Help talk:Citation Style 1/Archive 13

Archive 10

Archive 11

Archive 12

Archive 13

Archive 14

Archive 15

→

Archive 20

Open access icon

The Wikipedia Library's recommended citations often request that {{open access}} () be appended at the end of the citation. Is there a reason why this (and {{closed access}} ) isn't baked into the citation template itself? I would think the metadata alone would be worthwhile. czar 22:11, 12 March 2016 (UTC)

Link to where that recommendation is written? What metadata? The {{open access}} and {{closed access}} templates are just pictures. Are you looking for some sort of parameter support?

—Trappist the monk (talk) 22:29, 12 March 2016 (UTC)

Maybe something like a |url-free=/|doi-free= (and similar for the other identifiers), which could either be set to yes/no, or used instead of |url=/|doi=? This when a free link is identified, then it can be used to automatically populate the url field, like PMC {{{1}}} already does. Headbomb {talk / contribs / physics / books} 02:44, 13 March 2016 (UTC)

@Trappist the monk, examples of TWL recommended citations at Wikipedia:Newspapers.com#Citation and Wikipedia:Newspaperarchive.com#Citation, but it's the same for any TWL-partnered database with open access. czar 04:27, 13 March 2016 (UTC)

Is this is (over)stating the obvious? One can just click at a weblink – I think that an accessible link would (rightly) be considered the default state by the majority of readers. If the link has any sort of access requirements, there's already embargo, registration, and subscription parameters built into the templates. 72.43.99.138 (talk) 13:29, 13 March 2016 (UTC)

Thank you. And the answers to my other questions?

Does this icon only apply to TWL-partnered databases? Is a working doi, pmc, jstor, identifier a requirement for journal cites? For newspapers accessible through the two database you mentioned, do we require |via=?

And why is the icon orange? How does orange imply gratis or libre? Yeah, rhetorical question.

—Trappist the monk (talk) 13:43, 13 March 2016 (UTC)

I vaguely recall picking up this practice before TWL distributed database accounts—it went along with the marking of subscriptions within the citation (which I also think could use some reformatting but that's another story). I was helping someone at an edit-a-thon who wanted to know what param to add to get the orange (un)lock and I realized it would be a good idea to bring it here. Could you tell by going to the link directly? Sure. I believe the idea is to incentivize use of open access links in line with our mission. I also think the icon is a nice courtesy, as readers often skip the text, but a lock icon is straightforward. For metadata, I was thinking that it would be easier for someone to parse usage of open access links if they're clearly marked as such in the citation. A separate param (|open=y) could work, or Headbomb's solution works too—the idea is just to get the lock icon at the end of the citation so users don't need to add a template outside the main one. TWL recommends the db identifiers when they fit, I believe, and they appear to use {{via}} (separate template) sometimes and |via= in others. I use the latter. Are you suggesting that certain ISSNs automatically generate the lock icon in their citations? As for orange, after a quick search, I have no idea, but it is the standard icon. czar 15:20, 13 March 2016 (UTC)

I made no statement with regard to ISSNs. I was trying to get you to tell me if, as a requirement to display the oa icon, cs1|2 journal cites must have one or more of the normal journal identifiers like doi, like jstor, like pmc. In other words, is it appropriate to apply the oa icon to a {{cite journal}} that only uses |url= and |title=? What about {{cite web}}?

This is where the clutter becomes objectionable. It is the point of my unanswered question: Does this icon only apply to TWL-partnered databases? If all external links in §References are marked with little orange locks, no real usable information is conveyed.

—Trappist the monk (talk) 11:17, 14 March 2016 (UTC)

I don't think there exists such a rule but ostensibly the icon is for distinguishing any potentially closed-looking source as indeed open. If a citation is only using |url= and |title= (out of laziness), I doubt the OA icon will be the third used (over |work=, etc.) czar 14:04, 14 March 2016 (UTC)

I do not understand that last sentence. Clarify?

—Trappist the monk (talk) 15:16, 14 March 2016 (UTC)

is it appropriate to apply the oa icon to a {{cite journal}} that only uses |url= and |title=? What about {{cite web}}? Yes, I'd say. It's for any link the editor wants to specify as open. It isn't restricted to DOI or identifier use. If the New Yorker made its pre-1990 archives publicly available, I'd want to note that they were open access. (Also I didn't mean that you made that statement re: ISSNs—I was just affirming it as a neat possibility, if it could be implemented.) czar 22:51, 14 March 2016 (UTC)

The usefulness of this was questioned because imo most readers would reasonably expect external links in citations (or anywhere in article space) to be open; if so, only the exceptions to this expectation need to be signaled, which is already the case. So semantically this imo may be a non-issue. However, there may be a consensus to implement this for style reasons, and at the editor's discretion (following established guidelines, which allow for variance in application of style). At least this icon makes more sense than useless file icons such as the pdf icon which only adds clutter. 64.134.243.9 (talk) 22:01, 13 March 2016 (UTC)

Thanks for bringing this up, I think it is an important question because the accessibility of sources deserves to be better indicated. Many paywalls are not marked with the "subscription required" option as far as I can tell. In addition, citations often have multiple sources (say, a link, a DOI, an arXiv id, a PMID…) so it is often unclear where the full text can be found. Ideally, it could be useful to have a way to indicate the availability for all of them (individually), but that could be too verbose. Anyway, if these parameters were available, I think filling them would be a good job for OABOT (I am working on it currently). What do you think? Pintoch (talk) 22:30, 13 March 2016 (UTC)

For a citation with a link, a DOI, an arXiv id, and a PMID, couldn't the reader just click on the four links and look at the resulting web pages to see if there is a free full-text version available? That doesn't seem difficult. In any event, if the citation contains four links and one "open access" icon, how does that help the reader figure out which of the four links leads to the free full-text version? – Jonesey95 (talk) 03:10, 14 March 2016 (UTC)

That's my point: a separate, "global" \{\{open access\}\} icon does not help. About clicking on all links: the point of the icons we are discussing here is to avoid just that! There are also cases where people might not even try to click on a DOI, for instance, because by experience they know DOIs tend to be closed, but many of them are actually open. Pintoch (talk) 07:40, 14 March 2016 (UTC)

Too much orange is too much orange. This mocks up a {{cite journal}} from Apatosaurus with all links to the source text marked:

Bates, K.T.; Falkingham, P.L.; Macaulay, S.; Brassey, C.; Maidment, S.C.R. (2015). "Downsizing a giant: re-evaluating Dreadnoughtus body mass". The Royal Society 11: 20150215. doi:10.1098/rsbl.2015.0215. ISSN 1744-957X. PMID 26063751.

What are the rules for applying the oa icon? Are there any? Should there be some? If we are to do this, I would favor limiting the number of oa icons per template to one and only one. We could, instead of creating some sort of parameter that just tags the whole of the citation with little orange locks, create something perhaps like:

|oa-icon=<parameter>

where <parameter> is one of only a handful of parameter names that can be associated with the lock: doi, jstor, pmc. Writing |oa-icon=doi would append the identifier to the doi external link:

Bates, K.T.; Falkingham, P.L.; Macaulay, S.; Brassey, C.; Maidment, S.C.R. (2015). "Downsizing a giant: re-evaluating Dreadnoughtus body mass". The Royal Society 11: 20150215. doi:10.1098/rsbl.2015.0215. ISSN 1744-957X. PMID 26063751.

—Trappist the monk (talk) 11:17, 14 March 2016 (UTC)

I would think that a single orange icon once at the end of the citation would suffice. (Either way this would be limited to once per citation.) I would think that users would expect the title link to link to the open access resource, if the whole citation is tagged. The identifiers, in my understanding, are for those who want the database page, which is often hit or miss for access. czar 14:04, 14 March 2016 (UTC)

On this note, if there is an available open access link, why would one consider citing additional links? Should such practice be discouraged? I don't include discovery/supply-chain identifiers such as ISSN, since they are pointers to discovery of material and do not refer to the material itself (like a link does). 65.88.88.126 (talk) 14:32, 14 March 2016 (UTC)

Providing alternate paths to copies of the same source at perhaps different locations is not necessarily a bad thing.

—Trappist the monk (talk) 15:16, 14 March 2016 (UTC)

Others have noted that, as things stand today, without the icon users must click each link to discover which of them is open. They have noted that this 'requirement' is undesirable. Adding the icon at the end of a citation that has multiple possible links only tells the reader that at least one of them is open. Its sole benefit then is to encourage the reader to undertake the search. If we are going to do this shouldn't we identify that one so that readers don't have to hunt?

—Trappist the monk (talk) 15:16, 14 March 2016 (UTC)

If we go back to the Wikipedia:Newspaperarchive.com#Citation-style example, even if there are multiple identifiers, I think a single trailing lock icon at the end of the citation is sufficient to indicate that the title linked URL is the one that is open source. If someone is prone to being confused by multiple links, I don't see why they would click the PMID or JSTOR seemingly arbitrary numbers when the title is linked. czar 22:51, 14 March 2016 (UTC)

Thinking a bit upon my idea, here's a few ways my option could work, if it's implemented. If we have a free identifier, then the url is automatically generated from it. For arxiv (always free, but not official version), bibcode (sometimes free), and doi (sometimes free), we would have something like this. I'm assuming that the doi is free, but the bibcode isn't.

{{cite journal |last=Smith |first=J. |year=2012 |title=Awesome stuff is awesome |journal=Journal of Stuff |volume=1 |issue=2 |pages=3-4 |arxiv=1001.1001 |bibcode=2012JStuf...1....3S |doi-free=10.1234/0123456789}}

Or alternatively

{{cite journal |last=Smith |first=J. |year=2012 |title=Awesome stuff is awesome |journal=Journal of Stuff |volume=1 |issue=2 |pages=3-4 |arxiv=1001.1001 |bibcode=2012JStuf...1....3S |doi=10.1234/0123456789 |doi-free=yes}}

Option A) Autogenerate, mark everything that is open access.

Smith, J. (2012). "Awesome stuff is awesome". Journal of Stuff. 1(2):3-4 arXiv:1001.1001. Bibcode:/abstract 2012JStuf...1....3S doi:10.1234/0123456789.

Option B) Autogenerate, and only mark the link.

Smith, J. (2012). "Awesome stuff is awesome". Journal of Stuff. 1(2):3-4 arXiv:1001.1001. Bibcode:/abstract 2012JStuf...1....3S . doi:10.1234/0123456789.

Option C) Autogenerate, and only mark the identifiers.

Smith, J. (2012). "Awesome stuff is awesome". Journal of Stuff. 1(2):3-4 arXiv:1001.1001. Bibcode:/abstract 2012JStuf...1....3S . doi:10.1234/0123456789.

Obviously a hierarchy of identifiers should be established, with some always generating links (pmc), others only when their foobar-free version is declared (bibcode, doi), and others disallowed (e.g. issn/isbn) because they will never point to a free version. This hierarchy should be customizable, at least in the case of cite arxiv, where an autogenerated link from |arxiv= is desirable while it would not be desirable in others. Headbomb {talk / contribs / physics / books} 15:45, 14 March 2016 (UTC)

In the general case, I would be somewhat concerned with adding these links to e.g. DOI, because while the identifier is permanent, the redirected website may not be, and the requirements for access to documents at a changed website may make a document unavailable to open access. --Izno (talk) 16:26, 14 March 2016 (UTC)

The thing about implementing these options, is that bots can do this maintenance for us. But typically, open access articles remain open access. Entire journals are open access (some from the outset, some after an embargo periods), whereas other articles are only open access because their authors paid a fee. That that remains true regardless of a change of website or publisher. Something open access, rarely, if ever returned from open to closed access, short of a minority of journals which have reverse embargo periods, which again can easily be handled by bots. Headbomb {talk / contribs / physics / books} 16:43, 14 March 2016 (UTC)

Sure. My concern: Which bot op did you have lined up to program and run a bot associated with such functionality? --Izno (talk) 17:41, 14 March 2016 (UTC)

Pretty much whoever wants to take on the coding effort and goes through a WP:BRFA. Open access is something many bot ops support, and would be more than happy to help with such efforts. Some of this functionality could be implement into existing bots like User:Citation bot, or my own User:Bibcode Bot. A lot of this could also be done through AWB. Headbomb {talk / contribs / physics / books} 17:52, 14 March 2016 (UTC)

It is my understanding that the link target of a doi may depend on where you access it from. So some people may see an open access version while others with access to a subscription-only collection may see it there instead. So I think decorating doi's with temporary and contingent information may be a mistake, and goes against the very purpose of identifying things by dois. —David Eppstein (talk) 17:54, 14 March 2016 (UTC)

I have had similar experience with JSTOR links which sometimes encapsulate dois. There is also the situation where access is open in some geolocations and not open in others, for a variety of reasons, including copyright/legal. 72.43.99.130 (talk) 20:27, 14 March 2016 (UTC)

I haven't experienced anything based on location. I'm not saying it's impossible, but make sure you don't confuse this with institution-based access. I got access to Phys Rev D at work for example, but not at home. Headbomb {talk / contribs / physics / books} 00:57, 15 March 2016 (UTC)

You may not have experienced it, but our digital object identifier article says it can happen: "DOI name resolution may be used with OpenURL to select the most appropriate among multiple locations for a given object, according to the location of the user making the request." —David Eppstein (talk) 01:35, 15 March 2016 (UTC)

Yes, so this only happens when using a DOI in a particular OpenURL resolver, not with the official dx.doi.org resolver, right? Then the same comment that openness depends from the location would be applicable to any other identifier used with OpenURL, if I am not wrong. Pintoch (talk) 13:42, 15 March 2016 (UTC)

I'm afraid you are wrong. It is allowed to have multiple URLs corresponding to a DOI in the handle record, with optional extra information specifying how to choose between them. So whatever resolver is used, there can be no guarantee that the same URL results. See here for a detailed account of how choices are made. It would be odd to set up a handle record with a mixture of open and closed access URLs, but nothing appears to forbid this. Peter coxhead (talk) 16:48, 15 March 2016 (UTC)

Going to ping @DGG: for his input here. Headbomb {talk / contribs / physics / books} 16:36, 15 March 2016 (UTC)

• It can be surprisingly difficult to tell if an OA version exists, if the article does not come from one of the obvious all-OA sources, such as PLOS . It can in some cases be even more difficult to tell if an apparently available OA version is a legal copy: for example, the copies authors place on their own website is sometimes legal in the opinion of the author, but not of the publisher, and there can be no practical way for an outsider reaching the site to tell. Blatantly illegal copies of a great many articles exist on the internet, and not necessarily on obvious pirate sites. In the other direction, publishers have been know to claim copyright on material they clearly have no right for, and if they sell paywalled articles for $60, some of them will gladly sell you a pre-1923 article for the same money. There are furthermore some OA-advocates who to say that a copy available by pushing a web button to ask the author to mail you a copy is sort of OA. If such a button is present it works about half the time according to reports. Some publishers consider this legal; some not, some only with restrictions on the number of copies sent. When I was involved in running interlibrary loan services, this was a matter for specialists.

There is also the problem of works that change OA status. This is generally works that are non-OA for the first 6 months or some other period, and then OA, such as PNAS; it an be even more complicated, for articles with open access as "pre-publication", then closed, and then (sometimes) open again after a period. It can't even be done journal by journal, because some articles in even a toll journal will be OA because the authors paid a fee.

It would not be too complicated for us to programmatically add an OA marker for some specific journals known to be entirely OA--we might for example start by adding it to all PLOS journals. We should be able to add a marker for all articles with a PMCID, though I need to verify that there are no exceptions. We could a systematic run updating everything in PNAS that is from 2014 or earlier. These methods might however be misunderstood as saying that the others are not. It should be possible to develop a (complicated) procedure for identifying automatically a considerable portion of those papers that are in fact OA, at least in biomedicine, but it would have to run each time our article was displayed, but that's something I think should be left to Google.

We in any case are dependent on the person adding the entries to do so correctly. Location by country does is indeed part of the doi specification, but I do not know how widely it is used.

The default has to be not to mark as either OA or nonOA. If people choose to add the an OA marker, then they are responsible for seeing it is correct. We need some way of updating all this, & indicating when it will be updated--in fact , we need to verify and update all citations in WP. But what is annoying, is that at least he OA status needs re-verifying--t least its not bada problem as dead weblinks. Where this will really get to be a problem is the centralized database of citations, if we ever get there. DGG ( talk ) 02:38, 16 March 2016 (UTC)

I believe PMC articles can have an embargo period. The PMCID can be assigned during the embargo period, but the template currently handles that through |embargo=YYY-MM-DD. Which apparently is an undocumented parameter. Headbomb {talk / contribs / physics / books} 14:12, 16 March 2016 (UTC)

Concerning the task of finding OA versions of a given article, various resources have been listed on OABOT. Perfect coverage is of course beyond our reach, but the sources listed there already encompass tens of millions of documents. I am working on the Dissemin API, which can be used in a quite straightforward way for this task (here is a proof of concept). If anybody is interested in participating (especially working on the wikipedia-side of things, which I am not so familiar with), it would be great. Pintoch (talk) 17:58, 16 March 2016 (UTC)

• The original purpose of the OA icon was to indicate scholarly publications compliant with the BOAI definition of open access, which encompasses not just the freedom to read but full reuse rights — essentially, compatibility with CC BY (which only became available later). However, licensing (i.e. reusability) has long been (and still is) regarded as less of an issue than read access, and over time, the icon has been used in many different contexts that blurred the original meaning, to the point that many equal it now with "free to read", which is also the sense in which the icon is now increasingly being used for tagging news articles and other things that would not fall under the original scope of scholarly publications. As pointed out above, it is often difficult and sometimes impossible to decide whether a free-to-read copy of an article is actually legal (or in which jurisdictions, or when, or via which URL etc.).

Based on these and other considerations, the OA Signalling project is planning not to rely on the orange lock but to indicate licensing terms instead (or in addition to it), and to focus only on Wikimedia-compatible licenses (essentially CC BY, since almost no scholarly publications are using CC BY-SA) as well as CC0/ public domain. This addresses a number of points raised in this discussion: the presence of such a licensing indicator would signal that all copies of the article are

• free to read (thus encouraging people to click on any of the links and actually dive deeper into the sources without having to endure paywalls)

and

• openly licensed (thus making readers and editors more aware of the possibility to use the sources or parts thereof — e.g. images or media files — in new contexts, e.g. in their own talks or on Wikimedia projects).

If that second condition is absent (as seems to be the case in the discussion so far), we (or the OABOT team) will have to deal with lots of legal edge cases that

• are hard to tackle algorithmically

and

• were one of the main reasons why Creative Commons licenses have been created.

In addition to signalling the licensing of references on Wikipedia, the OA Signalling project also aims to import full texts into Wikisource, media into Commons (expanding on the Open Access Media Importer) and metadata (including licensing) into Wikidata (see here for a more detailed sketch of the envisaged workflows). The latter will hopefully at some point become (or closely integrated with) "the centralized database of citations" mentioned above and thus useful for handling source metadata across Wikimedia projects (see also the upcoming WikiCite 2016). Having a Wikipedia-cited source available on Wikisource would allow to deeplink into the particular statement in the source that is referred to in the Wikipedia statement citing the source, which should facilitate verifiability.

-- Daniel Mietchen (talk) 22:36, 16 March 2016 (UTC)

Why would we need to detect licenses before adding links to full texts in references? I understand it is a requirement if you want to import them in Wikisoource, but I do not see why adding links to arXiv preprints (mostly non CC-licensed) would be a problem. Most of the published versions are not openly licensed anyway, and they link to paywalls. Like you, I value open licenses, but if we only want to deal with openly-licensed content, we will not be able to do add many links. -- Pintoch (talk) 06:31, 17 March 2016 (UTC)

No one is proposing to make it a requirement that links are free. Headbomb {talk / contribs / physics / books} 11:41, 17 March 2016 (UTC)

Alright, I thought Daniel was (by arguing that we need to take into account his second condition). -- Pintoch (talk) 12:15, 17 March 2016 (UTC)

I think that was a remark on the aims of the OA Signalling project, and possibly the original intent/their guideline on when to use the orange lock. It might be that we'll need to use a different icon, or agree that the meaning of the orange lock has evolved. Headbomb {talk / contribs / physics / books} 12:45, 17 March 2016 (UTC)

I was not talking about adding/ removing/ forbidding arXiv/ DOI/ PMC/ PMCID links — just about indicating what papers are openly licensed. This is a minority, yes, but that makes the signaling all the more valuable. -- Daniel Mietchen (talk) 20:52, 18 March 2016 (UTC)

date=1011–922 BC

In MOS:DATERANGE 1011–922 BC is given as an example of a legitimate range. There was a discussion in Help talk:Citation Style 1/Archive 6#Time to show date error messages?, which is self explanatory. The archived talk section ends with:

Just to satisfy my own curiosity, what are you citing that is nearly 2,400 years old?

—Trappist the monk (talk) 21:24, 19 October 2014 (UTC)

"370 BC" could go in |origyear=, with the publication date of the source you are actually citing (and viewing with your own eyes), or to which you are referring readers, in |date=. – Jonesey95 (talk) 01:06, 20 October 2014 (UTC)

I am running an AWB script and in passing I came across an article (Aristotle) in which there is a citation:

{{cite web
| last =Cicero
| first =Marcus Tullius
| title =flumen orationis aureum fundens Aristoteles
| work =Academica Priora
| date =106–43 BC
| url =http://www2.cddc.vt.edu/gutenberg/1/4/9/7/14970/14970-h/14970-h.htm#BkII_119
| accessdate =25 January 2007}}

It seems to me that as the date given meets the requirements of MOS:DATERANGE one of three things ought to be done. Add code to parse BC dates correctly, or BC dates ought to be silently ignored and not reported as an error, or the advise given by Jonesey95 should be added to the CS1/CS2 documentation. I don't mind which approach is adopted, but it is out of order to flag an error on a correctly formatted (MOS compliant) date without an explanation.-- PBS (talk) 13:27, 19 March 2016 (UTC)

I have made what I think is a reasonable edit to Help:Citation Style 1. Help:Citation Style 2 seems to have been abandoned. I don't know the most efficient way to provide this information at Template:cite book, Template:cite web, Template:Citation, and so on. Jc3s5h (talk) 14:17, 19 March 2016 (UTC)

Do not report as error. Rendering BC dates per MOS:DATERANGE is not a matter of style. Date ranges should be forward (from older-to-newer date), a practice that is firmly established. It just so happens that BC dates count down instead of up. 72.43.99.146 (talk) 15:24, 19 March 2016 (UTC)

Thanks, Jc3s5h. Your edit seems reasonable to me. As for providing this information in the documentation for the cite templates, there is already a link to the section you edited, so that seems adequate for the concise instructions we try to provide on the template documentation pages.

I'm not sure what the IP editor above is getting at. Dates before 100 AD are marked as errors in the CS1 templates. If you want to reopen a discussion about that, please do so in a separate section. This discussion is about the recommendation to give to editors who want to cite such dates. – Jonesey95 (talk) 15:45, 19 March 2016 (UTC)

The problem is the software underlying the template can't handle any year less than 100. So if you read a stone tablet carved in 3 BC, you will either have to live with the error message or not use a template to cite it. Jc3s5h (talk) 15:38, 19 March 2016 (UTC)

@Jonesey95: I was offering my opinion on PBS' comment above: Add code to parse BC dates correctly, or BC dates ought to be silently ignored and not reported as an error, or the advise given by Jonesey95 should be added to the CS1/CS2 documentation. Not reporting the error (which doesn't mean it is ignored) seems like the least confusing/cumbersome option at present. It is not incompatible with adding proper advice to the doc. 65.88.88.69 (talk) 18:41, 19 March 2016 (UTC)

Use found for accessdate parameter with paper sources

Please see Wikipedia:Articles for deletion/Date of birth of Jesus of Nazareth. In this edit the article creator acknowledges the article was compiled by cutting and pasting from other Wikipedia articles without reading the citations. My current proposal is to delete the article for lack of sources. But if instead editors step forward to read and verify the sources, those editors could mark the sources they read with |accessdate=. Then other editors could keep track of which sources have been verified, and which have not.

The presence of accessdate parameters is also useful in detecting this approach to creating an article in the first place; if the access dates are earlier than the creation date of the article, that raises a red flag. Jc3s5h (talk) 13:23, 20 March 2016 (UTC)

There is a template, {{Verify source}}, that is better suited. Mark each reference in the article under discussion. As each reference is checked, remove the tag. By this means, the task that needs to be accomplished is obvious to editors not familiar with the AfD. This method does not require a change at cs1|2.

—Trappist the monk (talk) 14:39, 20 March 2016 (UTC)

Suggestion to add a printing= parameter similar to edition=

Mentioning the printing batch of a work has come out of fashion in recent years, but many older books mention both, an edition and a printing, and it may be even important to mention both since simple fixes of typographical errors and other slight corrections were often more or less "silently" incorporated into newer print runs, without marking this as a new edition. At present, both information would have to be put into the |edition= free-flow parameter, but this results in inconsistent formatting and more difficult (automatic) parsing, therefore a separate parameter |printing= appears beneficial to me. --Matthiaspaul (talk) 18:05, 15 February 2016 (UTC)

I oppose any automatic parsing of |edition= because of the wide variety of ways publishers may describe their editions. Automatic parsing would just be a stumbling stone to mess up editors. Jc3s5h (talk) 19:36, 15 February 2016 (UTC)

Hm, you must have misunderstood my proposal, as it is about adding a |printing= parameter, not about adding any automatic parsing.

|edition= is and |printing= should be free-flow parameters accepting any text. At most, I would add a special case for the case where the value given is a single numerical character ("1".."9"), as this cannot conflict with any other reasonable free-flow text (even not with other short forms like "3rd" or "3." discussed in a thread further above).

--Matthiaspaul (talk) 20:09, 15 February 2016 (UTC)

Well, dates used to be free-form, but now we can't enter February 29, 1900 because of all the error checking. I oppose trying to parse fields that are not standardized by the outside world. "Edition" is not standardized by the outside world, so I oppose parsing it. If |edition= isn't parsed, then printing information can be included in the edition parameter. Jc3s5h (talk) 20:53, 15 February 2016 (UTC)

Thanks for your reply, but I'm afraid there must be a misunderstanding, still. I don't propose at all to add code to somehow "automagically" make sense of whatever an editor might put in a |edition= parameter (it would be way too complicated for the template to sort it out). However, I also don't propose to put printing information into the |edition= parameter. While in lack of a proper parameter for this, |edition= might be the next-best place to put printing information, I don't think it is wise to mix together semantically separate information, this just creates the "mess" (non-standardized and difficult to parse text) you are complaining about. Editions and printings are different properties of a work, therefore we should have different parameters for them, in particular as the |edition= parameter automatically adds "(? ed.)", which clearly does not work for printings.

While (in the other thread above) I propose to add a non-conflicting special case (only for single-character values "1".."9") to |edition= (and |printing=), both parameters should, of course, (continue to) accept any free-flow text and pass it along unaltered. So, I don't see how my proposal could in any way get in the way of your or another editor's edit style.

--Matthiaspaul (talk) 22:01, 15 February 2016 (UTC)

Jc3s5h: I think you're raising your opposition in the wrong thread? --Izno (talk) 22:30, 15 February 2016 (UTC)

I would like to repeat my proposal to add a parameter for print runs like |printing=. To illustrate the fact, why it may sometimes useful/necessary to indicate a particular printing in addition to an edition, the well-known Handbook of mathematical functions by Abramowitz and Stegun was published in 1964 and saw only a single edition (not counting various reprint editions), but went through many printings with a large and growing number of corrections, so, if this work is used to source a formula it might be necessary to indicate the exact print run. Since |edition= automatically appends "ed.", it is difficult to put printing information into the |edition= parameter; it also violates the principle of trying not to combine various information into a single parameter. The idea is to display the contents of the optional |printing= free-flow parameter, if present, following the contents of the |edition= parameter (and separated by a comma), as in this example:

• Abramowitz, Milton; Stegun, Irene A. (December 1972) [June 1964]. Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables. Applied Mathematics Series 55 (1st ed., 10th printing with corrections). Washington D.C., USA: United States Department of Commerce, National Bureau of Standards. LCCN 64-60036.

The parameter may also be convenient to indicate reprints / facsimilies in a more organized way, as in this example:

• Abramowitz, Milton; Stegun, Irene A. (2005) [1965]. Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables. Applied Mathematics Series 55 (1st ed., 9th reprint with additional corrections of 10th original printing with corrections). New York, USA: Dover Publications. ISBN 978-0-486-61272-0. LCCN 65-12253.

If the contents of the |edition= parameter is put into meta-data and there is no meta-data entry for printing information, the whole string "1st ed., 9th reprint with additional corrections of 10th original printing with corrections" rather than only "1st ed." should be put into the meta data, so no information gets lost in the transition, regardless of how the information is given in our template.

--Matthiaspaul (talk) 15:59, 7 March 2016 (UTC)

Setting aside the possible corruption of the |edition= parameter, as I don't know how it's tracked elsewhere, I think an easy workaround to accommodate the "ed." suffix would be to precede the edition number with print run information, rather than trying to place it after the edition number:

• {{cite book |first1=Milton |last1=Abramowitz |first2=Irena A. |last2=Stegun |title=Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables |publisher=United States Department of Commerce, National Bureau of Standards |location=Washington D.C. |edition=10th printing with corrections; 1st |lccn=64-60036 |year=December 1972 |orig-year=June 1964 |series=Applied Mathematics Series 55}} yields:

• Abramowitz, Milton; Stegun, Irena A. (December 1972) [June 1964]. Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables. Applied Mathematics Series 55 (10th printing with corrections; 1st ed.). Washington D.C.: United States Department of Commerce, National Bureau of Standards. LCCN 64-60036.

• {{cite book |first1=Milton |last1=Abramowitz |first2=Irena A. |last2=Stegun |title=Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables |isbn=978-0-486-61272-0 |publisher=Dover Publications |location=New York |edition=9th reprint with additional corrections of 10th original printing with corrections; 1st |lccn=65-12253 |year=2005 |orig-year=1965 |series=Applied Mathematics Series 55}} yields:

• Abramowitz, Milton; Stegun, Irena A. (2005) [1965]. Handbook of Mathematical Functions with Formulas, Graphs, and Mathematical Tables. Applied Mathematics Series 55 (9th reprint with additional corrections of 10th original printing with corrections; 1st ed.). New York: Dover Publications. ISBN 978-0-486-61272-0. LCCN 65-12253.

Does that suffice?

I will also comment that neither |orig-year=, |series=, nor |lccn= are parameters that are included in any of the citation tools to my knowledge.

—D'Ranged 1 | VT^alk :  17:30, 8 March 2016 (UTC)

I have used that as well occasionally, even though I was a bit hesistant to put that info into a parameter it was not designed for originally and also because information on printings typically follows information regarding an edition, not the other way around. Editions and print runs are different properties of a publication. Therefore, lumping them together in a single parameter is not a good idea in general. While the exact future is not predictable, it is clear, that there is a general trend to improved machine readability to enable more advanced link and (re)search methods in the future. So, mixing different kinds of information in a single parameter will fall on our feet sooner or later and cause either valuable information to get lost or unnecessary maintenance overhead in the future when someone uninvolved (=not the original editor and therefore typically without direct access to the sources) will have to sort out what was meant when merged information will have to be split into separate parameters later on. I therefore consider it as a temporary workaround, but only a weak one. Since the template can be improved to fit all requirements I don't think we should settle for anything less than a perfect solution, in particular, if it is about something as easy to implement as this suggestion and does not break anything.

--Matthiaspaul (talk) 14:47, 20 March 2016 (UTC)

web.archive.org/save/...

I just found this template:

{{cite news |ref={{sfnRef|The Jakarta Post, 2002}} |author=<!--Staff writer(s); no by-line.--> |date=June 5, 2002 |title=Medan loses its historical buildings |url=http://www.thejakartapost.com/news/2002/06/05/medan-loses-its-historical-buildings.html |dead-url= |newspaper=The Jakarta Post |location=Jakarta |archive-url=http://web.archive.org/save/http://www.thejakartapost.com/news/2002/06/05/medan-loses-its-historical-buildings.html |archive-date=May 3, 2016 |access-date=May 3, 2016 }}

The May archive and access dates were causing error messages. When I looked at fixing them, I clicked on the citation's title but landed on a page that didn't look like a standard archived page: the header along the top that I usually see was missing. Assuming that I had misclicked, I backtracked and tried again landing in the same place.

Looking at the |archive-url= again, I noticed the 'save' in the path. I wonder then, does that cause archive.org to save a copy of the target url? If it does, we should not be making that kind of link active in cs1|2 templates.

I propose then, to create a test for the content of |archive-url= that looks for |archive-url=http://web.archive.org/save/ (and also https:...) When found, the module will emit an error message and disable the link.

—Trappist the monk (talk) 01:35, 6 March 2016 (UTC)

Yes, "/save/" is used to save a snapshot of the given page. It doesn't make sense in |archive-url= as it is a moving target, whilst |archive-date= specifies a specific date. If this was meant as an example, it should be replaced by "/web/20160503/".

Given that saving snapshots of web pages used in cites is good practise (to fight long-term link rot), but consumes precious time, and given that it is a frequently occuring procedure, perhaps the error message could be made "smarter" than usual to not only indicate the problem but help editors saving a snapshot. Let's assume the template code detects that |url= contains a well-formed link and it finds an |archive-url= containing either only "http[s]://web.archive.org/" (that is, an incomplete link indicating intended usage of archive.org), or a link starting with "http[s]://web.archive.org/save/", or - if we assume archive.org is our default archiving service - a |archive-url= parameter with empty content "", the template could display its error message but additionally provide a number of dynamically constructed links to archive.org to help editors select an existing snapshot link from archive.org or create a new snapshot:

• [https://web.archive.org/save/<contents of url= parameter> Save new page snapshot on archive.org?]
• [https://web.archive.org/web/*/<contents of url= parameter> Select existing snapshot on archive.org?]
• [https://web.archive.org/web/1/<contents of url= parameter> Check oldest existing snapshot on archive.org?]
• [https://web.archive.org/web/2/<contents of url= parameter> Check newest existing snapshot on archive.org?]

If |archive-date= contains a parseable date the template could convert it into ISO 8601 format (without separators) and implant it in the url as well:

• [https://web.archive.org/web/<iso date>/<contents of url= parameter> Check nearest existing snapshot to given archive date on archive.org?]

(Of course, the text in the links should be shorter to keep the error message reasonably short as well - this is just to illustrate the idea.)

This would help editors seeing this error message selecting or creating a suitable snapshot with only a minimal number of clicks and 1 or 2 copy and paste actions (for the link and possibly the date from archive.org back into Wikipedia).

In fact, it could become "standard procedure" for editors, who wish to add an archived version of a page, to just give an empty |archive-url= parameter, and use the error message then displayed in edit preview as a tool to create and/or select a snapshot and date then to be stuffed into the |archive-url= and |archive-date= parameters before saving their contribution.

--Matthiaspaul (talk) 11:48, 6 March 2016 (UTC)

Where is this documented? I did a quick look at archive.org yesterday and didn't find anything.

I do not think that we should offer the ability to save a new snapshot at archive.org. The error messages are visible to all readers so that provides a mechanism for abuse.

If the module had the ability to modify page sources, then offering to select an existing snapshot could be a useful tool. But, since the module can't ... I didn't find any uses of that form of url.

Apparently, https://web.archive.org/web/1/... doesn't always select the oldest snapshot:

https://web.archive.org/web/20000529023519/http://www.math.iastate.edu/abian/homepage.html

We could replace /*/, /1/, and /2/ with a transformed |archive-date=. That may not be the right thing to do because, apparently, archive.org will select the most recent snapshot from that date which may be wrong. This link shows that the last snapshot taken on 2016-03-05 was at 23:56:38:

http://web.archive.org/*/http://www.thejakartapost.com/news/2002/06/05/medan-loses-its-historical-buildings.html

which is the time stamp of the snapshot returned with this url:

http://web.archive.org/web/20160305/http://www.thejakartapost.com/news/2002/06/05/medan-loses-its-historical-buildings.html

I've hacked Module:Citation/CS1/sandbox to detect the /save/ version of an archive.org url. The template described in my first post renders this way:

"Medan loses its historical buildings". The Jakarta Post. Jakarta. June 5, 2002. Retrieved May 3, 2016. {{cite news}}: |archive-url= is malformed: save command (help); Cite has empty unknown parameter: |dead-url= (help)

Right now, the error message categorizes into Category:Pages with archiveurl citation errors. If we further modify the module to detect /*/, /1/, and /2/ (are there others?) then perhaps all of them deserve their own category.

—Trappist the monk (talk) 13:17, 6 March 2016 (UTC)

I have not seen official documentation for /1/ and /2/, but it is cited in various places. Perhaps an over-simplification (or a left-over from the old millennium). However, using f.e. /1000/ (derived from official documentation) instead of /1/ works to retrieve the oldest snapshot:

• https://web.archive.org/web/1000/http://www.math.iastate.edu/abian/homepage.html

A documented replacement for the /2/ thing is to just omit it entirely:

• https://web.archive.org/web/http://www.math.iastate.edu/abian/homepage.html

Official documentation can be found here: https://archive.org/web/web-advancedsearch.php

Some additional documentation can be found here: Help:Using the Wayback Machine

Regarding uses, the idea was to make it as simple as possible for editors to create or select snapshots. Without the links, the editor has to open a browser window, type http://web.archive.org, find the search box for the Wayback Machine and copy and paste the desired link from the WP edit window into there, and then later copy and paste the resulting link from the browser url line back into the WP edit window. Tiresome. With the links provided automatically, most of that procedure could be avoided, it would be down to clicking / selecting the desired snapshot and pasting back the link.

Regarding possible abuse of "save" links in error message, since crawlers will create snapshots (of possibly no longer valid contents) anyway from time to time, I don't think it would be a serious problem. After all, the easier we make it for editors to create and/or select a snapshot representing the desired contents, the sooner the error message will disappear.

Alternatively, is there a way for the template to detect "preview mode"? Does Mediawiki define some special symbol in preview? If so, the additional links could be displayed only in preview mode.

--Matthiaspaul (talk) 15:20, 6 March 2016 (UTC)

I understand the desire to make lives easier for editors but the primary purpose of the module is to render cs1|2 templates into correctly formatted citations. How those templates are created is beyond the scope of the module and properly belongs to source creation tools (Wikipedia:RefToolbar, Wikipedia:VisualEditor, etc.).

The archive link and the static text around it in a rendered citation implies a direct correlation between the original url and the archive url. The module should not render links to unspecified archives; somewhat akin to WP:ELNO item 9. For archive.org, that means that there should be a complete timestamp (YYYYMMDDhhmmss – 14 digits). This is easily checked. Perhaps we should.

—Trappist the monk (talk) 19:02, 6 March 2016 (UTC)

While this will create problems only in cases where a page changes frequently, I agree with you that the actual |archive-url= links to snapshots should be non-ambiguous. Yes, they should use all digits provided by archive.org. I never stated anything different for the normal rendering; those suggested links were only meant as part of the error message (that is, semantically outside normal reader context).

For the same reason, we cannot fully automate the selection process (just make it as easy as possible), it still requires a human to actually check if a particular snapshot supports a statement in an article or not. So, trying to be extra-smart and silently correcting invalid links isn't an option.

However, the question remains how to achieve the goal of minimize the time for error messages to show up and how to possibly fight link rot (IMHO by proactively archiving snapshots of pages used to support articles before they become dead).

I agree with you that the citation template is not the best place to integrate editing tools, but the basic rule to keep users' and editors' views separated is already violated by displaying error messages in the normal page rendering for readers at all. So displaying a few more links in those error messages would not violate that design principle more. But perhaps the idea can be reduced to a single extra link:

• [https://web.archive.org/web/<iso date>*/<contents of url= parameter> Check nearest existing snapshots to given archive date on archive.org?]

If |archive-date= is present, this should be stuffed into <iso date> (with the count of digits depending on what was provided by |archive-date=). The pending * causes archive.org not to select the nearest snapshot but to show all snapshots matching <iso date>. The oldest and newest snapshots (optionally narrowed down by <iso date>) are available from archive.org's list, so we don't need direct links. If the page wasn't archived at all, archive.org will display the dialogue to save the current page (similar to /save/). This doesn't cover the case of saving a new snapshot for pages of which some older snapshots exist already, but it would still be helpful in the majority of other cases, whilst not disturbing the error message display much.

Regarding RefToolbar and VisualEditor. They both require client-side scripting, so they are not an option for people using browsers not supporting client-side scripting (for security policies or in general).

--Matthiaspaul (talk) 00:38, 7 March 2016 (UTC)

I have further hacked Module:Citation/CS1/sandbox so that archive.org urls in |archive-url= cause an error message when it is the save command url and when the timestamp is not 14 digits:

"Title". Archived from the original on 2016-03-05.

"Title". {{cite web}}: |archive-url= is malformed: save command (help)

"Title". {{cite web}}: |archive-url= is malformed: timestamp (help)

—Trappist the monk (talk) 13:32, 7 March 2016 (UTC)

I don't know how frequently they are used, but perhaps the "id_" / "js_" / "cs_" / "im_" appendages to the timestamp (see second doc link above) should be allowed as well.

--Matthiaspaul (talk) 01:09, 21 March 2016 (UTC)

I haven't looked at the sandbox, Trappist, but have you included uses of url=//web.archive.org/web/...? For quite some time, there was an effort to make all web.archive.org urls relative, as they weren't accessible in countries blocking https. It's now a moot point, since both Wikipedia and the Internet Archive have adopted secure protocols, but there may be leftovers, although I have no idea how many.

As for making it easier for editors to save a cited page or find an archive of it, there are handy tools available at Help:Using the Wayback Machine#JavaScript bookmarklet that allow editors to add scripts to their toolbars that make both tasks simple. Where should we propose that these archiving tools be added to citation tools; I don't think it's a function of the citation templates but also wonder if there's some help that could be given via the templates? Evidently the availability of these scripts is not well-known; nor is proactive archiving of sites much promoted, to our detriment, I think. So, is there a way that the templates can help?

—D'Ranged 1 | VT^alk :  19:24, 17 March 2016 (UTC)

I continue to maintain my suggestion to add at least a single multi-purpose link to archive.org to the error message. It does not clutter the error message more than displaying convenience information such as the position of offending characters in a parameter (as we do already), and it works also for users who do not want to enable (for security reasons) or simply cannot use JavaScript (for lack of support in their browsers). So, instead of

Title |archive-url= is malformed: timestamp (help)

we could display

Title |archive-url= is malformed: timestamp (help, check)

Note the extra "*" blindly inserted into the check link by the suggested template enhancement after the incomplete or otherwise malformed timestamp and before the target link. This "*" forces archive.org to display the selection bar for all snapshots matching the (incomplete) timestamp instead of silently selecting the nearest snapshot. The user would then check and select the desired snapshot and copy the resulting url (now with valid date string) back into the |archive-url= parameter. If the link target hasn't been archived at all but exists in the live net, archive.org will ask the user if s/he wants to save a snapshot. If the link is invalid and does not exist in the live net, archive.org will ask if the user wants to broaden the scope to other snapshots at the target site. So, this trivial enhancement would already make it much easier for editors to select the desired snapshot and fix an error. In contrast to providing "/save/" links, there is no risk for abuse, as archive.org will only ask to save snapshots if no snapshots exist (so it will happen only once) and it will not save snapshots without explicit confirmation by the user, anyway (whereas "/save/" saves snapshots without user confirmation).

A slightly smarter implementation would provide even more utility value:

This improved enhancement would append the contents of the |url= parameter to the |archive-url= parameter if the template detects that the given |archive-url= does not contain a link target already. If the timestamp is not present at all, but |archive-date= is given, it would use the contents of |archive-date= to create an incomplete timestamp to be implanted into the resulting check url link. The extension would still insert the "*" as described further above.

This enhancement would allow content editors to just add |archive-url=http://web.archive.org/web/ (and optionally some form of |archive-date=) to the citation template, invoke edit preview, and click on the "auto-completed" check link in the resulting error message in order to select the proper snapshot at archive.org and copy the resulting url back into the |archive-url= parameter before saving their contribution. It would thereby making it trivially easy to select and/or create archive snapshots without external tools - while the "user interface" would remain completely unobtrusive (a single link in an error message typically displayed only in preview mode) and allow no abuse at all.

--Matthiaspaul (talk) 01:09, 21 March 2016 (UTC)

harvnb doesn't work correctly within the |in parameter

When using {{harvnb}} within the |in= parameter of {{cite book}}, the page is incorrectly appended following the word "and", while there should be just a comma. Outside {{cite book}} it works fine. See these examples:

Code

{{harvnb|DNC|2016|pages=100–159}}

Example 1 – {{harvnb}} used inside {{cite book}}

• Clinton, Hillary (2016). Who would trust a secret Neocon?. {{cite book}}: Invalid |ref=harv (help); Unknown parameter |in= ignored (help)

Example 2 – {{harvnb}} used outside {{cite book}}

DNC 2016, pp. 100–159

Unfortunately I'm not enough into the whole CS1 module to find the bug, but am hopeful someone else does. Regards, PanchoS (talk) 18:14, 21 March 2016 (UTC)

|in= is an alias of the parameter |language=.

The {{harvnb}} parameter doesn't belong inside {{cite book}}. Instead, do this:

{{cite book |last=Clinton |first=Hillary |year=2016 |title=Who would trust a secret Neocon? |ref={{sfnref|DNC|2016}}}}

Clinton, Hillary (2016). Who would trust a secret Neocon?.

—Trappist the monk (talk) 18:27, 21 March 2016 (UTC)

Usually, if the author is known, |ref=harv is the proper way to create a link from a shortened form to the full citation. It can be confusing to the reader to have differing author-date combinations for the two, which is why a non-"harv" value for |ref= is recommended when the author is unknown or cannot be determined. It may be that you are trying to cite a work (Clinton) that was included in another work (DNC). There are native template parameters that can handle these situations. 72.43.99.130 (talk) 19:01, 21 March 2016 (UTC)

Red text in citation tells me to delete math formula from quote

In the citation (from Alphabet (formal languages))

• {{cite book |last1=Ebbinghaus |first1=H.-D. |last2=Flum |first2=J. |last3=Thomas |first3=W. |date=1994 |title=Mathematical Logic |edition=2nd |publisher=[[Springer Science+Business Media|Springer]] |location=[[New York City|New York]] |isbn=0-387-94258-0 |page=11 |quote=By an ''alphabet'' <math>\mathcal{A}</math> we mean a nonempty set of ''symbols''. |url=http://www.springer.com/mathematics/book/978-0-387-94258-2}}
• Ebbinghaus, H.-D.; Flum, J.; Thomas, W. (1994). Mathematical Logic (2nd ed.). New York: Springer. p. 11. ISBN 0-387-94258-0. By an alphabet ${\mathcal {A}}$ we mean a nonempty set of symbols.

I see the red error text "delete character in |quote= at position 20 (help)" at the end of the citation. Position 20 of the quote is the math formula, which should not be deleted (it is an essential part of the quote), and formatting it as a math formula is necessary in order to accurately convey the quote (the fact that it is shown as a script font is a meaningful piece of mathematical notation and changing the font would change the meaning). So the error message is itself erroneous. In addition this is putting the article into the "CS1 errors: invisible characters" error category. I assume this is something to do with the wikimedia math formatting, since I don't think this passage actually has any invisible characters (I tried copying and pasting into an editor that would show me the fnords, and then back to here again, but this made no change, and I get the same behavior whenever I have math in a quote even when I type it myself with no invisible characters). The same problem also happens with math in titles; see e.g. two examples in squared triangular number. Can this be fixed, please? —David Eppstein (talk) 23:37, 14 March 2016 (UTC)

This may be related to recent changes to how wikimedia handles math, in order to fix https://phabricator.wikimedia.org/T103269 —David Eppstein (talk) 23:55, 14 March 2016 (UTC)

This is related to the stripmarkers issue. MediaWiki replaces <gallery>...</gallery>, <math>...</math>, <nowiki>...</nowiki>, <pre>...</pre>, and <ref>...</ref> tags with strip markers. The first and last characters in a stripmarker are delete characters. The error message identifies the first delete character in the strip marker. Here is a highly simplified version of your citation as the module renders it for further processing by MediaWiki:

'"`UNIQ--templatestyles-0000002A-QINU`"'<cite class="citation book cs1">''Mathematical Logic''. <q>By an ''alphabet'' '"`UNIQ--math-00000029-QINU`"' we mean a nonempty set of ''symbols''.</q></cite><span title="ctx_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Abook&rft.genre=book&rft.btitle=Mathematical+Logic&rfr_id=info%3Asid%2Fen.wikipedia.org%3AHelp+talk%3ACitation+Style+1%2FArchive+13" class="Z3988"></span>

Highlight and copy the text from the second apostrophe following ''alphabet'' (the last part of the italic markup) to the 'w' in we and paste it into the green box a this unicode decoder. Click the 'Convert' button and look at the content of the 'Percent encoding for URIs' box. You should see something like this:

%20%7F%27%22%60UNIQ--math-00000006-QINU%60%22%27%7F%20

That's the stripmarker. In the old days of just last week, it would have looked like this:

%20%7FUNIQ--math-00000006-QINU%7F%20

But, earlier this week MediaWiki changed how it renders stripmarkers and so made it unrecognizable to the code that used to skip over <math>...</math> tags. The issue is fixed in the sandbox:

Ebbinghaus, H.-D.; Flum, J.; Thomas, W. (1994). Mathematical Logic (2nd ed.). New York: Springer. p. 11. ISBN 0-387-94258-0. By an alphabet ${\mathcal {A}}$ we mean a nonempty set of symbols.

and from squared triangular number:

Benjamin, Arthur T.; Orrison, M. E. (2002), "Two quick combinatorial proofs of $\textstyle \sum k^{3}={n+1 \choose 2}^{2}$" (PDF), College Mathematics Journal, 33 (5): 406–408.

—Trappist the monk (talk) 00:23, 15 March 2016 (UTC)

Thanks! So as I suspected, it was something triggered by a change to Wikimedia. This is only visible because I've changed by rendering preferences to make these things visible, not by default, right? So putting it into the sandbox and waiting until the next major batch of updates, as you're doing, rather than rushing to get a fix out, seems like the right thing to do. —David Eppstein (talk) 16:18, 15 March 2016 (UTC)

I'm not sure because I'm not sure that I understand your question. If you mean the preferences at Special:Preferences#mw-prefsection-rendering, then no, none of those math options prevent the delete character error message. If you mean that you have been intentionally hiding all cs1|2 error messages by using the css specified at Help:CS1 errors#Controlling error message display then yes, because this error message is visible to all readers unless they specifically hide cs1|2 error messages.

—Trappist the monk (talk) 16:54, 15 March 2016 (UTC)

Ha! That's a cite and quote I added. Small world. Jason Quinn (talk) 17:57, 22 March 2016 (UTC)

Citing software

I vaguely remember seeing a template to cite software, but can't find it. The key thing that seems missing is a parameter to give the version number, which is often critical for software. Jc3s5h (talk) 20:01, 23 March 2016 (UTC)

I would use {{citation}} and |edition=v. xx.xx. 72.43.99.130 (talk) 20:39, 23 March 2016 (UTC)

Well, scratch |edition=. Citation includes |version=. 72.43.99.130 (talk) 20:41, 23 March 2016 (UTC)

That seems to work, thanks Jc3s5h (talk) 22:25, 23 March 2016 (UTC)

Requesting edit—there's a typo on the "Template:Cite web" page

Hi, I noticed a typo in the "Template:Cite web" article when I was reading it for help. I'm not sure if this is the correct place to be requesting edits, but I'd thought I'd let anyone who is able to edit the article know about the typo.

The typo is right under the "Choosing between {{Cite web}} and {{Cite news}}" subheading. Currently, it states, "Before 2014, editors needed to decided whether to use {{Cite web}} or {{Cite news}} based on their features." The word "decided" should be changed to "decide" no? 104.10.252.77 (talk) 04:20, 28 March 2016 (UTC)

 Fixed. Thanks. – Jonesey95 (talk) 04:43, 28 March 2016 (UTC)

Yes, thanks for catching this. Also, for future reference, unlike the template itself, the template documentation is not locked — anyone, even not-logged-in users, can edit it. —David Eppstein (talk) 04:45, 28 March 2016 (UTC)

Protected edit request on 26 March 2016 - LIBRIS

The "Cite book" template misses any suitable parameter for books that uses the Swedish LIBRIS system. These codes looks like 2219566. My suggestion is a parameter in accordance with the MARC Code for the Swedish Union Catalog "selibr". Ie .. Cite book | selibr=2219566 Ferrofield (talk) 21:25, 26 March 2016 (UTC)

What do you mean by misses any suitable parameter for books that uses the Swedish LIBRIS system?

—Trappist the monk (talk) 21:33, 26 March 2016 (UTC)

This is, I suspect, another attempt at Template talk:Citation#Swedish LIBRIS database. --Redrose64 (talk) 21:53, 26 March 2016 (UTC)

Ah, that. Now, there are 50 transclusions of {{LIBRIS}} and 246 pages that use the libris url (insource:http://libris.kb.se/bib/).

—Trappist the monk (talk) 00:10, 27 March 2016 (UTC)

Is there any parameter that can harbor free text?Ferrofield (talk) 22:37, 27 March 2016 (UTC)

For a libris identifier? |id=

—Trappist the monk (talk) 22:41, 27 March 2016 (UTC)

Yes, |id={{LIBRIS|2219566}} .Imzadi 1979 → 22:42, 27 March 2016 (UTC)

Thanks. I found it though by the hint of LIBRIS and what links here.. But it would be perhaps a good idea if the Cite book template documentation spells this out? and the LIBRIS template documentation is in essence empty. Ferrofield (talk) 23:00, 27 March 2016 (UTC)

With ~300 uses, that's above the threshold of what we typically implemented for identifiers (see Template_talk:Citation/Archive_4#Implementation). I would support adding this identifier. Headbomb {talk / contribs / physics / books} 12:52, 29 March 2016 (UTC)

Question: Which Template to use.

Where should I ask about which template to use if it is not clear from the documentation in the article? I'm specifically interested in how to cite a regulatory agencies FAQ's and other published opinions. I have seen these variations (the first four are for the exact same source):

1. found here
2. found: Genetically_modified_food's lede.
3. found Genetically_modified_food#Definition
4. found: Genetically_modified_crops's lede
5. found: Genetically_modified_organism#Mosquitoes

References

I prefer style #1 and #2's order, because it makes clear the author is the World Health Organization, and is consistent with MLA's style here, but I found no template for it. --David Tornheim (talk) 06:29, 29 March 2016 (UTC)

Perhaps a bit of a kludge, but one could replace |publisher=World Health Organization with |author=World Health Organization in {{cite web}}. Boghog (talk) 07:04, 29 March 2016 (UTC)

References

Thanks! Good idea. --David Tornheim (talk) 09:34, 29 March 2016 (UTC)

Except the Publisher isn't the Author, and MLA doesn't do it like you think it does. In this case, the author is unknown, even if he is known to be working on behalf of the WHO. MLA leaves the publisher after the title of the article (regardless whether the author is known or not)--see "A Page on a Web Site" For an individual page on a Web site, list the author or alias if known, followed by the information covered above for entire Web sites. Remember to use n.p. if no publisher name is available and n.d. if no publishing date is given. where the information above is Editor, author, or compiler name (if available). Name of Site. Version number. Name of institution/organization affiliated with the site (sponsor or publisher), date of resource creation (if available). Medium of publication. Date of access. --Izno (talk) 11:18, 29 March 2016 (UTC)

I looked and everything agreed with the first MLA citation style for I gave here where Agency name goes where the author is, even Purdue's OWL website that you cited agreed [1]:

"Online Sources: ...Authors for websites are often corporations, or governments."

"Entire Website: United States Environmental Protection Agency. Drinking Water Standards. EPA, 8 July 2004. Web. 24 Jan. 2006"

(emphasis added)

The other universities had similar responses on how to handle government issued reports, etc.: Radford, EasyBib, Cornell, Walsh, UNF.

Columbia State had a similar response [2]:

"Most government documents do not list a personal author. The agency that issues a government publication is usually considered the author."

This one was a bit more complicated: Univ. Nevada, Reno

Please let me know if you agree that calling the EPA the author is okay. --David Tornheim (talk) 05:08, 30 March 2016 (UTC)

It depends on the citation style established in the article, or, in the absence of an established style, on the style you want to use. CS1 is not MLA. It is being developed (one hopes) primarily for the needs and expertise of Wikipedia readers and editors. The guideline for CS1 is to not substitute publisher for author, even when the latter is unknown. Imo, this is the correct approach for citations in general, regardless of style. 65.88.88.174 (talk) 14:55, 30 March 2016 (UTC)

I'm surprised this is not more straightforward and established, given how common government documents must be in citations. I will agree with you that Wikipedia citations are not MLA, especially since MLA does not even require a URL. And I won't contest the guidelines for proper use of this or the other templates are without more information. My feeling is the templates support the use of citations rather than dictate what form the citations should take in the final product. So, as in my original question, perhaps this is just the wrong template to use for producing the correct citation form.

The definition of "author" and "publisher" is not so straightforward, without further evidence of an established definition on Wikipedia. Part of the reason I have been inclined to the form with Government Agency first (whether by using a template or otherwise) is because, unlike a book, a journal article, or conference presentation, the agency is taking responsibility for it, not whatever specific persons worked on preparing it. In fact, those who prepared the first version of the FAQ above might have been fired or reassigned and someone else became responsible for it--but these things are not transparent to the readers--by design. The agency often takes responsibility and credit for the document, not the author(s). That's one of the reasons the document has more authority and weight than other sources. So, it makes sense to me to put that first, rather than have the first thing be the title, which could be authored by anyone. So, I tend to agree with MLA that the author is the agency, even though this does not meet our normal understanding of an author.

As regards to publisher, this is not so straightforward either. My understanding of publisher is an entity like Random House or the Government Printing Office. These entities are somewhat independent of the author of the documents. If you look at some of the MLA references I gave you can see that they distinguish this [3]:

United States. Cong. House. Committee on House Administration. Bipartisan Campaign Reform Act of 2001. 107th Cong. 1st sess. H. Rept. 131, pt. 1. Washington: GPO, 2001. The Library of Congress. Thomas. Web. 12 January 2012.

The U.S. Congress is the author and the GPO is the publisher. This is less clear for documents that are self-published on the web, including the FAQ I presented.

I would like to bring this issue to the appropriate forum for dealing with standardization of Wiki-references, assuming we have one. After reviewing the MLA guidelines at the various universities, I am more firm in my belief that it makes more sense for us to follow MLA in putting the agency name first--as the "author"--than when I first wrote the post. Yet, I might agree that using an agency for author with this template may be a bad idea based on whatever assumptions the template has about the form of authors. I would like to know where to get the form I am suggesting confirmed, and how to use a template--if it exists--to accomplish the goal of getting it to produce as the end product with government agency first, as "author" and not misuse the template. I do think it is better to use a template whenever possible so our bots can review and check the citations, etc., which is why I came here rather than just doing the citation by hand in the form that I think is best. --David Tornheim (talk) 18:39, 30 March 2016 (UTC)

I believe the distinction between publisher and author is much clearer. The author is the person or entity that creates the work. The publisher is the distributor of the work. Congressional committees author laws, so by rights the specific (by year) committee is the author. These may or may not be published (pending approval by the entire congress). If they are published, they are citable as legal documents, and the publisher of first instance is GPO. In the case of the WHO faq, this should be treated as any other work produced by staff writers (unless you know that it was produced by a specific department/section of WHO). The guideline for staff- (or unattributed) works as far as CS1 is concerned, is to just mention the publisher, with the implicit understanding that in the absence of named author(s), the publisher assumes editorial (and legal) responsibility, much the same way articles without a by-line in news sources are considered as fully backed by the publisher. In filling a CS1 template, it is recommended that as a courtesy, these contingencies are commented, e.g. |author=<!--Staff writer(s); no by-line-->. 72.43.99.130 (talk) 20:40, 30 March 2016 (UTC)

I appreciate the clarity of your sentences and explanation for proper use of the CS1 template. Your example of Congress makes sense. Your definition of publisher makes sense also. I can't agree that it is appropriate to compare a FAQ on the WHO website as comparable to an editorial in a major newspaper. Newspapers often endorse political candidates or propositions in their editorial sections. I cannot even imagine we could use that as WP:RS most of the time. It is true that the paper is backing the opinion and taking legal responsibility if they lie or for libel, but the kind of opinion they state is of a very different form than something the WHO would publish and stand behind, especially when it is stated in form that is intended to be read as factual rather than opinionated. My sense is the CS1 template may not be designed for this particular application of RS. If so, that leaves the question of which template is most appropriate for a government agency. I am just surprised this question has not come up before and that it is not documented somewhere, like it is for MLA. --David Tornheim (talk) 08:28, 31 March 2016 (UTC)

I didn't read anything in IP editor's comments that made any reference to a newspaper's right to take an editorial position. [A]rticles without a by-line in news sources are considered as fully backed by the publisher says nothing about opinion. The New York Times commonly did not attribute articles to a particular author: example. No doubt other newspapers did, and perhaps still do, the same.

cs1|2 is not MLA, is not Chicago, is not ALA, nor any other TLA. It is cs1|2 and is an amalgam of all of those with our own ideas tossed in. In cs1|2 the parameters are intended to hold the information appropriate to the parameter name: |author= gets the author; |publisher= gets the publisher, |date= gets a date, etc. Beneath the bonnet, these parameters and their data are made available as metadata to reference management tools. It is important that that information be correct.

The advice quoted by IP editor was added to Help:Citation Style 1#Authors with this edit which suggests to me that, indeed, handling of sources without identifiable authors has been considered and that an acceptable method has been adopted.

—Trappist the monk (talk) 10:58, 31 March 2016 (UTC)

Thank you for pointing to this: Help:Citation Style 1#Authors, which says:

Editors should use an |author= organizational citation when the cited source, such as a committee report, specifically names an official body or a sub-unit of the publisher as the collective author of the work, e.g. |author=Commission on Headphone Safety or |author=Rules Sub-committee.

When using |author= avoid citations like {{cite news|work=Weekday Times|author=Weekday Times editors|title=...}}, unless the article is on a field in which the majority of professional journals covering that field use such a citation style.

This seems consistent with what MLA says about government organizations, and that hence putting World Health Organization as the author would be appropriate in this context. And the scholarly documents put entities like the WHO, FAO and FDA as the author almost all the time. For the particular document above: [4], [5]. More examples in GM literature where organization is first (as the author): see citation 78; (These four are from the same document: [6], [7], [8],[9]);[10]; see citation 16;see citations 47-52. So can we all agree that making the WHO or World Health Organization the author is appropriate in this case as being consistent with the instructions for this template? --David Tornheim (talk) 10:31, 2 April 2016 (UTC)

I looked at the FAO link you provided above. As is typical with many government organizations, their document repository links are lagging, many having not been updated. I was able to open a pdf document through their ftp server, and FAO's citation is incorrect as far as cs1|2 is concerned:

FAO/WHO. 2000. Safety aspects of genetically modified foods of plant origin. Report of a Joint FAO/WHO Expert Consultation on Foods Derived from Biotechnology, Geneva, Switzerland. 29 May-2 June 2000 (available at ftp://ftp.fao.org/es/esn/food/gmreport.pdf; accessed March 2004)

The authors in this instance are known (there's a list of the reporting experts on pages 26–27), and the publisher should not be used instead of author per cs1|2. Imo, the proper cs1 template would be {{cite techreport}}. 72.43.99.146 (talk) 15:24, 2 April 2016 (UTC)

Although I agree the work itself appears to be a "technical report", and indeed I saw the list of authors at the end of the report, I searched to see how that report is cited in the literature. With this Google search, every single one of the first five references was in a form similar to the above, with FAO and/or WHO as the author. That appears to be the proper way to cite it in this field. These are the first five that came up: [11],[12],[13],[14],[15] --David Tornheim (talk) 18:42, 2 April 2016 (UTC)

{{cite techreport}} was suggested under the assumption that cs1 would be used. As it is stated in its doc, it is appropriate for a variety of non-peer-reviewed scientific (or quasi-scientific "expert") literature. As has also been stated above, cs1 is not mandatory; you could use the citation style of your choice, taking into account any pre-existing contributions to the article. 65.88.88.127 (talk) 21:02, 2 April 2016 (UTC)

duplicated styling

In playing around with the code to support |dead-url=usurped no archive, I discovered that the styling used for the bad url error message is being added twice:

'"`UNIQ--templatestyles-0000004B-QINU`"'<cite class="citation web cs1">[http//www.exampl.com "Title"].</cite><span title="ctx_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Abook&rft.genre=unknown&rft.btitle=Title&rft_id=http%2F%2Fwww.exampl.com&rfr_id=info%3Asid%2Fen.wikipedia.org%3AHelp+talk%3ACitation+Style+1%2FArchive+13" class="Z3988"></span> <span class="cs1-visible-error citation-comment"><code class="cs1-code">{{[[Template:cite web|cite web]]}}</code>: </span><span class="cs1-visible-error citation-comment">Check <code class="cs1-code">&#124;url=</code> value ([[Help:CS1 errors#bad_url|help]])</span>

I have fixed that with a minor change to Module:Citation/CS1/Configuration/sandbox:

'"`UNIQ--templatestyles-0000004D-QINU`"'<cite class="citation web cs1">[http//www.exampl.com "Title"].</cite><span title="ctx_ver=Z39.88-2004&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Abook&rft.genre=unknown&rft.btitle=Title&rft_id=http%2F%2Fwww.exampl.com&rfr_id=info%3Asid%2Fen.wikipedia.org%3AHelp+talk%3ACitation+Style+1%2FArchive+13" class="Z3988"></span> <span class="cs1-visible-error citation-comment"><code class="cs1-code">{{[[Template:cite web|cite web]]}}</code>: </span><span class="cs1-visible-error citation-comment">Check <code class="cs1-code">&#124;url=</code> value ([[Help:CS1 errors#bad_url|help]])</span>

—Trappist the monk (talk) 13:17, 4 April 2016 (UTC)

Cite magazine

{{Cite magazine}} doesn't get a mention on the help page itself and the {{Cite journal}} use description includes magazines. Does this need updating or is there a reason? -- Cavrdg (talk) 19:30, 4 April 2016 (UTC)

I've added {{cite magazine}} to the table and tweaked |work= to include it.

—Trappist the monk (talk) 19:42, 4 April 2016 (UTC)

RfC: Placement of Internet Archive URLs in cite templates

There are some editors who say archiveurl should not be used. Archive URL's should be placed directly in the url argument replacing the original URL. There is an open RfC on this question. -- GreenC 14:24, 6 April 2016 (UTC)

Cosmetic difference: |access-date= vs. |archive-date= output.

Markup	Renders as
{{cite book|title=Title|url=http://dummy-url.com|archive-url=http://archive-of-dummy-url.com|access-date=2016|archive-date=2016}}	Title. Archived from the original on 2016. Retrieved 2016. {{cite book}}: Check date values in: |access-date= and |archive-date= (help)

Should they both render the same? (Either with or without "on" before the date). 72.43.99.130 (talk) 17:56, 4 April 2016 (UTC)

It's not a problem if the parameters are used properly, that is, with full dates. Partial dates are meaningless in these contexts. --Redrose64 (talk) 12:42, 5 April 2016 (UTC)

No there is still a difference in display:

• Title. Archived from the original on 2016-04-05. Retrieved 2016-04-05.

The archive date is preceded by "on". The access date is not. 72.43.99.146 (talk) 13:49, 5 April 2016 (UTC)

Oh. I thought you were complaining about the grammatical error of "on 2016". --Redrose64 (talk) 20:21, 5 April 2016 (UTC)

Well, not really complaining. I hope it didn't come across as whining, it's no big deal. 72.43.99.130 (talk) 17:27, 6 April 2016 (UTC)