Template talk:Reflist/Archive 25

Archive 20

←

Archive 23

Archive 24

Archive 25

Archive 26

Archive 27

→

Archive 30

Deprecating fixed number of columns

I would like to propose removing the capability to specify a fixed number of columns (like with |2 or |3) and move to width-based metrics (like |30em) for all of our multi-column needs. This appears to be the consensus for what will be implemented for the <references /> upgrade based on: [1]. The main reason is that narrow screens can't display 2 or 3 or more columns comfortably; they should be able to drop columns to fit the screen size. For example, anything that has 2 columns is slightly awkward to read on my Android phone, and anything with 3 or more is downright ugly. Another benefit is that very wide screens can make additional columns, resulting in a column width that remains easier to read, and saving space.

We can use [2] as a guide to which articles are using the feature I propose to deprecate; it looks like about 1400 articles are currently doing so, which is a number which can be fixed manually. (Or a bot could do it.) What do others think? -- Beland (talk) 20:49, 13 March 2014 (UTC)

Is anyone going to make this work across all languages? If not I oppose. Doc James (talk · contribs · email) (if I write on your page reply on mine) 22:04, 13 March 2014 (UTC)

I don't understand that, as there is no standard implementation of this template across the various language versions. And many, such as the German Wikipedia don't have this template. --  Gadget850 ^talk 23:15, 13 March 2014 (UTC)

I think he's talking about Bugzilla:51260, which talks about adding the columns feature into the Cite extension. — Edokter (talk) — 23:28, 13 March 2014 (UTC)

I understand Beland's reference, which I replied to below. But I don't understand Doc Jame's comment "make this work across all languages." We have no purview over the other Wikipedias, the reflist template is not consistently implemented across the various Wikipedias, and many don't have a version of reflist.

• You link to {{columns}}, which is not used for references (it's a table-based column template). By far the most uses for reflist already use the column-width functionality, and the few that don't can be edited, either by hand or bot. But I don't see a reason to deprecate column-count; it is merely a option, sometimes usefull, to tell the browser how to render columns. — Edokter (talk) — 23:23, 13 March 2014 (UTC)

• I would argue the missing reason for disallowing column-count is that it doesn't work on narrow screens, like smartphones. Only one-column layout works for those screens, so I would argue good editors would either never use columns or only use options that automatically determine the number of columns based on the screen size. -- Beland (talk) 21:05, 18 March 2014 (UTC)

• Oppose I edit across 60 languages and {{reflist}} appears to work in all of them. {{reflist|30em}} doesn't and {{reflist|2}} at least doesn't break anything. Doc James (talk · contribs · email) (if I write on your page reply on mine) 23:25, 13 March 2014 (UTC)
• Neutral appears issues fixed. Doc James (talk · contribs · email) (if I write on your page reply on mine) 22:45, 14 March 2014 (UTC)
  • The Guaraní Wikipedia has a different implementation of {{reflist}}. You need to use {{reflist|colwidth=30em}}. The Guaraní also supports a |title= parameter that we do not have here, but it does not have the fix for orphans/widows. --  Gadget850 ^talk 01:58, 15 March 2014 (UTC)
• And given the state of older and more serious bugs on the Cite bug list, I would not put a lot of faith in T53260 being implemented any time soon. --  Gadget850 ^talk 23:37, 13 March 2014 (UTC)
• Support Using {{reflist|width=30em}} {{reflist |colwidth=30em}} - which is best practice - wouldn't break anything when translated onto other wikis either. The continued use of two fixed columns looks ridiculous both on my phone and on my 1920px-wide monitor. With 2560px panels becoming more available, we really need to move towards columns based on a width in ems that will adjust easily no matter the resolution of screen in use. @Edokter: Unless we actually deprecate a parameter, we do not indicate that a better method is preferred and we end up repeating the same arguments time-and-again to editors who don't realise that their screen width isn't the standard for the whole world. --RexxS (talk) 22:09, 14 March 2014 (UTC)

It does appear that the issues around the 30em bit has been fixed. It doesn't break it is just not do anything [3] which is fine with me. Doc James (talk · contribs · email) (if I write on your page reply on mine) 22:43, 14 March 2014 (UTC)

RexxS, I don't like removing options that could be usefull in certian places. Shall we set up a tracking category (or better method) to see how often fixed columns are actually used? — Edokter (talk) — 23:17, 14 March 2014 (UTC)

@Edokter: I completely agree with your philosophy about removing options, but I'm at a loss to see where |2 could be more useful than e.g. |width=30em. Nevertheless I'd be happy to be enlightened if we can find any instances. There's no rush because this is a niggle, not a major problem, so let's do the tracking and see. --RexxS (talk) 00:56, 15 March 2014 (UTC)

Apologies to James, the correct parameter is |colwidth=, but I checked and it doesn't break anything on gn-wiki either. --RexxS (talk) 19:39, 15 March 2014 (UTC)

The cat already is picking up 23,526, thats a lot of editors choosing to use that option. Think its likely still populating so will be interesting to see the final figure, however surely this would need wider consensus before removing that option.Blethering Scot 23:04, 15 March 2014 (UTC)

• Support the concept but Oppose the default of 30em which is a tad too small on many devices and causes the second column to appear in too small of vertical space. Using 33em seems to work better on most of those devices. Vegaswikian (talk) 23:27, 22 March 2014 (UTC)

Tracking

Added tracking category Category:Articles using fixed number of columns in reflist to {{reflist}} and {{refbegin}}. Let's see in 30 days what comes up. — Edokter (talk) — 01:51, 15 March 2014 (UTC)

At least one blank line is now appearing after the list of references. I presume it is due to this change. Please correct it. Thanks. Nurg (talk) 02:58, 15 March 2014 (UTC)

I noticed it too, so I fixed it. --Redrose64 (talk) 08:04, 15 March 2014 (UTC)

Count has been steady for the last week and is now at 265,671 (that's about 17% of all 4.5M articles). I think we can remove the tracking categories. — Edokter (talk) — 01:24, 10 May 2014 (UTC)

Deprecate

What do we me by deprecate here? Would a bot go through 26k articles and update the parameter? Or would we update the template to convert a column value to a column width? --  Gadget850 ^talk 01:40, 16 March 2014 (UTC)

I prefer the bot, or simply let it 'break', so authors will fix the template use. Changing the meaning of "2" to mean "30em" was tried before, and it was not received well at all, so any static number should be ignored. How would people feel about making "30em" the default? — Edokter (talk) — 01:46, 16 March 2014 (UTC)

support 30em as the default value would go a long way toward improving the presentation of reflists; then a bot could be used to remove instances of col=2, etc. --User:Ceyockey (talk to me) 14:01, 16 March 2014 (UTC)

What is meant by default? That 30em is applied as the default, regardless of the number of references (which we cannot detect)? That is no going to be seen as visually appealing where there are one or two references. --  Gadget850 ^talk 12:54, 17 March 2014 (UTC)

Good point. In which case, "|2" and such should do nothing, ie. no columns. — Edokter (talk) — 13:07, 17 March 2014 (UTC)

Yes, I do mean 30em as the default in the absence of a parameter; not being visually appealing is a matter of debate. It is not uncommon for an article to have 0 or 1 reference, but that is not the desired norm - it is an indication of the poor sourcing that wikipedia has in some areas. Consider the lack of visual appeal in these cases as a sign of an underdeveloped article -- an indicator the work needs help, not unlike the more informative but less in-your-face stub-templates --User:Ceyockey (talk to me) 13:18, 17 March 2014 (UTC)

In which case, Oppose - where an article has primarily or exclusively shortened footnotes in the reflist, like NBR 224 and 420 Classes, narrow columns are beneficial. But where an article has primarily or exclusively full-length citations in the reflist, like Ipswich railway station, narrow columns are detrimental. --Redrose64 (talk) 17:54, 17 March 2014 (UTC)

Here is an example of a page with a 30em reflist - it has six entries. For me, there are three columns, and the third column consists of the single word "or" and a closing parenthesis. --Redrose64 (talk) 23:43, 17 March 2014 (UTC)

Curious what your browser is. I only get two columns, no matter how wide. The "or)" should not break off. — Edokter (talk) — 23:48, 17 March 2014 (UTC)

Firefox 27.0.1 - where I've just noticed that in Gadget850's example below, the "edit" link is similarly orphaned to a third column. --Redrose64 (talk) 23:58, 17 March 2014 (UTC)

Firefox still needs to implement it. — Edokter (talk) — 00:49, 18 March 2014 (UTC)

In Chrome, 1920px wide, I get refs 1-4 in col 1; ref 5 in col 2; ref 6 in col3. But when I lessen the window width I get two columns which remains even when I go back to 1920px wide. Or are we being too optimistic by expecting <ref>...</ref> tags to work properly with {{harv}} nested inside them? --RexxS (talk) 01:17, 18 March 2014 (UTC)

With Firefox 27, a single citation always gets orphaned across the columns, whereas two or more do not. More examples added. --  Gadget850 ^talk 02:19, 18 March 2014 (UTC)

I'm pretty sure FF27 is capable of orphaning two citations across multiple columns just as easily - see screenshot below. Nevertheless I'm convinced the problem above in Devon heraldry is an artefact of trying to nest references improperly. --RexxS (talk) 14:08, 18 March 2014 (UTC)

In that example I get the same weirdness but 33em, as I suggested above, works just fine. Vegaswikian (talk) 23:31, 22 March 2014 (UTC)

"deprecate" can also mean that it gets added to AWB and bots that do general cleanups, rather than having a bot edit all the articles. Anomie⚔ 11:29, 17 March 2014 (UTC)

I wasn't proposing that we change the default one-column layout if no parameters are supplied; that's actually my preferred layout, so I would oppose that as well. 8) I was thinking we could edit (either by hand over time or by bot - bot would make a lot of sense if it really is 26K) all the articles that currently use the fixed-column layout into the variable-column layout, and then once the number of articles is zero, either neuter the fixed-column parameter so it shows one column, or cause it to print an error so editors can learn about the new way of doing things. -- Beland (talk) 21:00, 18 March 2014 (UTC)

Examples

Here is an example at standard and at 30em:

--  Gadget850 ^talk 16:03, 17 March 2014 (UTC)

Two citations at 30em:

Six citations at 30em:

--  Gadget850 ^talk 02:16, 18 March 2014 (UTC)

Colwidth 30 in Firefox27 at ~2200px

Firefox 27 is quite happy to break two citations across columns as well as breaking one citation across multiple columns - with a near-comical orphaning of the [edit] link as shown in this screenshot at around 2200px wide. IE11 behaves almost identically. Chrome and Opera behave much better, never breaking a ref across columns, with Chrome sometimes placing the second ref of two in the third column. If we are going to accept that readers will view references at very wide-screen resolutions, then to accomodate FF27 and IE11 we ought to be advising against using columns when there are less than (perhaps) four references. --RexxS (talk) 13:58, 18 March 2014 (UTC)

Agreed. Forget the default. The code for avoiding breaks in columns is still experimental. I thought Firefox would repond to it, but it apparently does not (at least with one item). — Edokter (talk) — 16:35, 18 March 2014 (UTC)

Personally I never go to multi-columns unless there are at least ten refs, or the article uses Shortened footnotes exclusively. --Redrose64 (talk) 18:32, 18 March 2014 (UTC)

Mobile CSS

Image demonstrating how a fixed 2-column reflist shows on Mobile

Having done some testing, this is a serious problem on Mobile. Now, we can deprecate fixed columns alltogether (which I think is the correct thing to do in the long term), or I can add a CSS rule for Mobile to disable fixed columns. — Edokter (talk) — 12:37, 25 March 2014 (UTC)

Proof, please of this "serious problem". Thanks. Lugnuts ^{Dick Laurent is dead} 12:41, 25 March 2014 (UTC)

If anybody wishes to test that example on their mobile device, see here. --Redrose64 (talk) 15:43, 25 March 2014 (UTC)

I take the silence as a no. So there is no problem. Lugnuts ^{Dick Laurent is dead} 13:16, 26 March 2014 (UTC)

Discussion

I am going to disable column-count on Mobile (but not column-width). In the mean time, let's round up opinions to determine if the column-count option should be removed from {{reflist}} and {{refbegin}}. — Edokter (talk) — 20:03, 26 March 2014 (UTC)

Solution to a problem that doesn't exist in the first place

To quote the OP - "For example, anything that has 2 columns is slightly awkward to read on my Android phone, and anything with 3 or more is downright ugly." Then stop looking at WP on a phone. Lugnuts ^{Dick Laurent is dead} 08:51, 24 March 2014 (UTC)

Corrected header, and noting that above comment should not in any way be considered a constructive contribution to this discussion. — Edokter (talk) — 09:46, 24 March 2014 (UTC)

Corrected it back - please don't change my comments again. You're missing the point. If one editor has a problem on their phone, why is that everyone elses problem? Is it worth changing 150,000+ pages to appease that one person? The correct answer is no, BTW. Lugnuts ^{Dick Laurent is dead} 10:00, 24 March 2014 (UTC)

Do you really think only one person has a problem with this? Mobile use is on the rise, check the stats. Your comment is as ignorant as it is non-constructive, and I reserve the right to remove such comments in the future. — Edokter (talk) — 14:24, 24 March 2014 (UTC)

No you don't. You need to assume good faith and stop crying. Your attitude is terrible. Lugnuts ^{Dick Laurent is dead} 14:25, 24 March 2014 (UTC)

Mobile phone users make two billion pages views of Wikipedia per day. That's a bit more than one editor, isn't it? --RexxS (talk) 20:19, 24 March 2014 (UTC)

But only one of them is actually bothered about it. One. Out of two billion. What an utterly trivial matter. The needs of the many, outweigh the needs the of the few. Lugnuts ^{Dick Laurent is dead} 11:33, 25 March 2014 (UTC)

I also didn't consider this feature "broken". Is there evidence (e.g. screenshots) that it is? Lesion (talk) 10:05, 25 March 2014 (UTC)

“ ...or simply let it 'break', so authors will fix the template use...--Edokter (talk) ”

And agree with Lugnuts that your attitude is concerning. Suggest not stick a pointless category on so many articles until some consensus has been reached. Lesion (talk) 10:10, 25 March 2014 (UTC)

Then stop looking at WP on a phone... Please tell me how that remark is of no concern, and by any standard constructive. Ignoring close to half our readers with a comment like that is ignorant, patronizing, arrogant and not deserving any serious considerarion. Please consider the root of my attitude before jumping to conclusions. — Edokter (talk) — 11:17, 25 March 2014 (UTC)

"before jumping to conclusions" Irony. You shit it. Lugnuts ^{Dick Laurent is dead} 11:34, 25 March 2014 (UTC)

Evidence this is something is actually broken please. Lesion (talk) 11:22, 25 March 2014 (UTC)

Any article listed in Category:Articles using fixed number of columns in reflist, viewed on a phone. — Edokter (talk) — 11:25, 25 March 2014 (UTC)

That doesn't answer the question asked. That could be zero or 160,000. Next. Lugnuts ^{Dick Laurent is dead} 11:35, 25 March 2014 (UTC)

Per Wikimedia Analytics - User Agent Breakdown by Browser, February 2014 saw 67,579 M page views by mobile browsers; 27.2% of all traffic. How many viewed reference lists that used fixed columns is unknown. Several research articles anticipate that mobile browsing will overtake fixed browsing in the next few years. Thus, we need to optimize all aspects of Wikipedia for the best reading experience for all devices. --  Gadget850 ^talk 15:09, 25 March 2014 (UTC)

In the absence of systematic bias, we can assume that about a quarter of the views of the 160,000 articles will be made on mobile phones. That's a lot of views that give mobile phone users a sub-optimal experience unnecessarily, in my humble opinion. We also need to be aware that not everybody has a choice over their platform. Wikipedia Zero is driving the use of mobile platforms for viewing Wikipedia in developing nations. I would not be surprised if mobile phones soon became the predominant means of viewing the English Wikipedia in much of Africa. It's not a difficult fix to deprecate |2 and |3, etc. and to encourage fixed width for articles that have more than a handful of references. --RexxS (talk) 17:54, 25 March 2014 (UTC)

"...we can assume...." *Claxon goes off* No original research, please. Lugnuts ^{Dick Laurent is dead} 08:12, 26 March 2014 (UTC)

Given your title for this section, your lead comment and your following comments, you have already made up your mind. Thus, further discussion is useless. Bye. --  Gadget850 ^talk 11:02, 26 March 2014 (UTC)

No, I'm inviting others to add to the discussion. Unlike you and your tag-team friend. Bye? What a child. Lugnuts ^{Dick Laurent is dead} 13:12, 26 March 2014 (UTC)

Break

Comment I have a couple of questions/observations:

1. Since we have a dedicated platform for mobile users, then why can't the ref column feature be simply disabled on that platform? It seems to me this is something a mobile edition of the site was designed to accommodate.
2. How do you reconcile the decision to scrap the feature with MOS:ACCESS#Resolution which seems to back up Lugnuts' "we don't care" stance?

I am largely ambivalent about scrapping the parameter since I try to use the "em" system where possible, but I can imagine a scenario where editors may just want to limit the number of columns to two or three for purely aesthetic reasons. For instance, a dozen references split across two columns looks ok; a dozen refs across three or four columns not so much. Betty Logan (talk) 14:31, 26 March 2014 (UTC)

1. That was exactly what I was about to suggest, were I not interupted by Lugnuts (edit conflict). That is when to ANI. I actually plan to implement that momentarely.
2. That section on screen resolutions was written before columns became mainstream. Using a fixed number of columns is not a matter of "limiting" a maximum number of columns, rather then "forcing" that exact number of columns, no matter what the resulution. This creates problems for anyone who's display is not optimized for that specific number of columns. So it basically makes no sense to use them at all. Either use adaptive columns by using width, or no columns when there are less then 10 references. — Edokter (talk) — 14:48, 26 March 2014 (UTC)

"This creates problems for anyone who's display is not optimized for that specific number of columns." Not really, as there is no problem, as we've already established. Lugnuts ^{Dick Laurent is dead} 15:05, 26 March 2014 (UTC)

You deny that there is a problem on mobile devices then? --Redrose64 (talk) 17:06, 26 March 2014 (UTC)

Pretty much. As I've said, one user reports a problem. Apparently, there are billions of page views on mobiles, but no-one else seems to be logging this issue. Therefore... Lugnuts ^{Dick Laurent is dead} 18:53, 26 March 2014 (UTC)

Comment I really don't see a problem to be solved here, can you actually give an example where a user has reported a problem? I view wikipedia regularly on a mobile device and have never noticed an issue. Wee Curry Monster talk 18:02, 26 March 2014 (UTC)

I've asked several times for other issues to be reported, but none have been forthcoming. Lugnuts ^{Dick Laurent is dead} 21:00, 26 March 2014 (UTC)

Comment-- usage of WP is probably different between desktop users and mobile users. Mobile users less likely to be trawling through references to potentially notice that the layout is not optimized... in my experience, people access Wikipedia on a phone when a laptop is not near, e.g. in a pub. But this is OR Suggest if this is indeed depreciated, to make the change with minimal interference to normal editing, and not to let it break and wait for editors to correct it themselves, as was proposed above. If it is not depreciated, suggest delete the otherwise pointless category. Lesion (talk) 23:26, 26 March 2014 (UTC)

On my phone, when a wiki page loads, each subsection is autocollapsed, including references. Not sure I have ever bothered to open the references subsection on any article when using mobile. Lesion (talk) 23:31, 26 March 2014 (UTC)

Thanks Lesion. Lugnuts ^{Dick Laurent is dead} 08:15, 27 March 2014 (UTC)

Lesion, I would appriciate if you do have a look, so at least you know what we are actually discussing. Watch this page on your mobile and tell us what you see, and more important, if you think the references are readable. — Edokter (talk) — 11:02, 27 March 2014 (UTC)

Still waiting the billions to log this problem. How's your Arbcom report coming along? Chortle. Lugnuts ^{Dick Laurent is dead} 12:11, 27 March 2014 (UTC)

Those of us maintaining the technical side of Wikpedia must take into account any potential problem that occurs now or may occur in the future. This is called anticipation; it is an important process necessary to ensure Wikipedia remains accessable to everyone. One report is enough to recognize an issue, and several editors have confirmed this to be an issue. Ignoring it does not make it go away and is ultimately detrimental to the reader experience.

This may all seem like a big fat joke to you, and Wikipedia may be nothing more then your average joke site, but please stop antogonizing those who actually want to make it a better experience for our readers. If tripping up editors while making light of the situation is your only way of conveying your opinion, or interact with certain editors, then I can no longer feel any respect for you. I woke up this morning thinking I'd start fresh and continue in a civil discussion. But your remark has crushed that sentiment once and for all. This will be my last communication towards you and from now on, will be on my ignore list. Have a good life. — Edokter (talk) — 12:34, 27 March 2014 (UTC)

Don't worry, I'll get over it. Now educate yourself on the polices that you seem to know nothing about. Shame on you. Lugnuts ^{Dick Laurent is dead} 14:26, 27 March 2014 (UTC)

Wow... just wow... just ...... —TheDJ (talk • contribs) 18:50, 31 March 2014 (UTC)

The comments in the sections above just make me wonder what happens if I'm gonna question the words people add to articles... I mean who has been complaining about the article ? You really think hundreds of people are just waiting for the words you are adding ? Who still needs an education in such a world... —TheDJ (talk • contribs) 18:54, 31 March 2014 (UTC)

Edokter, I do "actually know what we are talking about", but thanks for the concern =D Lesion 16:36, 30 March 2014 (UTC)

I don't really want to jump into an argument, but if I see something ugly on my mobile phone (such as an ugly reference list), I'm not going to report it as a problem--I'm just going to ignore the reference list. (I don't even know *how* to report it, so I doubt many 'normal' people know how to report it either.) The number of people reporting a problem really is irrelevant to the problem's level of importance. Metamaterial is currently in the "fixed column count" group with two columns wired in to its reference list. On my Android, when I open the reference list. it shows up as one column on the "Mobile" version. I'm not sure where the "magic" for that is located. Maybe MediaWiki's software or the templates or the mobile interface. I'm just reporting what I see on my current device. (I now think it's the mobile site that has the magic because my Android tablet also puts the reference list as one column when I open it up with he default mobile interface even though it's much wider.) OR or not, that's what I see on my devices. Consider it a point of information. Your mileage may vary. Other devices may do it different ways--what does a WebTV do or a iDevice or...? The list is endless. SesquiZed (talk) 02:05, 5 April 2014 (UTC)

Thanks for the comment. The fixed columns have in fact been disabled on the mobile site. The question is wether we want to disable it for desktop as well. — Edokter (talk) — 02:20, 5 April 2014 (UTC)

The answer is no. Lugnuts ^{Dick Laurent is dead} 09:59, 6 April 2014 (UTC)

Comment – The tracking category fails to take into account usage of the template {{Refbegin}}, which also supports a fixed amount of columns. The combined total of both using fixed amount of columns would probably be freakin' enormous, however. I definitely prefer the look of 30em on larger monitors, though, and would support it being used as the default value, though that may be difficult for reasons stated. (And, for the record, it shows as one column on my phone.) Whisternefet (t · c) 04:17, 6 April 2014 (UTC)

{{Refbegin}} should be tracked as it was updated on 15 March. --  Gadget850 ^talk 13:02, 6 April 2014 (UTC)

Restart 2

I'm gonna try this once more with a fresh proposal, since I think there were some incorrect assumptions in the initial proposal. I make the following proposal: To change the current behavior of a fixed amount of columns and to reinterpret this value as a maximum number of columns with a minimum column width. The number of columns will be reduced with 1 (until column count == 1), if the width of a single column is fewer than a value between 20 and 35 that we can fight over in a follow up proposal. Specifying a manual column width would not set a fixed number of columns (just like now) and specifying both will also work just as before. When your screen is more than the specified columncount * the implied column width, there will also be no difference. This is good for mobile and should not affect other users unless their screen is very small. I have implemented this in the sandbox and it can be tested against the testcases. Please discuss —TheDJ (talk • contribs) 21:38, 6 April 2014 (UTC)

I can see the reasoning here. But I would would slightly simplify it by not have a colwidth parameter. In other word, when |2 is given, it should give a max number of 2 columns with a default width of 30em (instaed of 20em). — Edokter (talk) — 22:53, 6 April 2014 (UTC)