTC)

I find it easy enough to use {{sfn}} (or actually I prefer sfnp) for the simple cases, and <ref>{{harv...}}</ref> for the more complex ones. Still, using "and" instead of "&" seems like a reasonable stylistic choice that we might want to support better. —David Eppstein (talk) 08:04, 5 March 2017 (UTC)

@David Eppstein: why? This is yet another example of "I like it like this so you have to support me". How is it helpful to readers or new editors? Peter coxhead (talk) 09:38, 5 March 2017 (UTC)

Well, from a purely technical perspective, I could probably hack up a naïve implementation of this in pretty short order (and I am decidedly inexpert at this). We could have a parameter analogous to |ps=, where you add |as=a to get " and " instead of " & " in front of the last author (as here short for "author separator", and just "a", for "and", for brevity when using). Alternately, we could allow editors to specify the full string—|as=" and "—which would allow more flexibility for things like |as=", and " (for the adherents of the Oxford comma), but at the expense of having to actually specify the full string for every short reference (including correct use of spaces, which probably means things like &nbsp;).

However, this does add a bit of complexity to a template that appears designed specifically for simplicity; and there is a question of whether it is desirable to encourage such variations. cf. Peter coxhead's comment above (which question I am not taking a position on, just noting the objection). --Xover (talk) 09:53, 5 March 2017 (UTC)

You can use {{sfnref}}:

{{sfn|Black and White|2017|loc=1}}

{{cite book |title=Title |last=Black |first=D |last2=White |first2=B |ref={{sfnref|Black and White|2017}} |date=2017}}

Black, D; White, B (2017). Title.

References

—Trappist the monk (talk) 10:05, 5 March 2017 (UTC)

• Harvard refs are the tool of Satan. That's all I have to say. EEng 10:16, 5 March 2017 (UTC)

Something from Harvard that you're against? Wow. —David Eppstein (talk) 22:46, 5 March 2017 (UTC)

Smartass. As I've explained before [1] AFAIK no one at Harvard calls them Harvard refs. Same with the Harvard comma. It's blatant trademark infringement. EEng 23:07, 5 March 2017 (UTC)

SarahSV, if you're editing an article written by other editors that has the citation templates including {{sfn}} as its established style, it seems to me it would be contrary to WP:CITEVAR to go against that style. If you're writing a new article, or overhauling an existing article with hopelessly messed-up citations, and want to use citation templates that are similar to, but not quite the same as, the existing ones, go right ahead and create your own set of templates and see if they catch on. Jc3s5h (talk) 16:19, 5 March 2017 (UTC)

Jc3s5h, I don't normally use citation templates, but I'm considering dipping my toe in the water by using {{sfn}} on an article I've written that currently using manual refs. So I'm trying to figure out whether I can do that while preserving some or all of the appearance of the current refs (in particular, no "p.", and "and" instead of "&". SarahSV ^(talk) 21:13, 5 March 2017 (UTC)

Mob of {sfn} zombies pursues {r} interloper

Don't go over to the dark side, SSV! It's not too late! Salvation can be had with {{r}}! As it happens I just did a demo on another article to pitch using {r} on it. [2] Give it a look. You'll never go back! EEng 22:12, 5 March 2017 (UTC)

EEng, the clickiness of {{sfn}} is seductive. I'm trying to figure out whether it's worth the extra work, and what I'd lose by doing it. Instead of short refs, I've tried {{rp}} a few times, but it's ugly when there are lots of refs together, especially with page ranges. SarahSV ^(talk) 23:13, 5 March 2017 (UTC)

I personally find {{sfn}} much easier to use than either manual refs or <ref> ... </ref> pairs. The references are normalised and alphabetised. You only need to do the refs once, and can detect if someone else has done the same ref. Page numbers are a doddle, they apply and are linked to the ref which can be found either by clicking or simply by looking. Have a look at Subhas Chandra Bose for an extreme case where intermingling the page citations and references would be an absolute nightmare (actually, was a nightmare, see this old version for how bad a random mix gets). Is the &/and really an issue or just a stylistic preference? If you don't want page numbers in a particular citation note that |p=n is optional. Martin of Sheffield (talk) 23:39, 5 March 2017 (UTC)

Hi Martin, the "and" is just a stylistic preference; "and" seems more professional than "&", I suppose. I'm glad to find a way not to have to repeat "p."; that was more of an issue for me, as I'm removing those from articles I write, so I didn't want to have to reintroduce them with {{sfn}}. SarahSV ^(talk) 23:19, 6 March 2017 (UTC)

MoS, the old version you link mixes footnotes ("explanatory" footnotes) with source citations, and no one's suggesting doing that. I'm afraid this thread will bog down, as usual, in the lack of agreed meanings for terms like citation, reference, footnote, etc. And of course, p= is optional with {r} as well, plus it has a way less fragile syntax. The loss of the alphabetized bibliography is unfortunate, but I don't think readers care much. EEng 00:05, 6 March 2017 (UTC)

@Martin of Sheffield: There are a handful of errors at Subhas Chandra Bose where the short form references don't correctly link to their attendant long-form citations. Consider using the User:Ucucha/HarvErrors tool which is very helpful in making sure that all of those links work.

—Trappist the monk (talk) 23:49, 5 March 2017 (UTC)

A great example of the fragility of {sfn} and its friends. EEng 00:06, 6 March 2017 (UTC)

@Trappist the monk: thanks for the link to User:Ucucha's script. @EEng: actually it shows how easy it is to fix a couple of minor typos. Martin of Sheffield (talk) 09:49, 6 March 2017 (UTC)

Everything in a wiki is easily fixed, because everything's preserved and all changes are reversible -- once you figure out what's wrong. The problem with sfn is the fiddly, comparatively verbose syntax which requires that the author's name, year, and sometimes more be repeated on every invocation -- compact, simple refnames can't be assigned as with < ref>< /ref> and {r}. EEng 15:16, 6 March 2017 (UTC)

Break 1

Hi, Sarah. {{Harvid}} allows creation of any form of cite you want. Something like [[#{{harvid|Smith|Jones|2017}}|Smith and Jones 2017]] will give you this: Smith and Jones 2017. (Which, of course, can be embedded in <ref>...</ref> tags, or not, as preferred.) This links to Smith, A.; Jones, J. (2017), Dummy citation.

If you need any help I would be pleased to be of assistance. ~ J. Johnson (JJ) (talk) 23:05, 6 March 2017 (UTC)

Thanks, J. Johnson, that's very kind. I'm currently testing on preview without saving; looking at trickier examples (linking page numbers to Google Books pages; book chapters with authors' and editors' names; multiple refs with same author and year, etc. I might give you a shout at some point. SarahSV ^(talk) 23:23, 6 March 2017 (UTC)

So you're going over to the dark side after all. Sometimes I just don't know why I get up in the morning. EEng 23:32, 6 March 2017 (UTC)

EEng, I haven't decided yet. It would be a big step. I'll keep you posted! SarahSV ^(talk) 00:45, 7 March 2017 (UTC)

Point me to your article and I'll demonstrate The One True Referencing Method™ in my sandbox. EEng 01:46, 7 March 2017 (UTC)

EEng, the one I had in mind was The Life of Mary Baker G. Eddy and the History of Christian Science. SarahSV ^(talk) 02:31, 7 March 2017 (UTC)

I just looked at that article, and (ha ha ha) she's already using short cites. (She's mine, mine!!) :-)

Sarah, don't let the big engineer fellow scare you, you're already halfway to enlightenment, and the rest can be done in easy steps. One of the reasons I advocate short cites is that then the dreary masses of bibliographic detail in the full citations can be pulled out of the text (where they just make the text harder read, and the citations harder to maintain) into their own section. But you have already done that! In this case, using {sfn} would be putting the <ref>...</ref> tags full citations back into the text, a definite (and big!) step backwards[oops, I was thinking of something else]. And you are already using short cites in the notes. So just replace all the stuff like <ref name=Fraser1999p139>Fraser 1999, 139.</ref> with <ref>{{Harvnb|Fraser|1999|p=139}}.<ref>. That will give you formatted short cites in the same style you're using now, without the hassle of of chasing "named refs" around. At that point you won't be linking (yet!) to the full citation, but there is More Than One Way to do that, which I'll explain on the Talk page. ~ J. Johnson (JJ) (talk) 00:05, 12 March 2017 (UTC)

BTW, I have made one such change to the article to illustrate. ~ J. Johnson (JJ) (talk) 00:15, 12 March 2017 (UTC)

J. Johnson, thanks for illustrating that. I wasn't thinking of putting long refs back into the text, but of writing {{sfn|Fraser|1999|loc=139}}, then converting the long refs in the Works cited section to templates. What are the differences between sfn and harvnb? SarahSV ^(talk) 00:23, 12 March 2017 (UTC)

{{sfn}} incorporates <ref>...</ref> tags and names the tag using the parameters in the template. So, your example:

{{sfn|Fraser|1999|loc=139}}

produces:

<ref name="FOOTNOTEFraser1999139">[[#CITEREFFraser1999|Fraser 1999]], 139.</ref>

{{harvnb}} with the same parameters:

{{harvnb|Fraser|1999|loc=139}}

produces:

[[#CITEREFFraser1999|Fraser 1999]], 139

—Trappist the monk (talk) 00:55, 12 March 2017 (UTC)

Yes. And if you drop an {{anchor}} incorporating those parameters – something like {{anchor|CITEREFFraser1999}} – in front of your full citation the link above will have some where to go. (Go to note #11 in the article and click on "Fraser 1999".) It won't highlight the whole citation; for that use either the {cite} template (with |ref=harv) or {citation} template. Some people don't like to use templates, but the slight extra effort has some advantages, like not having to futz around with links or formatting. ~ J. Johnson (JJ) (talk) 03:49, 12 March 2017 (UTC)

Sarah: I see your recent edit where you use [[#Bates|Bates and Dittemore 1932]]. Yes, it can done that way, too, but.... If you are adding these manually, "#Bates", being less typing, might seem preferable to (say) "#BatesDittomore1932". But if you have another Bates you have to use another tag, and then you are creating arbitrary tags you have to keep track of. It is simpler and less trouble in the long run to have the tags follow the form of the cite. And if you keep the "CITEREF", conforming with the default pattern of {harv} and {citation}, then you have no problem intermixing all of these. Also less breakage if you want to do some of these changes section by section. ~ J. Johnson (JJ) (talk) 05:09, 12 March 2017 (UTC)

J. Johnson, I see what you mean. The more I see this, the more I like it. I may try it out on that article myself. SarahSV ^(talk) 03:15, 13 March 2017 (UTC)

You may have won this, time, J. Johnson, but I'll be back! – EEng

• FTR, while I continue to believe that {r} is better, as between the manual syntax and the harvnb/sfn/whatever machinery, I'll take the harvnb/sfn/whatever machinery. EEng 00:29, 12 March 2017 (UTC)

(Maybe I'd better head into the History department. He'll never think of looking for me there! ~ J. Johnson (JJ) (talk) )

• EEng (he's back!!) has been demonstrating use of the {{rma}} ("reference, manually anchored") and {ran} templates on the article. Quite interesting, but I will take issue with the documentation where it says "any string (preferably very short)" can used as a tag. While it is true that any unique string can be used, "very short" is not necessarily preferable. In practice, it is harder to keep track of the sources when editing an article where the sources are referred to rather cryptically as "B1", "B2", etc. (Which one is Bates? or Brown? or whatever the other one is?) Regardless of what template (or even no templates) one uses – whether 'named refs', {harv}, {rma}, whatever — in all cases it is a better practice to use tags that carry some minimal identification of the source. Such as "Fraser 1999". Which, being the same form as the short cite (though other variations are possible), provides a clue as to which source is referred to. (The same issue programmers have in the naming of variables.)

Though this becomes a problem with {ran}, as the ugly note-links it creates – such as ^{[B2]:288-94[S3]:xvff} — become even more intrusive, reminiscent of the much despised {{rp}}, and making the often hated parenthetical style look better. And it appears that {reflist} doesn't catch instances of {ran}, so there is no handy list (or the list is incomplete) of where a source is used in the text. Handcrafting links in <ref> tags is distinctly better than using {ran}. ~ J. Johnson (JJ) (talk) 22:40, 12 March 2017 (UTC)

• The superscript callouts in the style of {rp} startle some editors at first, but you soon don't notice them. I get a lot of private feedback on Phineas Gage because of my personal work on Gage, and no one's ever said anything about them unless I've asked, and the response is always, "Oh, I didn't notice them". I think it's something editors are sensitive too (because they're always fussing with refs in the course of editing) but readers don't even notice. The Eddy article unfortunately has a lot of roman page numbers, so indeed there are pimples like ^{[B2]:288-94[S3]:xvff}. If a particular string is too intrusive they can be banished thus [3]
• The lack of backlinks (the "handy list of where a source is used") is not always desirable -- se this, where the list of backlinks are completely useless. {rma}/{ran} works best in an article where, among possibly a large number of sources which are mostly used only a few times each, there's a relatively small subset of sources that are used many, many times each – you use ran/rma for the much-used sources, and the usual < ref> machinery for the little-used sources. If, for the rma/ran sources, you want to find where it's referenced, just use <ctrl>-F.

I think if you look closely at Phineas Gage#References you'll see why we did it that way we did. EEng 00:16, 13 March 2017 (UTC)

EEng, your system seems to have made some short refs disappear. Compare current (38 entries under References) and previous (59). SarahSV ^(talk) 03:19, 13 March 2017 (UTC)

They have not disappeared, but you only need to look at two places to figure out what is being referenced (the superscript annotation in the article text and the list of references), not three (the superscript annotation, the footnote, and the full reference that the footnote sends you to). For instance one of the references now looks like ^[B1]:258,274 — the superscript B1 (instead of a number) means to see B1 in the list of references. The drawback of this system is that it works only when the added information in each citation is just a page or pages. With short footnotes, I can mix them up with longer and more informative footnotes like <ref>Brakke (1978), Appendix B, Proposition 2, p. 230; Chou & Zhu (2001), Lemma 5.5, p. 130; "6.1 The decrease in total absolute curvature", pp. 144–147.</ref> when I want to. —David Eppstein (talk) 03:33, 13 March 2017 (UTC)

Eppstein's right. The point of rma/ran is to eliminate the simple short refs of the form Smith 1999, p. 5. But tsk, tsk, Eppstein, you're wrong about it works only when the added information in each citation is just a page or pages. With short footnotes, I can mix them up with longer and more informative footnotes -- I can do that too. See [4]. So the majority of short refs (of the type Smith 1999, p. 5) are eliminated, but where you need something more complex, you can still have any complicated ref you want. EEng 03:59, 13 March 2017 (UTC)

Yes, I see now, thank you. But if I want to type in Bates (in read or edit mode) to find where Bates is used, it doesn't appear. I have to know (or remember) that it's called B1, or know to look in the other section for it. Reference 7 is a short ref, "[B2]:288-94[S3]:xvff" (perhaps just a mistake). 24 and 31 are odd too. SarahSV ^(talk) 04:08, 13 March 2017 (UTC)

Ok, I stand corrected. I've tended to waver between a style that uses short footnotes and a separate alphabetized list of longer references (e.g. Curve-shortening_flow, the one I took that example from), and a different style where everything is in the footnotes but sometimes one footnote refers to another using a harv link (e.g. Dehn invariant). But EEng's style is tempting too. One question, though, EEng: suppose you want to make an external link on a page number, to the Google books preview for that specific page. Is it easy to do? —David Eppstein (talk) 04:14, 13 March 2017 (UTC)

[5] EEng 04:37, 13 March 2017 (UTC)

I suppose {{ran|S3|p=[https://books.google.com/books?id=bfUVPgvoNCkC&pg=PT18 xviii]}} works too. —David Eppstein (talk) 05:27, 13 March 2017 (UTC)

How wickedly clever. Yes. But I think the subtlety of two links squeezed into that little space would be lost on most readers, plus you get the little link icon thingamajig in the superscript callout, making the callout bulkier. But the important point it, that with ran/rma you keep all the choices you have with harv/sfn, but can lose the trivial Smith 1999 p.5-type intermediate notes. EEng 05:57, 13 March 2017 (UTC)

[24] used one particular way of doing an Eppstein-type "longer and more informative footnote" -- but here [6] I change it to two a more conventional format. It's completely flexible. To find where Bates is cited, yes you have to look in the source list to find that it's B1, but then it's easy to <ctrl>-F for the string [B1]. Again, the ran/rma system works best in articles where at least some of the sources get many, many refs each -- again, look here for the mess you get when you use < ref> in this case.

[31] was just a typo. EEng 04:37, 13 March 2017 (UTC)

Why "lose the trivial[ha]Smith 1999 p.5-type intermediate notes"? In the first place, they are not "trivial". The full citation for each source should appear only once in each article; the short cites (not "intermediate notes", because they are not "notes", but cites) allow links from multiple points in the text to the full citation, each link customized for that point in the text. Named-refs try to do this by making a single note (with the full citation) appear in more than one place in the text. But this is entirely one-size-fits-all, with no scope for customization aside from using something like {rp}. (Which I do notice. Superscripted in the text is no place for bibliographical detail.) Second, "author-date" is the conventional form of short cites (for plenty of good reasons I won't go into), and there is little or no reason for inventing our own peculiar form.

That previous version of Phineas Gage was pretty durn bad, but I think not simply for using <ref> tags; those cites and notes were rotten through and through. The current version is much better, but still grates. Having the pagination and other details in the text (like ^{[M]:1,378[M3]:C[3]:1347[4]:56[K2]:abstr}) is just too gruesome to contemplate. If {ran/rma} could satisfy some need where no other alternative existed I might allow it for that particular need, but not generally. And not even in this case: I think a better job could be done with the other tools. Even if one had "many, many refs", each with the same page number, and didn't want long series of notes with "Smith, 1999, p. 5", one could (grudgingly) put that short cite into a named ref. And if one still, really didn't want a long line of back-links, one could still manually construct links (like we have covered above) Even using {harvid} with an "author" of "B1" would be preferable to what I have seen so far with {ran}. ~ J. Johnson (JJ) (talk) 22:32, 13 March 2017 (UTC)

To me, having to click once on then click again on Smith 1995, p. 5 to finally arrive at Smith, J. (1995) "A recent article on the subject" is annoying. I think going from straight to Smith, J. (1995) "A recent article on the subject" makes much more sense, and for the occasional (very occasional) reader who's really studying the material it's helpful to be able to see at a glance that "Ah yes, here's another thing cited to [S] for Smith, got it" without having to click.

Re superscripted in the text is no place for bibliographical detail: Part of what's funny about Wikipedia's "Harvard" system is that the actual Harvard system is short cites directly in the text e.g.

The leading expert, however, disagrees (Smith 1995, p.5).

As I'm always joking, no one at Harvard thinks of it as the "Harvard system", nor is it particularly popular there (wherever "there" is at Harvard). Someone in Europe in the 19C was charmed by the style when he saw it used (instead of the usual footnotes signaled by asterisks and daggers) in a paper by someone at Harvard; thus it's thought of as "the Harvard system" primarily in Europe. Anyway, if an unsuperscripted (Smith 1995, p.5) in the text is acceptable, surely a superscript should be. And as demonstrated somewhere above, if a chain of superscripts is so big it's an eyesore^{[M]:1,378[M3]:C[3]:1347[4]:56[K2]:abstr} it's easily banished to the foot -- see [7]. ran/rma is very flexible in that kind of thing.

But honestly, I think all this worrying about the aesthetics of citations matters a lot to editors but little if at all the readers, as long as it's functional. I believe strongly with Herostratus (Wikipedia_talk:Manual_of_Style/Archive_188#How_to_indicate_which_person_is_which_in_a_caption) that

"Thou shalt not muzzle the ox that treadeth out the corn" (1 Timothy 5:18, paraphrased from Deuteronomy 25:4) which updated means "Let the editor who did the actual work of looking up the refs and writing the friggen thing -- you know, the actual work of the project -- be at least allowed the satisfaction of presenting it as she thinks best, within reasonable constraints".

EEng 23:24, 13 March 2017 (UTC)

That's quite the red-herring, as Wiikpedia doesn't have a "Harvard" system of referencing. The actual Harvard referencing style (or "system"), a form of parenthetical referencing developed by Charles Sidgwick Minot at the Harvard Medical School, should not be confused with the Wikipedia {{Harv}} family of templates. Which may be used, OR NOT, for implementing Harvard referencing. People really, really should get over this MIS-BEGOTTEN NOTION that if you use {{Harv}} you are stuck with parenthetical citations in your text. ~ 22:21, 14 March 2017 (UTC)

It's certainly hard to understand how people could conclude that the {Harv} templates produce Harvard referencing, but in any event the templates' clear, easy-to-digest documentation clears everything up without delay. Since we seem to be at the let-me-gently-correct-your-confusion stage, I beg to inform you that it was certainly not Minot who devised the Harvard system. Rather, Minot wrote a paper (Minot 1896) ascribing the system to E.L. Mark, who seems to have been the first to use it in a monograph (Mark, 1881), though he may have been adapting a system used by Harvard's Museum of Comparative Zoology library (Chernin 1988). EEng 23:29, 14 March 2017 (UTC)

Yes, I struggle to understand why people confound Harv templates with Harvard referencing (something to do with the first four letters?)(and who bothers to read the documentation???). As to the origin, I relied on the Wikipedia article. I possibly and conditionally regret the error. :-(     J. Johnson (JJ) (talk) 08:39, 15 March 2017 (UTC)

I'm fine with letting editors choose their poison, which is why I object to these bot-heads that keep merging "duplicate" notes into named-refs. But what we are discussing here is what style or method an editor might find "best", noting that most editors, having struggled to learn some method, then prefer to stick with what they know rather than learn another. "Best" usually becomes "easiest in the near-term view", which is what one already knows, and considerations of "best" often become just rationalizations of what one has already chosen. Right?

As to superscript ornamentation: the whole point of a note is take something (citations, comments, whatever) off the page and put it somewhere else. Some methods of referencing don't leave any indication that there is something pertinent to be found elsewhere, but our requirement for in-line citation is generally taken to mean that if you don't include either the full citation (!) or a short cite in text (whether "parenthetically", or not), then you must leave a link, which is usually the bracketed and superscripted numbers. But given that the key info of which source – e.g., "Smith 1995" – has been moved off of the page, why retain the piddly secondary details of which page in the so far unidentified source? The main reason seems to be that various methods of referencing (including named-refs) don't have any other way of handling those details. Editors often get used to this, but it's still a wet diaper. ~ J. Johnson (JJ) (talk) 22:30, 14 March 2017 (UTC)

Yes, you have put your finger on exactly why I don't like {{rp}}. It puts the less-important information front and center (the page number) and hides the more-important information (you have to go elsewhere to find out who is being cited). I don't mind the fact that these reference systems hide information, in general — that's what they're for, to avoid cluttering the article text with it — but if you're going to choose to hide some parts and not hide others, you should make that choice more rationally. EEng's system has much the same drawback, unless one gets so familiar with the references by obsessively re-editing Phineas Gage that one memorizes which cryptic initial goes with which source. —David Eppstein (talk) 22:46, 14 March 2017 (UTC)

When you are designing a database one of the key steps is "data normalisation". What that means is removing duplicate information by abstracting fixed information from multiple entries. When citing works there is exactly the same issue: removing the fixed information (ie the bibliographic reference) from the multiple entries (page numbers or locations). It's easy to get hung up on letting the comp[uter do all the work: "just click on the link", but it is easier for printouts if the bibliographic information is in alphabetic order. If citation number 64 is Smith and citation 125 references Jones, where the heck are the details of Jones' magnum opus – before or after Smith? For this reason I would always move the references into a bibliography arranged alphabetically. With computers there is an increasing tendency to use titles as keywords, but the traditional way in a library is to use the author as the key (when I worked in a public reference library the indices were typewritten 5x3 cards). Data normalisation works two ways: there may be multiple references to a particular page range, so show the page ranges in a single location. I say "show" because {{sfn}} does the filing and agglomerating for you. {{r}} may be adequate, if ugly and non-normalised, for simple page numbers but a more complex citation such as <ref>{{harv|Smith|1736|p=28}} cited in {{harv|Jones|2005|loc=Appendix A}}</ref> would be a bit harder! I'm not saying that {{sfn}} is perfect, but it follows good IT practice, is easy to use (a lot easier than learning Wiki's syntax) and is a good analogue of traditional library systems.

References

Bibliography

• Jones (2005), A new survey
• Smith (1736), An old book

- Martin of Sheffield (talk) 23:12, 14 March 2017 (UTC)

You're confusing the conceptual schema, which in general should be normalized, with the external schemata, which often are not. (Long ago in a universe far, far away, Phil Bernstein was my advisor. Thus I outrank you on this.) EEng 23:44, 14 March 2017 (UTC)

I used 3x5 cards, but perhaps that's like the difference between square holes or round holes on your punch cards? Anyway, I'm glad we all share some basic grasp of data normalization. I certainly agree with moving "references" (i.e., citations) into a separate section; in many articles it's undoubtedly the biggest improvement to be made in regard of citation.

As I've said before, I used to be tolerant of {sfn}. But then I realized it wants to make named-refs, for which I hate. Better to do the whole <ref>{{harv|xxxx}}</ref> enchilada than {{sfn|xxxx}}. ~ J. Johnson (JJ) (talk) 08:43, 15 March 2017 (UTC)

Break 2

[A little experiment, to see if this "Side-trip into 'blue'" would be well-handled as a side-bar. Okay to revert. ~ J. Johnson (JJ) (talk)]

EEng, I'm in two minds about this. It's nice to have the ref highlighted more, but it's more fiddly. I'd like to keep fiddling to a minimum, while still being able to point to the long ref. SarahSV ^(talk) 00:07, 14 March 2017 (UTC)

I lost my mind some years ago so if you have two I hope you can see your way clear to lending me one. When you click on e.g. Smith 1995, p. 5 it jumps down to the right bit of the page, but the problem is that when there are multiple columns of bibliography (on my screen there are three columns) you have to kind of hunt around to find which is the entry to which you're being sent to. Which you're sent being. To. Which. When it lights up it's immediately obvious. EEng 00:37, 14 March 2017 (UTC)

Sorry, I have to hang onto every bit of mind I have left. If would be nice if the blinking thing were bigger. You're right: you do momentarily have to hunt for the blink. SarahSV ^(talk) 00:43, 14 March 2017 (UTC)

The "lighting up" effect (in pale blue) is found not just with {{wikicite}}, but also with {{citation}}, {{cite book}} and the rest of them. You will also see it appear on the short cite in pages that use {{sfn}} or <ref>{{harvnb}}</ref>, when clicking from the superscripted numbers in the text. It works with most browsers; however, in Internet Explorer, it doesn't highlight and there's nothing we can do about that. --Redrose64 🌹 (talk) 00:47, 14 March 2017 (UTC)

Redrose, it's harvnb we're discussing, used with manual long refs. That leaves a blinking vertical line at the long ref, but it doesn't light it up. See Bohlke, L. Brent, the second ref in Works cited, for an example. SarahSV ^(talk) 01:09, 14 March 2017 (UTC)

I see no "blinking" - if I did then I would have serious concerns, as this could be an accessibility violation.

At the version of 23:32, 13 March 2017 the long ref is no different from normal text in its appearance, whereas at the version of 23:36, 13 March 2017 the long ref has a pale blue background. The technical details of the pale blue background are unimportant (it is an effect of using the :target pseudo-class (some information here if you really want it) in conjunction with some other CSS) but the thing to understand is that the pale blue background extends only to the text enclosed by the target element.

With the later version, of 23:36, 13 March 2017, the Wikimarkup is

{{wikicite| ref=CITEREFBohlke1982 | reference = Bohlke, L. Brent. "Willa Cather and The Life of Mary Baker G. Eddy", ''American Literature'', 54(2), May 1982, 288–294. {{jstor|2926137}} }}

for which the emitted HTML is

<span class="citation wikicite" id="CITEREFBohlke1982">Bohlke, L. Brent. "Willa Cather and The Life of Mary Baker G. Eddy", <i>American Literature</i>, 54(2), May 1982, 288–294. <a href="/wiki/JSTOR" title="JSTOR">JSTOR</a>&#160;<a rel="nofollow" class="external text" href="https://www.jstor.org/stable/2926137">2926137</a></span>

Notice here that the whole of the long ref is enclosed in a span element, and importantly, that this span element has (amongst others) the attribute id="CITEREFBohlke1982" - this means that when you click a link that leads to the anchor named CITEREFBohlke1982, the target is the whole of the long ref, from Bohlke, L. Brent all the way through to that JSTOR link. By contrast, the earlier version, of 23:32, 13 March 2017, the Wikimarkup is

{{anchor|CITEREFBohlke1982}}Bohlke, L. Brent. "Willa Cather and The Life of Mary Baker G. Eddy", ''American Literature'', 54(2), May 1982, 288–294. {{jstor|2926137}}

for which the emitted HTML is

<span id="CITEREFBohlke1982"></span>Bohlke, L. Brent. "Willa Cather and The Life of Mary Baker G. Eddy", <i>American Literature</i>, 54(2), May 1982, 288–294. <a href="/wiki/JSTOR" title="JSTOR">JSTOR</a>&#160;<a rel="nofollow" class="external text" href="https://www.jstor.org/stable/2926137">2926137</a>

Again, there is a span element with the attribute id="CITEREFBohlke1982" but the significant difference is the position of the closing </span> tag - it is before the start of the long ref, so none of that long ref is enclosed - in fact, nothing is enclosed, so when you click a link that leads to the anchor named CITEREFBohlke1982, the target is an empty string - there is nothing there to which a pale blue background could be given.

This is why for this specific case, {{wikicite}} is better than {{anchor}}. --Redrose64 🌹 (talk) 08:36, 14 March 2017 (UTC)

The <span>...</span> tags should really be <cite>...</cite> tags. I have changed {{wikicite}}.

—Trappist the monk (talk) 09:29, 14 March 2017 (UTC)

Break 3

J. Johnson, I've started slowly converting the Eddy article to see whether I could work with harvnb and anchors. How do you produce the short refs? My brain is so used to writing <ref>Fraser 1999, 1.</ref> that it's there in the blink of an eye; {{harvnb|Fraser|1999|loc=1}} is a big change. Do you have a quick way of doing it? SarahSV ^(talk) 23:45, 14 March 2017 (UTC)

For starters, let's be clear on terminology, as confused terminology is a large part of why citation, and talking about citation, is so painful. I do not recognize "long refs" or "short refs". "Refs" undoubtedly is intended as short for "reference", but that term is so broadly and ambiguously used we are better off to avoid using it. "Short cite", or "short citation", is preferred for describing something like "Fraser 1999". A key point is that the <ref> tags are not part of the citation.

So what you are trying to do is change stuff like <ref>Fraser 1999, 1.</ref> to <ref>{{harvnb|Fraser|1999|loc=1}}.</ref>. Now if you are adept in an editor with a search-and-replace function you could use that to make some of the change. (E.g.: "<ref>" → "<ref>{{Harvnb".) In some cases I also write scripts to do this. But in this case (aside from whatever you can do with S&R) I think it would be easiest to just type it in. If you want, I can do some of that for you. I take it that for now you are not going to put the full citations into {cite} or {citation} templates, so wrapping them in {wikicite} seems reasonable. ~ J. Johnson (JJ) (talk) 09:06, 15 March 2017 (UTC)

J. Johnson, thanks for this. I was wondering how you do it when you start an article, rather than convert it. Do you type in <ref>{{harvnb|Fraser|1999|loc=1}}</ref> each time?

Please don't go to any trouble at the Eddy article, as I'm not sure yet whether to keep it. It's an article I want to develop, and I don't know whether I'd want to type all that in each time. It also makes it harder (for now) to find things, because I'm so used to searching for "Fraser 1999". SarahSV ^(talk) 22:10, 15 March 2017 (UTC)

For you, it would be an honor and a delight, no trouble at all!

When preparing text (off-line) I usually type something like "<<H|Smith|1999|p=1>>" (I am quite used to typing "Smith|Jones|Brown|2003" with the vertical bars instead of spaces). When I'm ready to upload I use my editor's Search&Replace function to globally replace every "<<H" with "<ref>{{Harvnb", and ">>" with "}}.</ref>". I could get fancier, but this works for me. ~ J. Johnson (JJ) (talk) 23:19, 15 March 2017 (UTC)

P.S. I forgot to mention another way I go about this. When I put together a longer article I sometimes have a file with a suitable Harv template for each source. As I write the article it is a simple matter to just copy in the template. ~ J. Johnson (JJ) (talk) 20:38, 18 March 2017 (UTC)

J. Johnson, sorry, I didn't save my reply to your earlier one (wanted to edit it, but must have closed the tab and forgot I hadn't saved). Copying in a template makes sense. The more I look at this, though, I wonder whether the extra work is worth the effort. SarahSV ^(talk) 20:50, 18 March 2017 (UTC)

I have a spreadsheet on my computer, it has one row for each of my books that I have used as a source at least once. There are columns for first name, last name, title, date, publisher, ISBN and so on. At the start of each row there are two columns containing formulae based upon the other cells in that row - the first emits a {{cite book}}, the second a {{harvnb}}, and both are fully-constructed except that the |page= or |p= parameters have no values. Having written some text into a Wikipedia article, I find the row in my spreadsheet that matches the source that I am using, copy the cell from the first or second, paste it into the appropriate place (possibly also using the <ref></ref> link in the "Insert" thing below the edit box, or possibly altering harvnb to sfn), fill in the page number and save. --Redrose64 🌹 (talk) 10:16, 19 March 2017 (UTC)

I am impressed. It is essentially what I do (more or less, at least some of the time!), but better organized. I especially like having the "formulas" set up to produce harv or cite templates. Well, if you produced {citation} templates you would be covered for all sources. Or I suppose you could have a field that indicates type of source, to get the appropriate {cite} template. Something I have found useful: when I find a potential source, I add codes for particular material I might be interested in, so I can find it later.

This does lead up to the several forms of bibliographic software. I looked at several packages some years ago, but didn't like the overhead, or the citations they produced. But probably a viable option for many editors.

Sarah: all of these approaches involve extra work; the question is really what do we get for that work. If the controlling criterion is the least work that passes WP:V, then straight text with no templates, no hyperlinks, etc., would seem the winner. But then you're always futzing with the formatting of the citations (though some folks seem to think that is a plus), and as soon as you use a source more than once there are all sorts of complications. And not having hyperlinks (implicit in hypertext, which is the initial letter in HTML and HTTP) seems lame, even sub-standard. Having the full citations in templates (for ease of formatting as well for meta-data purposes) and in their own section looks to be what citation practice is evolving towards, with short cites in the text. Given that, the remaining issues are the nature of the short cites (e.g., whether "Smith 2014", "SMIT14", "S14a", ":0", "NASA-20151112-ab", "Blue", etc.) and how to implement the links. I happen to favor the {harv} family of templates, but allow that explicit links are reasonable. When the criterion applied is a high standard of quality, the little extra work of doing citation right is a minor matter. And much less than the work of straightening out badly done citations. ~ J. Johnson (JJ) (talk) 22:20, 19 March 2017 (UTC)

I'm also impressed. I tend to edit from different computers, so imitating (the "sincerest form of flattery") exactly wouldn't be appropriate. I have just set up a sortable table in a subpage into which I can put the citations, and then find the blighters when I want. I have spent ages in the past searching for a reference I know I've done, and just want to "save time" copying. (oops, missing sig) Martin of Sheffield (talk) 23:13, 19 March 2017 (UTC)

There's an interesting idea I missed: keeping them on a Wikipedia subpage (presumably under your user page). I'm not certain how I would organize that – text? a simple (??) sort table, or what. Would you have sample we might examine? ~ J. Johnson (JJ) (talk) 04:59, 20 March 2017 (UTC)

I did that early on, but it didn't last long. My last edit of that nature was more than seven years ago. --Redrose64 🌹 (talk) 07:51, 20 March 2017 (UTC)

As I said I have only just set it up, it's still experimental but feel free to have a look here. I changed it only this morning after seeing RedRose's work from 7 years ago, adding the third column. I'm not sure about the template, I may add it, or more likely just list it as an aide memoir. Martin of Sheffield (talk) 09:42, 20 March 2017 (UTC)

I'm impressed and depressed, because I do none of these things. I'm full of admiration that you're all so organized. SarahSV ^(talk) 23:19, 19 March 2017 (UTC)

Ah, don't be depressed. None of us do as much as we could, and most of us just get on doing the most immediate tasks from day to day, with what we have. Just start doing a few simple improvements, get used to them, the try some more. I'm sure that eventually your methods would be as impressive as Master Martin's. ~ J. Johnson (JJ) (talk) 05:22, 20 March 2017 (UTC)

Too much honour you do me, I think. :-) Yodo aka Martin of Sheffield (talk) 09:42, 20 March 2017 (UTC)

Combining

Hello. Is {{sfn}} the only template we have which can automatically combine redundant entries? Here is an example article, Continental Motors Company where I hoped to use efn instead of refn (sfn works but results in unwanted hyperlinks, and ends up in the wrong list at the bottom). So I wondered if we have another combining template, but could not find it in the documentation. Thanks, —░]PaleoNeonate█ ⏎ ?ERROR░ 05:05, 1 May 2017 (UTC)

I am not at all sure what it is that you are asking. Continental Motors Company does not have any recent edits by you that would give a clue to what you tried and did not like.—Trappist the monk (talk) 10:04, 1 May 2017 (UTC)

I only have tested changes in preview mode so have not edited that article yet, indeed. I'm no longer convinced that my idea of a combining notes template is ideally what's needed for that article. However, I still scribbled an idea of what results a hypothetical {{efnc}} may be able to produce, near the bottom of Special:Permalink/778167597. —░]PaleoNeonate█ ⏎ ?ERROR░ 15:08, 1 May 2017 (UTC)

What does this do that the |name= parameter of {{efn}} does not do? For an example of |name= in use, see List of The Goon Show episodes#Vintage Goons series (1957–58), the two in the last column of the table. --Redrose64 🌹 (talk) 15:47, 1 May 2017 (UTC)

If two different elements share the same name, the multiple entries would be merged instead of being rejected. The name would also not need to be filled explicitely (it would be autogenerated from the first non-text field(s)), and duplicate entries+content would be automatically merged as for sfn. I realize that they would be heavier to process than simpler less automatic references, though. It's just an idea, I have no actual need for it at current time. I was impressed by what sfn can do but found no equivalent for non-hyperlinked alpha-indexed text notes in terms of internal automatic processing. —░]PaleoNeonate█ ⏎ ?ERROR░ 16:21, 1 May 2017 (UTC)

It is not clear what you want. I suspect your concepts are tangled (no surprise, it's endemic on this topic), and your terminology is, ah, curious? E.g., what precisely do you mean by "elements"? Or "alpha-indexed text notes"? And, do you understand that {{sfn}} is basically a way of incorporating {{harvnb}} templates into <ref>...</ref> tags? I don't believe there is anything {sfn} does that can be done using those components. If you want to explore this some more you might copy some article text into a sandbox, and then we could play around with different formulations. ~ J. Johnson (JJ) (talk) 23:05, 1 May 2017 (UTC)

Ideally, put the proposed variant template at Template:Efn/sandbox, and set up some demonstrations at Template:Efn/testcases - more at WP:TESTCASES. --Redrose64 🌹 (talk) 23:26, 1 May 2017 (UTC)

Before setting up some variant template he should figure what the supposed problem is. If he sets up a sandbox (perhaps under his user space) with some dummy text he could demonstrate what he thinks the problem is, and we could try some solutions using the existing tools. Before making new tools we really should check 1) what the problem is (for real, not just some hazy conception of a problem), and 2) that that the existing tools are inadequate. ~ J. Johnson (JJ) (talk) 00:14, 2 May 2017 (UTC)

I don't think there's a problem, it would only be convenience features. It is even possible that it's not worth implementing considering that text explanatory footnotes are less commonly used than references. I initially hoped that someone would point me at an existing template which I was ignorant about, but when I realized that it may not exist, I also realized that I may be wasting everyone's time. If so, my apologies.
I will answer the questions for clarification: by alpha-indexed, I meant in the "lower-alpha" (i.e. a-z, etc) default group-type (like efn). By combining, what I meant was that sfn can merge redundant footnotes automatically rather than needing to be deduplicated/merged manually/explicitly via the name= reference tag (without needing to care much about using unique reference tags for every variant, they are generated and matched by the common template parameters supplied). Yes I understand that sfn is for harvard-style citations (and I think that sfn makes using those very user-friendly, versus more basic ref, refn, harvnb, etc). My constatation was that efn seems to be the counterpart to sfn, but that it seemed closer to ref/refn than to sfn in terms of user-friendliness. In a way, sfn is more featureful than most footnote templates (and has its custom lua implementation). —░]PaleoNeonate█ ⏎ ?ERROR░ 09:50, 2 May 2017 (UTC)

You have several concepts mashed together that confuses things. Note first that a note – subdivided into footnotes and endnotes — such as created with the <ref>...</ref> tags, is just .. a note. Which is to say, any kind of text that explains or augments something in the main text, but was considered too tangential, or just not important enough, to be included in the main text itself. Such usage also includes citation of sources, and often comments on the sources. Key point: in all conventional citation practice there is no inherent distinction between "explanatory" (it's all "explanatory"!) or "citatory" uses. It is only at WP we have this distinction, which is based on nothing more than long-running confusion on how to do these things.

Second point: you seem a little vague on the distinction between "redundant" and "duplicate". The former implies more than is needed, and duplication is often taken as redundancy, but that is not necessarily the case. In the case of full citations, yes, standard practice is one per source per article. The problem is where editors think that full citation must be in the note, and they need to cite that source in more than place. This led to the development of "named refs" (the "<ref name=...>" form), where a single note is linked from multiple places in the text. The problem with such multiple uses is they can't be customized for specific uses: one "ref" has to fit all occasions. Which seems to lead many editors into the bad practice of omitting in-source specifiers like page numbers, to avoid having nearly duplicate notes that duplicate a full citation, but differ in the in-source location.

So while duplication of full citations is to be avoided, this confounding of full citations with notes (where they are most frequently placed) has led to this mistaken notion that "duplicate" notes should be suppressed (or "de-duplicated"!), which then leads to various problems. What you consider a convenience feature I see as coping with very poor practices. As I said before, if you want to throw some sample text into a sandbox I'd be happy to demonstrate what I think are better, and even more convenient, practices. ~ J. Johnson (JJ) (talk) 21:54, 2 May 2017 (UTC)

I appreciate the terminology clarification. Thanks, — PaleoNeonate — 03:09, 8 May 2017 (UTC)

First name, last name

J. Johnson, and anyone else who might know, I have a question about the order of names.

I've used one citation template in an article that is otherwise written with manual citations. I've done that because I want to use sfn with one citation that is repeated throughout in short form. But that means I have one citation that says "Smith, John" (because sfn needs that), while the manual citations say "John Smith" (which is my preference).

Is there a way to make sfn work with "John Smith" in the long cite? Alternatively, is there a way to write the long cite so that it says "first name=John; last name=Smith; display name=John Smith"?

SarahSV ^(talk) 16:25, 12 May 2017 (UTC)

Write the cs1 template |author=John Smith and |ref={{sfnref|Smith|2017}}. The first two {{sfnref}} parameters must match the first two {{sfn}} parameters.

—Trappist the monk (talk) 16:33, 12 May 2017 (UTC)

Trappist the monk, thank you! SarahSV ^(talk) 16:39, 12 May 2017 (UTC)

Spoke too soon. Can't get it to work. Instead of ref=harv, should I write "ref=sfnref|Smith|2017" or "ref={{sfnref|Smith|2017}}"? I've tried both. SarahSV ^(talk) 16:47, 12 May 2017 (UTC)

Here is a sentence that ends with a {{sfn|Smith|2017}} template. Clicking the superscript will take you to the {{reflist-talk}} box below. Clicking the link in the {{reflist-talk}} box brings you this citation:

John Smith (2017). Title.

written like this:

{{cite book |title=Title |author=John Smith |date=2017 |ref={{sfnref|Smith|2017}}}}

References

—Trappist the monk (talk) 17:21, 12 May 2017 (UTC)

Thank you, it worked! SarahSV ^(talk) 17:46, 12 May 2017 (UTC)

|Author= really shouldn't be used like that, as it confuses what the key is for sorting and indexing the author (and citation), and corrupts the metadata. One does not have to be an advocate of COinS to understand the benefit of labeling what a datum is. E.g., is an author "Marvin Lee", or "Lee Marvin"? The original editor presumably has that information at hand, and if s/he takes the effort to put it into separate named parameters we have some confidence in the result. With something like "John Smith" we rely on our knowledge of the two names to determine which is the personal name and which the surname. But I know someone whose last name is "John", and a long time ago knew someone with a first name of "Smit". This suggests "Smit John" as proper name, which would undoubtedly prompt quite a few editors to "fix".

When full citations are placed at the bottom (foot) of a page, I believe it has been quite acceptable to have the lead author's name in "normal" order. However, when full citations are collected in lists it greatly helps finding them and keeping them straight if they are ordered (sorted). Which is usually by the "last" name (surname or family name) of the lead author. This isn't just editorial preference, it corresponds to the universal practice of how authors are collated: by surname. (Which is probably why CS1/2 don't have a "normal order" option.)

BTW, your attempt with "ref=sfnref|Smith|2017" was close. Try it with "ref=CITEREFSmith2017". I.e.,with a CITEREF prefix, and the parameters concatenated without spaces or other characters. ~ J. Johnson (JJ) (talk) 23:54, 12 May 2017 (UTC)

In works-cited lists at the end, it makes sense to order alphabetically by last name, first name. But when long cites are within a text, it makes no sense to do that. If for other reasons we need to know "first= and last=", it would be good to have a "display=" to override it. SarahSV ^(talk) 00:11, 13 May 2017 (UTC)

I disagree that there is "no sense" in using last-first order in isolated cases (for various reasons), but I do agree it is acceptable to use first-last order in such isolated cases, such as at the foot of a printed page. But I also disagree that full citations should be "within the text", or even in notes (foot or otherwise). Such use clutters the working text, makes them hard to find, and hard to manage consistently. Even so, it might be better to talk Trappist into adding a |author-display= parameter, or perhaps a "first-last" option, than to misuse |author=. ~ J. Johnson (JJ) (talk) 20:00, 13 May 2017 (UTC)

Re: "But I ... disagree that full citations should be 'within the text'", it's usually unavoidable, because drive-by editors add full citations within the text. Unless someone is willing to stand guard and move them all, it's better to go with the flow. It's only on articles where few editors are anticipated, or where an article has FA status and people have come to respect the style it uses, that we can easily maintain a different citation system. SarahSV ^(talk) 20:15, 13 May 2017 (UTC)

Certainly not avoidable, for the reason you cite, but also because we don't enforce CITVAR in that respect, and we don't even set a good example. Editors do things in crappy ways because they come in, look around, and that's what they see other editors doing, so they conform. Crap is the standard. But it is correctable, even on a large scale. I have done it on several global warming articles, and on smaller articles. I would like to write some tools for that someday, but too much other stuff to do. ~ J. Johnson (JJ) (talk) 20:49, 13 May 2017 (UTC)

I agree regarding CITEVAR. Yes, it's correctable, but as you suggest it's the time factor. And people will keep arriving to mess it up. SarahSV ^(talk) 20:56, 13 May 2017 (UTC)

The approach above is fine if there is a need to append variable additional information to each short note, such as different page numbers which support different claims. But if the citation is to be exactly the same each time the source is cited, one could write the first <ref parameter like <ref name="Smith2017">.....</ref> and subsequent cites as <ref name="Smith2017"/>. Jc3s5h (talk) 00:15, 13 May 2017 (UTC)

Yes, we know about named-refs. And subject to the problems I have mentioned before. Perhaps the biggest objection is the implicit disparagement in your "if there is a need" for insource specification. "Need" is a funny word here, as we don't actually require insource specification. But to cite a (say) 600-page book without providing a page number is rather a mockery of WP:V. Perhaps within the "letter" of WP:V, but hardly its spirit. ~ J. Johnson (JJ) (talk) 20:03, 13 May 2017 (UTC)

There is an explicit obligation for in-source location in WP:V#Responsibility for providing citations: "Cite the source clearly and precisely (specifying page, section, or such divisions as may be appropriate)." – Finnusertop (talk ⋅ contribs) 01:43, 14 May 2017 (UTC)

@Jc3s5h: The point about {{sfn}} is that you don't need to name references, whether with <ref name=...>...</ref> or otherwise; it's done automatically. --Redrose64 🌹 (talk) 12:00, 14 May 2017 (UTC)

I find it immaterial whether you add named-refs manually, or use a template that does it for you. The bad results are same. Indeed, some new editor punching in a lot of new material with {sfn} but as yet inexperienced in working with existing text might believe they are not working with named-refs. ~ J. Johnson (JJ) (talk) 21:45, 16 May 2017 (UTC)

support for author names without year

Because of a discussion at Template talk:Citation, I have added code to the Module:Footnotes/sandbox to support citations that don't have years:

{{sfn/sandbox|One}}

{{sfn/sandbox|One|Two}}

{{sfn/sandbox|One|Two|Three}}

{{sfn/sandbox|One|Two|Three|Four}}

{{sfn/sandbox|One|Two|Three|Four|Five}}

References

• One, Title1
• One; Two, Title2
• One; Two; Three, Title3
• One; Two; Three; Four, Title4
• One; Two; Three; Four; Five, Title5

In October 2016 I added this support to two names without year and left myself a TODO to fix the rest. I think that I have done this now.

—Trappist the monk (talk) 12:02, 19 June 2017 (UTC)

Thank you. I was surprised to see the header, as I've been using (not voluntarily) {{sfn}} without year for some time, without any problems. It seems they must all have been two or fewer authors! --NSH001 (talk) 12:33, 19 June 2017 (UTC)

The fix only changes the rendering of the short form citation. Compare the sandbox above to the live version here:

{{sfn|One}}

{{sfn|One|Two}}

{{sfn|One|Two|Three}}

{{sfn|One|Two|Three|Four}}

{{sfn|One|Two|Three|Four|Five}}

References

• One, Title1
• One; Two, Title2
• One; Two; Three, Title3
• One; Two; Three; Four, Title4
• One; Two; Three; Four; Five, Title5

—Trappist the monk (talk) 13:29, 19 June 2017 (UTC)

Thanks Trappist. I've been caught with this problem in the past and it was annoying. I assume the sandbox code will move into the main template at some point? BTW, will the new code cope with {{sfn|Bede|721}}? Regards, Martin of Sheffield (talk) 12:42, 19 June 2017 (UTC)

Yes, three- and four-digit years, with or without disambiguators.

{{sfn/sandbox|Bede|721}}

{{sfn/sandbox|Bede|721a}}

References

• Bede (721), Title
• Bede (721a), Title

—Trappist the monk (talk) 13:29, 19 June 2017 (UTC)

Thanks and well done. Martin of Sheffield (talk) 14:29, 19 June 2017 (UTC)

And then there was distraction. These fixes now made to the live module.

—Trappist the monk (talk) 09:31, 11 July 2017 (UTC)

In my view making sure the documentation and the code are in agreement is as important as other development task as writing the code. Help:Citation Style 1#CS1 compliance with Wikipedia's Manual of Style states that years greater than or equal to 100 are supported. Is this how the logic of the change is written? Jc3s5h (talk) 12:15, 11 July 2017 (UTC)

Yes.

—Trappist the monk (talk) 12:31, 11 July 2017 (UTC)

Query re: loose ampersand using sfn and harvid

G'day all, for some years, I've been using sfn like {{sfn|The Register|14 May 1920}} for example, with a cite news field formatted like ref={{harvid|The Register|14 May 1920}} to cite newspapers and the like. So far as I can remember, it used to just appear as "The Register 14 May 1920" in the footnotes, but now it is appearing as "The Register & 14 May 1920". Any ideas how to switch off the loose ampersand? For a practical demonstration, see Bill McCann or Raymond Leane. Thanks. Peacemaker67 (click to talk to me) 04:53, 12 July 2017 (UTC)

@Peacemaker67: Use the year alone, not the full date, in both {{sfn}} and {{harvid}}. --Redrose64 🌹 (talk) 08:44, 12 July 2017 (UTC)

I'm pretty certain that wasn't the case before. So if I want to cite the same paper twice from the same year, I need to use 1923a, 1923b etc? Peacemaker67 (click to talk to me) 08:49, 12 July 2017 (UTC)

This may be related to #support for author names without year above. --Redrose64 🌹 (talk) 09:23, 12 July 2017 (UTC)

It did seem to be a recent change, which might explain it. Perhaps Trappist the monk can shed some light? The recent change appears to make using full dates with harvid not work properly. It is a pain in the proverbial to have to go back and change properly formatted dates to 1932a 1923b etc, especially when building an article, but more generally as well. I have a dozen or so FAs that are affected, let alone GAs. They should be able to stand alone with full dates as before. Peacemaker67 (click to talk to me) 09:37, 12 July 2017 (UTC)

The template expects the last unnamed parameter to be a 3- or 4-digit year, a 3- or 4-digit circa year (c. 1920), or either of the two no-date abbreviations 'nd' or 'n.d.'; all of these with or without the optional lowercase-letter disambiguator.

Isn't it the purpose of the {{Shortened footnote}} template to make short footnotes? It appears to me that the template's documentation has never suggested that use of a full date is a proper use of this template. Isn't the full date a bibliographic detail that correctly belongs in one place: the full citation? Is it really necessary or appropriate to fully duplicate bibliographic details in the short footnote?

Off topic: when looking at the code of your {{sfn}} example and then Bill McCann, and Raymond Leane, and then later, Bill Denny, the thing I noticed first, before I noticed the ampersand, was that the newspaper names are not italicized. Oughtn't they be?

—Trappist the monk (talk) 11:01, 12 July 2017 (UTC)

You're right about the italicisation, I'll go back and fix it, but my point is that when I am citing two editions of the same newspaper in a given year (for example), I now cannot use the dates to differentiate them in the shortened footnote, I have to differentiate them by putting an a or b after the year only, which becomes a problem when I add another citation from the same year from the same paper and want to maintain chronology with the a, b etc, as I have to go back. Shortened footnotes for books include the author, year and page number, so there is no real difference in the amount of information included with newspaper, date and page number, as they are the relevant parameters for a newspaper. I don't see why this functionality, which has been in place for years, has been lost. To what benefit? I am not the only editor who has been using the newspaper name and date of issue with a harvid template. harvid is also used to do sfn for news websites, as seen at Kragujevac massacre fn 77, but the ampersand doesn't appear there for some reason. Peacemaker67 (click to talk to me) 01:45, 13 July 2017 (UTC)

Without having expended any significant cycles on this from a design perspective, my intuitive take is that for this case using the full date makes sense. A typical short citation to a newspaper article (a review of a theatre performance, say) without a byline is {{sfn|The Guardian|13 July 2017}} or, with a byline, {{sfn|Billington|13 July 2017}}. You could of course do 2017a and so forth, but it seems needlessly obtuse and error prone, and, anecdotally, an unusual way to format it. Including the page numbers in the short citation appears wrong though; for citing a newspaper article, journal article, or book chapter these belong in the full citation. --Xover (talk) 07:25, 13 July 2017 (UTC)

I think the page numbers are needed in most instances (although perhaps not for newspapers, as you say), as even book chapters and journal articles generally use a page range in the full citation (covering the full page range of the article or chapter), and the sfn should specify which page the cited information is drawn from. Peacemaker67 (click to talk to me) 07:45, 13 July 2017 (UTC)

Actually, I just had an instance of needing separate citations (and links) to different pages of the same newspaper, so there are situations where page numbers are needed. Peacemaker67 (click to talk to me) 00:37, 14 July 2017 (UTC)

Going back through my FAs, I can see that I have been using the full date functionality of this template since 2012, and I know I am not the only one who has been using it, as it has been used on the FAs Harry S. Truman and Corona Australis, which I've never edited. I've made mention of this change on the talk pages of both articles to get the views of others who have been using it in the same way. I really think this change should be rolled back and re-examined, as it has had unintended consequences. Peacemaker67 (click to talk to me) 02:28, 14 July 2017 (UTC)

@Xover: I always include page numbers for newspaper refs (I know of no newspaper larger than a single folded sheet which doesn't number its pages). If verifying somebody else's referenced edit, I don't want to have to flip through the whole newspaper to find one item. --Redrose64 🌹 (talk) 08:52, 14 July 2017 (UTC)

@Redrose64: My apologies; I was a bit quick above. I meant to refer to the full page range for a chapter in a book, or a journal article, and by extension the same for a newspaper article; all of which belong in the full citation to which the short footnote refers. However, as you and Peacemaker67 point out, the specific page number within that range, when the range is too large (however you define "too large"), does obviously belong in the short footnote. --Xover (talk) 09:02, 14 July 2017 (UTC)

As the one who amended the template, Trappist the monk needs to respond here, so we know where to go next. This is a very widely used template, and the recent change has reduced important functionality for marginal benefit in a very specific instance. A rollback is appropriate at this point, in my view. Peacemaker67 (click to talk to me) 09:50, 15 July 2017 (UTC)

@Peacemaker67: I disagree. The change made was for a specific, desired, purpose and only as a side effect broke the behaviour you relied on; and that behaviour was both undocumented and not intentional. I have argued above that this functionality should be supported, but that's quite different from rolling back the existing code. --Xover (talk) 10:42, 15 July 2017 (UTC)

Frankly, any consensus that this change relied upon was amongst a very small group of editors who watch this page and work on template code, not among the wider content creators affected by it. Where do I go to escalate this, then? Peacemaker67 (click to talk to me) 10:47, 15 July 2017 (UTC)

You may recall that I have responded to this discussion. I have relatively little to add to what I have already written. Before the recent change, the template was broken. Now it isn't. Because it is not broken, there can be no reason to revert.

You might do this: {{sfn|''The Register'' 7 July 1923|p=6}} and |ref={{sfnref|The Register 7 July 1923}}

"Soldiers' Children's Education". The Register. Vol. LXXXVIII, no. 25, 749. South Australia. 7 July 1923.

References

Still a misuse of {{sfn}} so I do not advocate this method, but it does do what you want.

—Trappist the monk (talk) 13:09, 15 July 2017 (UTC)

Trappist, we do need to be able to cite newspapers using the full date, rather than Smith 2017, for the obvious reason that journalists publish more often than do the authors of journal articles and books. Using year alone would lead to too much of Smith 2017(a), Smith 2017(b), Smith 2017(c), and so on. SarahSV ^(talk) 15:38, 15 July 2017 (UTC)

Are there really enough page references to that single news article to use shortened footnotes? Why not use {{cite news}}? Another possibility, if long and short notes are both inadequate, would be the use of {{rp}} suffixes... —PaleoNeonate - 15:45, 15 July 2017 (UTC)

Please, never use {{rp}} in conjunction with Shortened footnotes. --Redrose64 🌹 (talk) 23:04, 15 July 2017 (UTC)

I reject the suggestion that this is some sort of "misuse" of sfn. It is a functionality (that is of benefit to the encyclopaedia and to content creators) that was in place (whether documented or not, it was being used, I found two FAs (not mine) that use it, very quickly). It is a classic example of IAR. It worked just fine, but the functionality has been broken by this change. What is the benefit of the change? My understanding is that this change was made to accommodate sources that don't have a year. What sources don't even have a year of publication? Peacemaker67 (click to talk to me) 01:15, 16 July 2017 (UTC)

And, in what way was the template "broken" before this change? Peacemaker67 (click to talk to me) 02:26, 16 July 2017 (UTC)

An {{sfn}} template that had three authors but no date written like this {{sfn|First|Second|Third}} produced an output that looked like this:

<ref name="FOOTNOTEFirstSecondThird">[[#CITEREFFirstSecondThird|First & Second, Third]].</ref>

There is a 'work-around' described in the template documentation. I suspect that that section of the documentation can (should) be rewritten to remove the 'work-around' and encourage the use of {{sfn|...|nd}} or {{sfn|...|n.d.}} with matching values in the cs1|2 |date= parameter.

—Trappist the monk (talk) 09:37, 16 July 2017 (UTC)

Maybe I'm being thick here, but a situation where there are more than two authors, but actually no year of publication, is, in my experience, about as rare as rocking horse shit. Even the example used to justify the change in the first place was on a page with a WWF copyright 2017 tag at the bottom, which could have been used as the year when citing it. However, the scenarios for which I (and others) have been using sfn and harvid is a common garden variety citation of a newspaper (or news website) that should be supported by sfn, even if it was not formally documented, it was being used that way in practice, as I have demonstrated. In terms of the common good, this change has benefited an incredibly small number, but disadvantaged many. I should make clear that I'm not having a crack at anyone here, god knows I couldn't find my own arse with both hands when it comes to writing the code that supports this template, I just think, with these changes, the downside outweighs the upside. Peacemaker67 (click to talk to me) 09:55, 16 July 2017 (UTC)

On the basis that several of the editors here have expressed support for the idea that this functionality is desirable, can restore it? Cheers, Peacemaker67 (click to talk to me) 04:07, 18 July 2017 (UTC)

@Peacemaker67: you've shown that {{sfn}} has been used with a full date, but not – to my satisfaction at least – that it needs to be used in this way. The point of {{sfn}} is to handle cases where different parts of the same source are used repeatedly, usually different pages of a book, so that without this template, the reference for the source would need to be repeated. At Harry S. Truman, for example, the short footnotes for Life and Time have no page numbers and lead to single article, so there's no point in using {{sfn}}. Can you point me to examples where {{sfn}} with a full date is really necessary? Peter coxhead (talk) 05:55, 18 July 2017 (UTC)

If we want to maintain consistency of the citation template used in any given article, any case where a newspaper article is being used as a reference and there is more than one use of that newspaper in a given year. For example, in the Raymond Leane article I'm writing at present, there is a particular incident that occurred in September 1928, several newspapers are cited several times each. If the suggestion of using 1928a, 1928b is followed (which frankly is pretty obtuse, and not useful to the reader or editor), each time a new citation is created to that newspaper in that year, the a, b etc may need to be changed in order to maintain chronology. What I don't understand is, the functionality existed, it was useful, but there is some problem with restoring it because it doesn't meet some narrow definition of what sfn is meant to be for. The example used to justify this change wasn't even of the type you describe. Peacemaker67 (click to talk to me) 06:12, 18 July 2017 (UTC)

@Peacemaker67: sorry, but I still don't understand. If each edition of the newspaper is cited once and once only you don't need {{sfn}} – it's only needed if two conditions are met: (1) the same date is used more than once (2) different pages are cited for the same date. (I can't speak for Trappist the monk, but it's not as simple to fix as you seem to think. It's not a simple matter of reversion because the previous behaviour was wrong. To be correct, the last unnamed parameter has to be validated as a date; your request means that any of the date formats accepted by the MoS have to be allowed: as well as "YYYY", "YYYYl", "nd", "n.d.", the code will have to check "DD M YYYY", "M DD, YYYY", "M YYYY", etc. And then for periodicals, people will ask for "Summer YYYY", ranges of months, ...) Peter coxhead (talk) 08:54, 18 July 2017 (UTC)

I feel like I'm beating my head against a brick wall here. Firstly, I don't know what others do, but I use the same citation template for every citation in any given article, per WP:CITE. And that is (or perhaps was) {{sfn}}, because it was clean and simple. Secondly, there was an existing and documented work-around for the rare "problem" that was "fixed" by this recent change. Thirdly, there was a functional and useful employment of the {{sfn}} template that made it easier for readers when hovering over the citation in the text, linked to the full citation, and differentiated citations within the Footnotes section. So, the outcome here is that a ridiculously minuscule proportion of {{sfn}} citations which have three or more authors and no year of publication (which already had an existing work-around) is now fixed, and we lost a significant functionality. Great, I look forward to the next "improvement" to the template. Peacemaker67 (click to talk to me) 10:10, 18 July 2017 (UTC)

@Peacemaker67: I can only say that I feel the same about my head and a brick wall. :-) {{sfn}} isn't a "citation template", it's a way of linking to a citation template. If I understand you correctly, what you mean is that you want to use {{sfn}} when it isn't necessary, i.e. when the short form has a 1:1 relationship with the actual citation. Since this isn't its purpose, at least as I understand it, you're asking Trappist the monk, or some other template editor, to do quite a bit of work (since, as I noted above, any date used as a value in the template should be validated) in order to support functionality which wasn't part of the intended use of the template, so used to work essentially by accident.

To move forward, I suggest that you accept that it would be wrong to revert the changes made, and instead make a case for the functionality you would like to be added to the template. I'm not personally convinced at present that this would be a good idea, but I could be persuaded if you make a good case, which can't simply be that "it used to work like this", given than it shouldn't have. Peter coxhead (talk) 11:37, 18 July 2017 (UTC)

I don't agree that {{sfn}} isn't a citation template; clearly it is only useful as a member of a group of templates used to create citations. More importantly, I don't agree that it shouldn't be used when a source is only cited once (or when all the citations are to exactly the same page).

1. There may be a desire to indicate the page range in the long citation and the specific page that supports the claim in the short citation.
2. The style of the article may be to list all the sources in one alphabetical list, so the bibliography becomes a handy overview of the sources related to the article, no matter how many or few times a particular source is cited.
3. It may be likely that a certain source will be cited to support other claims, so the article is set up to make the insertion of these future short cites easy.

Jc3s5h (talk) 12:52, 18 July 2017 (UTC)

I don't understand (1), but (2) and (3) make sense to me, so perhaps Trappist will consider adding this facility. Peter coxhead (talk) 19:09, 18 July 2017 (UTC)

• {{sfn}} might be used for newspapers when the editor wants to use short refs and is citing different page numbers of the same edition—{{sfn|The Times|18 July 2017|p=1 and later {{sfn|The Times|18 July 2017|p=3—or when no page numbers are involved but the editor nevertheless wants to use shortened refs for all citations. SarahSV ^(talk) 19:24, 18 July 2017 (UTC)

For an example of no. 1, see a particular version of Coordinated Universal Time. The first footnote cited page S181 of Guinot's 2011 paper, and the full citation in the bibliography specified that the full article is on pages S181 through S185. (This was a special issue of Metrologica, which is why the page numbers started with S.) Jc3s5h (talk) 20:25, 18 July 2017 (UTC)

@Jc3s5h: the case for a change, as I understand it, is that some editors want to have one list consisting only of short footnotes that in every case link to the full citation in another list. But your Coordinated Universal Time example isn't like that: the Notes list contains some full citations that don't simply link to the Bibliography. So it doesn't, to me anyway, support the case for having full dates in {{sfn}}. Peter coxhead (talk) 05:48, 19 July 2017 (UTC)

@Peter coxhead: Try Hamlet. In particular notes 203–9 (Michael Billington is a well known and prolific theatre critic; he publishes a lot of reviews in a single year). These don't happen to use full dates, but if I'd needed to cite more than one article in the Daily Mirror or Broadwayworld from a single year, I would have had to start using 2009a, 2009b, etc. absent support for full dates. And if you have Harv warnings on, you can see the problem with that on the full cite for Chambers 1923a: an editor came along and tried to "fix" the cites and ended up leaving them dangling (Chambers 1923a isn't cited in the article). It's a necessary workaround some times, but it's not the preferable approach in cases where one can use full dates instead. As for SarahSV's point above, I believe it relates to things like very long journal articles (whatever you think "long" is: 10, 15, 30, 50, 100 pages). In the full citation you give the page range for the whole article, and in the short citation you provide the specific page or pages that support the specific information it is attached to. The same journal article will be cited many times, but with different specific page(s) for each instance. --Xover (talk) 10:05, 19 July 2017 (UTC)

@Xover: re your interpretation of SarahSV's point, you mention journal articles, but for these the full date definitely isn't needed. When for newspaper articles do you need to give a page within the page range of the article?

The point is that there's a trade-off between the processing needed to validate a full date and the convenience that it might provide in a limited set of cases, for which Trappist the monk has suggested a work-around. To be persuasive, there needs to be a clear and precise specification of why and when the "YYYYa", "YYYYb", etc. notation, which is standard in the Harvard style, isn't appropriate and should be replaced by the full date. Anyway, I think I've said enough on this matter. Peter coxhead (talk) 20:34, 19 July 2017 (UTC)

@Peter coxhead: The journal example is by analogy; the situation is exactly the same for newspaper (or magazine) articles. If what you're citing is a part of a larger work (chapter, newspaper/magazine article, journal article, etc.) but still of any significant length in itself, you will need to specify page numbers. This will be substantially less common in modern newspaper articles (the tabloidization of the popular press discourages long articles through several mechanisms), but still occurs, and is not uncommon in magazine articles or in older newspapers. For example, Edmond Malone's 62-page Cursory Observations (1792) was first published in The Gentleman's Magazine (in two parts, but still).

And the tradeoff you mention is rather exaggerated. Both as a practical matter, and as a matter of policy (WMF stance is that "server load"-type considerations should not be taken into account), the amount of processing required for this is utterly insignificant. A more salient tradeoff is added complexity (of use and code), and resources required for implementation and maintenance. I don't see the added complexity rising to a level that is significant here, and the existing date validation code should be close to a drop-in implementation for this (there's other work needed too; but writing date validation from scratch is major undertaking). In other words, I do not believe these to be valid counter-arguments. However, someone that is actually familiar with the code and will in practice have to maintain it in the future is the right person to speak to these aspects.

The two remaining aspects are then a) "Is there any use for this?", and b) "Does it fit within the architectural design and scope of this template?". I believe this discussion has demonstrated that there is indeed a need for, and desire for, this functionality (in fact, scanning this thread you are the only one that is actually arguing against adding this functionality, but this may be due to the misunderstanding regarding the change that broke the historical unsupported use of it). And I have as yet not seen any actual argument that the functionality would not fit within the template's scope (a shortened footnote is no less of a shortened footnote just because it uses a full date rather than just a year). --Xover (talk) 11:40, 20 July 2017 (UTC)

at parameter

I would request that {{sfn}} do something when given the a parameter named at. I occasionally attempt to use this parameter instead of loc, intuitively following the usage of the at attribute for {{cite}}. Unfortunately, if I do this, there is no indication that the parameter was not recognized, eg.

{{sfn|Author|2017|at=Chapter 4}}{{reftalk}}

References

I am open to any of three options:

1. Treat at as a synonym for loc, following the usage in {{cite}}
2. Display a warning message in article previews.
3. Display a warning message in final pages which use this template.

I prefer the first option, but would gladly accept any of these over the present situation. Daask (talk) 17:39, 20 July 2017 (UTC)

@Daask: Use |loc=Chapter 4. --Redrose64 🌹 (talk) 19:25, 20 July 2017 (UTC)

@Redrose64: I am aware that loc is the correct way to do this, but am making this request because there is no acknowledgement when I make a mistake and use at instead. I can almost guarantee I have left a {{sfn|at=Chapter 4}} somewhere on Wikipedia because of this. Daask (talk) 13:59, 21 July 2017 (UTC)

Loose ampersand again

When working on historic buildings in England there are a couple of useful templates: {{NHLE}} and {{Listed building England}} which rely on using the listing number as part of the link. For instance ...text.{ {sfnp|Historic England|1336169}} ... References { {reflist}} ... Bibliography * { {NHLE | num = 1336169 | desc = Miller monument | access-date = 2 September 2016 | fewer-links = yes | mode=cs2 }} Which used to produce: References Historic England 1336169. ... Bibliography * Historic England, "Miller monument (1336169)", National Heritage List for England, retrieved 2 September 2016 Unfortunately this now falls foul of the changes discussed above giving references in the form Historic England & 1336169. The links between the references and the citations work correctly, it's just the display that is an issue. Peter and others have made it plain above that the changes are an improvement and usages such as this are incorrect. Therefore I have two questions:

1. Is it possible to set a switch to turn off the checking? Either as an extra parameter or as a global parameter on the whole page. If not then:
2. Which was the last version of sfn which supported this abuse? An obvious solution is to fork that version off as, say, sfnx (and likewise for sfnp -> sfnpx).

Regards, Martin of Sheffield (talk) 16:30, 30 July 2017 (UTC)

{{sfn}} just calls a Lua function in Module:Footnotes, so there's no version of it that can be used in the way you suggest. If there is a real demand for short footnotes which don't follow the Harvard style and in which the last value isn't a year but some other identifier, then someone (in practice Trappist the monk, I suspect) will need to write new code in Module:Footnotes which "sfnx" and "sfnpx" can then pick up. Peter coxhead (talk) 17:00, 30 July 2017 (UTC)

Issue with sfn - no author - sfnref attempted

Could someone who understands the complexities of this template possibly help? On West Somerset Mineral Railway there are two citatiomns to The Railway Magazine - Vol 86 (December 1940) and Vol 87 (May 1941). No authors are given. An attempt has been made to use |ref={{sfnref|The Railway Magazine|1944}} however they still give an sfn error "Harv error: link from #CITEREFRailway_Magazine1940 doesn't point to any citation." & in the source list "Harv warning: There is no link pointing to this citation." I have been struggling to fix these for ages and can't seem to make it work. Any advice or fixes appreciated.— Rod ^talk 17:10, 6 August 2017 (UTC)

@Rodw: See NBR 224 and 420 Classes, East Linton railway station or Reston railway station for some examples of how I handled such situations. --Redrose64 🌹 (talk) 20:18, 6 August 2017 (UTC)

I see that Trappist the monk already got to it, thankfully. For my own edification, why was italics necessary for matching here? Is it because of the magazine field? Thanks, —PaleoNeonate – 20:21, 6 August 2017 (UTC)

Italics not necessary for matching. Magazine titles should be italicized per MOS:ITALIC.

—Trappist the monk (talk) 20:30, 6 August 2017 (UTC)

Thanks. —PaleoNeonate – 20:39, 6 August 2017 (UTC)

Thanks all for comments & especially to Trappist the monk for fixes.— Rod ^talk 20:22, 6 August 2017 (UTC)

Postscript revisited

See Template talk:Sfn/Archive 2#Postscript

I have not come across the use of the ps= parameter before today. The article I have found it being used is Siege of Oxford. I have User:Ucucha/HarvErrors installed and so the article is awash with red-warnings, all of a similar type. The first of these is:

• {{sfn|Walker|1705|pp=18-21|ps=, entries for June 1, 2, 3 and 4}}
• {{sfn|Walker|1705|pp=18-21}}
• {{sfn|Walker|1705|pp=18-21}}

It seem to me that this is an abuse of what the post script parameter is there for and as it breaks multiple citations to the same page that it is more than a nuisance. As the same affect can be achieved by

• {{sfn|Walker|1705|pp=18-21, entries for June 1, 2, 3 and 4}}

I think that the ps= parameter should be restricted to a single character or "none" as a special case. -- PBS (talk) 12:47, 2 September 2017 (UTC)

The documentation describes the ps parameter as follows:

Parameter |ps= can be used to add quotes or additional comments. Note that this effect can also be achieved using {{harvnb}}, if done as shown in the last example. The postscript is only effective the first time {{sfn}} is used for a particular author, year and location.

I think that makes it clear the endnote in the article mentioned by PBS is using the parameter as intended. Using it to modify the punctuation at the end of the endnote would defy the documentation, and "pp" would not be an appropriate name for a parameter that was only about terminal punctuation.

But I think the parameter is badly designed. It has the same problems as "ibid." If an editor uses the parameter on the first spot where a particular page is mentioned, and a later editor rearranges the text so a different reference to the same page is first, the extra information will be attached to the wrong citation. (That's assuming the parameter works as described, which I have not tested.) Jc3s5h (talk) 13:13, 2 September 2017 (UTC)

Just because the documentation says it it does not have to remain that like that for ever. In the version of the article I liked to above it means that the first instance was visible (with a warning) but the second and third were not. It seems to me that this is a dumb thing to do. -- PBS (talk) 13:22, 2 September 2017 (UTC)

The errors I see are primarily same-reference-name-with-different-content errors:

{{sfn|Walker|1705|pp=18-21|ps=, entries for June 1, 2, 3 and 4}}

{{sfn|Walker|1705|pp=18-21}}

produce:

<ref name="FOOTNOTEWalker170518-21">[[#CITEREFWalker1705|Walker 1705]], pp. 18-21, entries for June 1, 2, 3 and 4</ref>

<ref name="FOOTNOTEWalker170518-21">[[#CITEREFWalker1705|Walker 1705]], pp. 18-21.</ref>

which shows that the reference names are the same but clearly, the content is different. These are not User:Ucucha/HarvErrors but are MediaWiki errors.

The fix, I would think for this problem is to use the correct template parameter: |loc=

{{sfn|Walker|1705|pp=18-21|loc=entries for June 1, 2, 3 and 4}}

<ref name="FOOTNOTEWalker170518-21entries for June 1, 2, 3 and 4">[[#CITEREFWalker1705|Walker 1705]], pp. 18-21, entries for June 1, 2, 3 and 4.</ref>

• Walker (1705), Title

References

No need to 'fix' {{sfn}}.

—Trappist the monk (talk) 14:06, 2 September 2017 (UTC)

While there may or may not be a need to fix snf, this problem will continue while setting ps= can alter the content of the citation while not altering the reference-name. However the quick an easy fix that will sort out most of these problems is to alter the documentation to emphasise your solution of loc= in the section "Additional comments or quotes", rather than the currentps= parameter. -- PBS (talk) 14:28, 2 September 2017 (UTC)

There is a slight difference in syntax with using p or pp with ps and loc. With loc the page[s] string ends with a comma with ps the terminating comma after the page[s] is missing. So the examples in the documentation are not a one to one substitute between ps and loc. In the example the string passed into the parameter starts ": ..." this works fine with ps but with loc produces ",:". Does this matter (or do we change the documentation to use an example string that does not start with a punctuation mark)?

References

-- PBS (talk) 14:36, 3 September 2017 (UTC)

Why are you surprised that the two different parameters should render differently? They serve different functions.

Module:Footnotes inserts a comma and space between the value assigned to |loc= and whatever precedes it (this is a function of |loc= not of |p= or |pp=). The module does not insert any characters ahead of the value assigned to |ps= because to do so could produce odd combinations of punctuation characters.

—Trappist the monk (talk) 15:11, 3 September 2017 (UTC)

I thought |ps= was intended for overwriting the terminal period that {cite} automatically adds, with the typical usage being to replace it with a comma. This business of adding quotes, etc., would be better done using |quote=, except for that not being a good idea to start with. At any rate, the documentation should not be suggesting use of |ps= for "additional comments", etc. ~ J. Johnson (JJ) (talk) 22:21, 3 September 2017 (UTC)

Treating year range as second author?

Hi, I just noticed that when the year is a range it treats it as a second author even when it's the last field. For instance:

Test. ({{sfn|Lastname|1999–2000}})

References

Works Cited

Lastname (1999–2000). Title. {{cite book}}: Invalid |ref=harv (help) ({{cite book|last=Lastname|date=1999–2000|title=Title|ref=harv}})

It still works as an anchor to the citation, but it displays as "Lastname & 1999–2000" instead of "Lastname 1999–2000". What should I be doing instead to force it to treat the final field as a year and not an author? Is there a better way, or should I just do:

Test. (<ref>[[#{{harvid|Lastname|2000–2001}}|Lastname 2000–2001]]</ref> )

References

Works Cited

Lastname (2000–2001). Title2. {{cite book}}: Invalid |ref=harv (help) ({{cite book|last=Lastname|date=2000–2001|title=Title2|ref=harv}})

Thanks! Umimmak (talk) 20:14, 18 September 2017 (UTC)

When would you need to use a range? Please give examples of pages where this problem occurs. --Redrose64 🌹 (talk) 22:02, 18 September 2017 (UTC)

I see Trappist has fixed it. It used to also work with 1974–75, but I can change that to 1974–1975 if necessary. Kanguole 23:26, 18 September 2017 (UTC)

Thanks, Trappist_the_monk! Umimmak (talk) 05:16, 19 September 2017 (UTC)

Treating month as second author

Does sfn allow months as well as years? I've just added:

• {{sfn|Fleming|April 2014}}
• Fleming, Michael (April 2014). "Allied Knowledge of Auschwitz: A (Further) Challenge to the 'Elusiveness' Narrative". Holocaust and Genocide Studies. 28 (1): 31–57. doi:10.1093/hgs/dcu014. {{cite journal}}: Invalid |ref=harv (help)

And the short cite rendered as Fleming & April 2014.

I added April because there is another Fleming 2014 (a book), so I wanted to add the month to his paper. SarahSV ^(talk) 02:35, 30 September 2017 (UTC)

Disambiguation of same-author-same-year is by addition of a lower case letter to the |date= parameter in the cs1|2 template (for the journal: |date=1 April 2014a; for the book |date=2014b) then {{sfn|Flemming|2014a}} and {{sfn|Flemming|2014b}}

—Trappist the monk (talk) 02:48, 30 September 2017 (UTC)

Gday SlimVirgin. It's not an intended use, but in that situation {{sfn|Fleming April 2014}} works as "Fleming April 2014" without the ampersand. Cheers, Peacemaker67 (click to talk to me) 02:49, 30 September 2017 (UTC)

Peacemaker67, thanks. The sfn displays that way, but the link doesn't work. The book short cite (Fleming 2014) is linking to the Fleming April 2014 long cite, so it looks as though I will have to ditch the month. SarahSV ^(talk) 03:02, 30 September 2017 (UTC)

It will if you format the ref field as {{harvid|Fleming April 2014}}. Peacemaker67 (click to talk to me) 03:47, 30 September 2017 (UTC)

Well that should be {{harvid|Fleming|April 2014}}, but really (as I keep telling Nishidani) the solution described by Trappist is better, and is the standard academic convention anyway. The only potential problem is the limitation to a single lower-case character, so it won't work if you're citing more than 26 works by the same author in the same year. Very unlikely, since even the most prolific author couldn't produce 27 or more books or journal articles in a year, and even then you are very unlikely to want to cite all of them in the same article. The only case I can think of where it might happen is a newspaper journalist writing daily reports on events eligible for inclusion in a list article. --NSH001 (talk) 04:18, 30 September 2017 (UTC)

Actually, putting the pipe in causes the ampersand because of the change Trappist did earlier this year, as was discussed ad nauseum. Not using the pipe avoids it. Peacemaker67 (click to talk to me) 06:31, 30 September 2017 (UTC)

True, but it's not what it "should" be. --NSH001 (talk) 06:49, 30 September 2017 (UTC)

My point precisely. Peacemaker67 (click to talk to me) 06:58, 30 September 2017 (UTC)

• Thank you, everyone who responded! Trappist the monk, that's what I couldn't figure out, how to do 2014a and 2014b, but still have April 2014 in the long cite. Thanks for the solution. SarahSV ^(talk) 04:51, 30 September 2017 (UTC)

Okay, NSHOO1. No need to blurt the beans about my template-citing follies, it was supposed to be a well-kept secret that I can't remember how to do them. Now I'll be conscience-stricken as I try to take the above on board, and no doubt rewrite the Axel Munthe article thinking I have to fix refs to his classic as being written by Axel and Munthe? :)Nishidani (talk) 07:04, 30 September 2017 (UTC)

"The templates assume that the last field is the year"

Right now the documentation says "The templates assume that the last field is the year", but this no longer seems to be the case. Now, it has to be a year, any deviation will make it think it's a name and stick in unwanted "ands". I already had a note about a range of years in the last field, and I just noticed it's treating a date as an author.

Since it now changed and seems to err on assuming things are names when in doubt, this should probably be cleared up.

Umimmak (talk) 12:26, 9 October 2017 (UTC)

Problem with "ps"

I want to write:

• Gerlach writes that over three million Jews were murdered in 1942 ... {{sfn|Gerlach|2016|p=99}} At least 1.4 of these were in the General Government area of Poland.{{sfn|Gerlach|2016|p=99|ps=, note 165}} [8]

This is causing an alert: "defined multiple times with different content". Help:Cite errors/Cite error references duplicate key says: "ps= must have the same value in each use of {{sfn}} where the other parameters are identical. An assigned value in ps= cannot be combined with an omitted ps= elsewhere on the page. If there is a blank ps= then remove it."

How can I cite p 99, then p 99 with a footnote? SarahSV ^(talk) 03:59, 9 October 2017 (UTC)

Two other ways which I commonly see is including quotes as part of the page: {{sfn|Gerlach|2016|p=99 "quote"}} or probably better, using {{harvnb}} for this instance which does not have implicit <ref>...</ref> tags or autocombining. i.e. <ref>{{harvnb|Gerlach|2016|p=99}} – "quote"</ref>. I hope this helps, —PaleoNeonate – 04:21, 9 October 2017 (UTC)

The above is an example for quoting, but it would be the same to add other types of notes of course. —PaleoNeonate – 04:27, 9 October 2017 (UTC)

@PaleoNeonate: thanks. This worked: {{sfn|Gerlach|2016|p=99, note 165}} Sorry, I should have tried that before posting here. For some reason I assumed it had to be in the "ps" field. SarahSV ^(talk) 04:33, 9 October 2017 (UTC)

It's possible that the documentation could offer more obvious examples for this situation. Welcome, I'm glad that I could help, —PaleoNeonate – 04:36, 9 October 2017 (UTC)

Perhaps even better would be to reserve |p= and |pp= for page numbers and put the additional location information in |loc=, e.g. {{sfn|Gerlach|2016|p=99|loc=note 165}}. Kanguole 08:18, 9 October 2017 (UTC)

Thanks, Kanguole. Another problem:

{{sfn|Stone|2010|p=109|ps=, citing {{harvtxt|Bajohr|Pohl|2008|p=10}}}}

produces:

Stone 2010, p. 109, citing Bajohr & Pohl (2008, p. 10)

How do I get rid of the brackets around 2008, p. 10? SarahSV ^(talk) 20:06, 9 October 2017 (UTC)

{{sfn|Stone|2010|p=109|ps=, citing {{harvnb|Bajohr|Pohl|2008|p=10}}}} produces:

References

HTH, Martin of Sheffield (talk) 21:53, 9 October 2017 (UTC)

I'm not sure sfn is the right tool for this – it might be better to use

<ref>{{harvnb|Stone|2010|p=109}}, citing {{harvnb|Bajohr|Pohl|2008|p=10}}</ref>

Kanguole 21:31, 9 October 2017 (UTC)

Martin of Sheffield and Kanguole, thank you. Kanguole, I'm not clear about the difference between sfn and harvnb. The article uses sfn, so I've been sticking to that. SarahSV ^(talk) 22:09, 9 October 2017 (UTC)

{{sfn|Stone|2010|p=109}} produces the same output as <ref>{{harvnb|Stone|2010|p=109}}.</ref>, except that it merges repeated references. So if you use sfn for simple references and ref+harvnb for more complex citations like this one, they will produce compatible formatting. Using |ps= for this purpose will lead to the problem at the top of this section. Kanguole 22:27, 9 October 2017 (UTC)

(edit conflict)SlimVirgin As a good first approximation {{sfn|author|date}} is the same as <ref>{{harvnb|author|date}}</ref>. Likewise sfnp and harvp are closely related. Moving beyond the first approximation, sfn will group identical references together:

harvnb|smith|2020|p=5, harvnb|smith|2020|p=5, sfn|smith|2020|p=5, fn|smith|2020|p=5

References

to get an identical effect with harvs would require a name parameter to the ref tag. The other advantage is that the syntax for sfn is easier (no refs to remember) so less liable to typos! Where harv and friends are useful is when you are doing something like adding an in-line reference as in citing {{harvnb|Bajohr|Pohl|2008|p=10}}. Regards, Martin of Sheffield (talk) 22:33, 9 October 2017 (UTC)

Thanks, Martin, these explanations are very helpful. I'm curious why, in Kanguole's view, <ref>{{harvnb|Stone|2010|p=109}}, citing {{harvnb|Bajohr|Pohl|2008|p=10}}</ref> is better than {{sfn|Stone|2010|p=109|ps=, citing {{harvnb|Bajohr|Pohl|2008|p=10}}}}. Is it better for some reason not to mix sfn and harvnb within one citation? SarahSV ^(talk) 22:48, 9 October 2017 (UTC)

The problem is the use of |ps=. If {{sfn|Stone|2010|p=109}} occurs elsewhere, you'll get the "defined multiple times with different content" error again. Kanguole 23:47, 9 October 2017 (UTC)

Sfn with "ps=, citing" works if you follow it with {{harvnb|Bajohr|Pohl|2008|p=10}} But it doesn't work if you follow it with {{harvtxt|Bajohr|Pohl|2008|p=10}}. SarahSV ^(talk) 00:07, 10 October 2017 (UTC)

For any given combination of author(s) year, and page/loc info, {{sfn}} expects that there will be no more than one distinct version of the |ps= parameter; being absent counts as one version. So

First sentence.{{sfn|Stone|2010|p=8|ps=postscript 1}}
Second sentence.{{sfn|Stone|2010|p=9|ps=postscript 2}}

is fine, because the page numbers differ; but

First sentence.{{sfn|Stone|2010|pp=8-9|ps=postscript 1 & 2 together}}
Second sentence.{{sfn|Stone|2010|pp=8-9}}

will throw an error because the |ps= differ but the rest is the same. The error checking is not part of {{sfn}} but is part if the "cite.php" extension - that's the code in the MediaWiki software that handles the <ref>...</ref> and <references /> tags. As noted above, {{sfn|Stone|2010|p=8}} is functionally identical to <ref name="FOOTNOTEStone20108">{{harvnb|Stone|2010|p=8}}</ref> so whatever goes into that name= attribute is the key here. --Redrose64 🌹 (talk) 10:24, 10 October 2017 (UTC)

@Redrose64: thank you, that's a very helpful explanation. Given that {{sfn|Stone|2010|p=8}} is functionally identical to <ref name="FOOTNOTEStone20108">{{harvnb|Stone|2010|p=8}}</ref>, what allows this to work: {{sfn|Stone|2010|p=109|ps=, citing {{harvnb|Bajohr|Pohl|2008|p=10}}}}, while {{sfn|Stone|2010|p=109 is elsewhere in the text? SarahSV ^(talk) 01:45, 11 October 2017 (UTC)

Eh? It throws a great big red error message:

References

Which means that it doesn't work. --Redrose64 🌹 (talk) 17:06, 11 October 2017 (UTC)

I'm not seeing an error message. It works in The Holocaust, currently ref 47. SarahSV ^(talk) 17:25, 11 October 2017 (UTC)

It 'works' at The Holocaust because there is only one Stone 2010 P. 109. Add anther reference to Stone 2010 @ p. 109 (with a different |ps= or without) and there will be an error.

—Trappist the monk (talk) 17:38, 11 October 2017 (UTC)

Sorry, I got confused with another example of doubling up that worked. I've just added {{sfn|Stone|2010|p=109}}, and it's not returning an error message. [9] SarahSV ^(talk) 17:47, 11 October 2017 (UTC)

I see one.

—Trappist the monk (talk) 17:57, 11 October 2017 (UTC)

I'm sorry, you're right! I was looking at the top of the page, because the last time this happened (see my first post in this thread), there was a red error message at the top on preview. So is it better always to write {{sfn|Stone|2010|p=109, citing X}}, rather than using "ps", because then the reference won't be functionally identical to any other instance of p. 109? SarahSV ^(talk) 18:05, 11 October 2017 (UTC)

Always? Meh. I think that |p=109, citing X is semantically incorrect; ', citing X' isn't an in-source location so really doesn't belong in |p=. Making it part of |ps= is, as we've seen, problematic because there are probably times when an editor would like to cite p. 109 where the author isn't citing some other author. I guess, for me, without we apply some sort of technical fix to Module:Footnotes, I would probably implement Editor Kanguole's solution for those 'unique' cases:

<ref>{{harvnb|Stone|2010|p=109}}, citing {{harvnb|Bajohr|Pohl|2008|p=10}}</ref>

But, I do have to wonder, why not just cite Bajohr & Pohl 2008, p. 10 directly and avoid all of this mess?

—Trappist the monk (talk) 19:25, 11 October 2017 (UTC)

Per WP:SAYWHEREYOUREADIT. Stone 2010 is our source for a figure I find unlikely (in the sense that I can't imagine how anyone arrived at it), so I've added Stone's source, which is Bajohr & Pohl 2008, p. 10, a German-language source I haven't seen.

The reason I'm asking these questions is that I'm trying to understand how the templates work, rather than just mimicking what I see others doing, without understanding why they're doing it—which means I have to remember all uses, instead of being able to work out for myself which is best. So, in that spirit, can you say why it matters that |p=109, citing X is semantically incorrect? SarahSV ^(talk) 20:15, 11 October 2017 (UTC)

Because the |p= parameter is intended for a page number. Nothing else. --Redrose64 🌹 (talk) 20:55, 11 October 2017 (UTC)

I was wondering whether it might cause a clash with some other template, or something like that. But I do take Trappist's point that it's better to write <ref>{{harvnb|Stone|2010|p=109}}, citing {{harvnb|Bajohr|Pohl|2008|p=10}}</ref>. SarahSV ^(talk) 22:07, 11 October 2017 (UTC)

Loose ampersand using "1996–97" as the date

This is perhaps related to the "Query re: loose ampersand using sfn and harvid" above. There, however, the problem was created in attempting to lengthen a date (14 May 1920 instead of 1920), whereas here the problem is created in trying to shorten a date.

A few pages (Guilden Morden boar, Sutton Hoo helmet) use an article that appears in the "1996–97" issue of a journal. Using this date generates a stray ampersand: {{sfn|Meadows|1996–97|p=193}} becomes Meadows & 1996–97, p. 193.. My current fix is to turn 1996–97 into 1996–1997, which solves the issue ("Meadows 1996–1997, p. 193."). Is there a way to retain the journal's given, shorter, date without resulting in a stray ampersand? Thanks, --Usernameunique (talk) 12:42, 13 October 2017 (UTC)

Above in the "Treating month as second author" section Peacemaker67 had a workaround for a similar issue; it'll look the way you want if you just get rid of the pipe: {{sfn|Meadows 1996–97|p=193}}, just make sure you set |ref={{harvid|Meadows 1996–97}}. I wish there were a better way to tell the code "this is the date" but it'll look right if you tell it "this is part of the name". Umimmak (talk) 12:57, 13 October 2017 (UTC)