Wiktionary:Votes/2010-06/Setting lang attribute for transliterations

Note (informative, not part of what's being voted on): The HTML used to encode Wiktionary pages can include a lang specification for the entire page, and a separate one for any part of it. We include such specifications for some things, and they can be added without a vote to others, but this probably requires a vote, as it goes against the current practice of how to edit pages. The template {{xlit}} supplies HTML code including a lang specification.

Currently, transliterations into English of non-Latin-script quotations and example sentences are rendered italicized using ''...'' (common practice: ELE is silent). This vote is trying to change that wikicode for reasons specified in the Grease pit discussion that led to it. (This will change the wikicode and resultant HTML code, but not the display as italicized, except for users who choose to change the display through fiddling with their personal CSS or who have very old browsers that don't use CSS at all.)

Voting on: Should transliterations of quotations and example sentences in entries and citations-namespace pages use (not ''...'' but rather) {{xlit}}?

Example (informative, not part of what's being voted on):

# ]
#: {{Hebr|'''חם''' לי.|lang=he}}
#:: {{xlit|lang=he|'''cham''' li.}}
#:: I am '''hot'''.

• Vote starts: 00:00, 4 July 2010 (UTC)
• Vote ends: 24:00, 17 July 2010 (UTC)

• Vote created: ​—msh210℠ (talk) 04:24, 29 June 2010 (UTC)
• Discussion:

  WT:GP#transliterations_of_usexes_and_quotations

1. Support as vote creator.​—msh210℠ (talk) 03:46, 4 July 2010 (UTC)​—msh210℠ (talk) 04:04, 11 July 2010 (UTC)

   Support: Yes, that way such transliterations become more customizable.   AugPi 03:53, 4 July 2010 (UTC)

   Support Anything that makes our data more highly structured without adding more demands on our editors is a good thing in my opinion. -Atelaes λάλει ἐμοί 04:12, 4 July 2010 (UTC)

   Disclosure (though you probably realize this and were merely being hyperbolic): It does demand something of editors, specifically writing "{{xlit|lang=foo|...}}".​—msh210℠ (talk) 13:44, 6 July 2010 (UTC)

   That's a good clarification, but I guess I was just thinking that everyone should be using quote templates of some sort, which would include this code within them. I know I always do. -Atelaes λάλει ἐμοί 13:56, 6 July 2010 (UTC)

   This is veering off-topic, but: I don't. I hate those templates, and I know I'm not alone, though I've grudgingly accepted that many editors do choose to use them. And per AGF, I assume that they're not doing it just to spite me. ;-)   —Ruakh_TALK 18:34, 7 July 2010 (UTC)

   Actually, I use quote templates solely to spite you. And just so I can feel my deep and abiding spite for you when I use them, may I ask why you dislike them? It seems like more consistent formatting and less coding to me. -Atelaes λάλει ἐμοί 11:59, 8 July 2010 (UTC)

   (1) I find it easier to remember the correct formatting than to remember the names and supported parameters of each of the quote templates. (2) The "indent" thing is ugly as heck. (3) I've grown curmudgeonly in my middle twenties. (4) Everyone seems to put the parameters in a different random order, and some editors put different parameters on different lines, so maybe the resulting format is consistent, but the wikitext is frequently confusing. —Ruakh_TALK 12:49, 8 July 2010 (UTC)

   Ah, those are reasonable reasons. I can now feel the full satisfaction of the irritation I will cause you by using said templates. Thanks.  :-) -Atelaes λάλει ἐμοί 12:54, 8 July 2010 (UTC)

   Support. I'm with Atelaes on this one. Thryduulf (talk) 13:41, 6 July 2010 (UTC) per Prince Kassad below. Thryduulf (talk) 18:55, 9 July 2010 (UTC)

   Supportuntil the dispaly issue is sorted out. --Bequw → τ 17:25, 9 July 2010 (UTC) This wouldn't prohibit the creation of templates that replace {{xlit}}, right? For example it might be nice to have an {{ex-sentence|example=...|lang=...|tr=...|translation=...}} that does the formatting of all lines of an example. --Bequw → τ 14:01, 6 July 2010 (UTC)

   It would not (as I read it) prohibit such. In fact, it's expected, if this passes, that the quote- templates will have transliteration support added to match the format here proposed.​—msh210℠ (talk) 14:11, 6 July 2010 (UTC)

   Support —Ruakh_TALK 18:34, 7 July 2010 (UTC)

2. Oppose it causes certain problems with CJK, therefore it should probably not be done. -- Prince Kassad 22:01, 7 July 2010 (UTC)

   May I ask what sorts of problems? -Atelaes λάλει ἐμοί 22:58, 7 July 2010 (UTC)

   Some browsers will switch to a CJK font if they see that the lang is zh/ja/ko, even if the text contains only Latin letters. This in turn leads to a rather ugly visual appearance. -- Prince Kassad 08:34, 8 July 2010 (UTC)

   Could we get around that by explicitly setting the script to Latin? Thryduulf (talk) 10:43, 8 July 2010 (UTC)

   A couple of things. First, it should be noted that we're not setting the language to "zh", we're setting it to "zh-Latn", which I suspect will get treated differently. Thus, we're setting the language and the script explicitly. Quite frankly, I wonder if most browsers won't know what to do with it at all, and thus simply go with the default. If you can find a browser which does go crazy if given "zh-Latn", by all means present it. I suspect that many would be willing to reconsider their votes (I know I would). However, I certainly won't for the mere possibility that some browser out there might do something bad. -Atelaes λάλει ἐμοί 11:57, 8 July 2010 (UTC)

   If you can find a browser which does go crazy if given "zh-Latn", by all means present it. - Firefox. -- Prince Kassad 15:26, 8 July 2010 (UTC) addendum: here's a testcase so you can see what the problem is. -- Prince Kassad 15:44, 8 July 2010 (UTC)

   I saved the text there as an .html file and tried it in Firefox (Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.9.0.19) Gecko/2010040116 Ubuntu/9.04 (jaunty) Firefox/3.0.190), and the display was five identical-looking lines of normal-font "test".​—msh210℠ (talk) 16:20, 8 July 2010 (UTC)

   You may not have any Japanese fonts, in which case you won't see a change. Or they may not be set properly in the preferences dialog (but normally, they're supposed to). For reference, the version of Firefox I used was Mozilla/5.0 (Windows; U; Windows NT 5.1; de; rv:1.9.1.9) Gecko/20100315 Firefox/3.5.9 (.NET CLR 3.5.30729) -- Prince Kassad 19:27, 8 July 2010 (UTC)

   Is this unique to zh, ja, and ko? (And, if so, why?)​—msh210℠ (talk) 19:32, 8 July 2010 (UTC)

   Hmm, good question. I tried it with Firefox, and surprisingly, the answer is no. You can also use codes such as am or km and it will exhibit the exact same behavior. -- Prince Kassad 20:25, 8 July 2010 (UTC)

   I've added a little bit more text to that test and it produced the following with Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.2.3) Gecko/20100408 Ubuntu/9.04 (jaunty) Firefox/3.6.3. . It does display in different fonts, which isn't what I expected. Thryduulf (talk) 20:50, 8 July 2010 (UTC)

   Yeah, I'm getting different fonts myself. It looks terrible. -Atelaes λάλει ἐμοί 23:18, 8 July 2010 (UTC)

   Bother. I'll post a notice on the usertalkpages of those who voted in support here informing them that they should revisit it in light of this information.​—msh210℠ (talk) 15:31, 9 July 2010 (UTC)

   (Especially Prince Kassad and Thryduulf:) What if you add class="Latn" to it?​—msh210℠ (talk) 15:38, 9 July 2010 (UTC)

   Alas adding class="Latn" to the <p> (which is what I think you mean) doesn't make any difference to what I see :( Thryduulf (talk) 18:55, 9 July 2010 (UTC)

   Well, that's sort of what I meant. What I really meant was adding class="Latn" to a similarly marked-up <p> on an enwikt page (as that's where class="Latn" will mean something). Would you mind trying that? I suspect that may work, and if it does, then we have no problem: we can add class="Latn" to {{xlit}}.​—msh210℠ (talk) 19:06, 9 July 2010 (UTC)

   It works in Konqueror but not Firefox. see User:Thryduulf/langscript for the code and screenshots (feel free to use that page for your own tests too). Thryduulf (talk) 20:16, 9 July 2010 (UTC)

   No wonder it won't do anything - MediaWiki:Common.css does not have any declarations for .Latn, so adding the class has no effect. -- Prince Kassad 20:42, 9 July 2010 (UTC)

3. Oppose I don't see how this will help 'customization' in a meaningful way. A template that seems to do nothing at all. — opiaterein — 16:00, 9 July 2010 (UTC)
   • Customization is a side benefit, not the point, and occurs through the fact that transliterations will be italicized through inclusion in a CSS class (which you can choose to format differently than as italicized, or not display at all, etc.) rather than by use of <i>. The reason for the template's introduction is discussed in the GP discussion that led to this vote, and is to increase legibility by screen readers and search engines.​—msh210℠ (talk) 18:32, 9 July 2010 (UTC)
4. Oppose per Prince Kassad.   AugPi 16:05, 9 July 2010 (UTC)
5. Oppose until such time as the display issues are sorted out. Thryduulf (talk) 18:55, 9 July 2010 (UTC)
6. Oppose, seems like we're not ready for this yet. Mglovesfun (talk) 10:14, 10 July 2010 (UTC)
8. Oppose While I would really like to add this structure, I must concede to Prince Kassad that this makes transliterations look inconsistent and generally terrible. If there is some sort of reasonable workaround, perhaps setting "lang" to the language and "script" (can html tags take a script parameter?) or whatever else, I'd be very open to exploring it. However, that's not what this vote is proposing, in any case. -Atelaes λάλει ἐμοί 12:07, 10 July 2010 (UTC)

This may be a bit presumptuous, but I'm going to go ahead and call this a fail, as all of the supporters have removed their support. It should be noted that this vote proposed a fairly specific technical implementation, which should be considered to have community consensus against. However, that does not preclude similar approaches from being devised in the future. This may, again, be presumptuous, but I strongly suspect that many (especially those who initially voted support, but later withdrew their votes) would support a similar implementation which did not cause common browsers to render transliterations in incorrect fonts. Also, may I just say that, in my opinion, there was no need for this vote. I think that, in general, technical implementations of this nature should be implemented in a few places as testing grounds, and should be pushed more widely if general consensus of interested parties can be made, say on the GP or BP. -Atelaes λάλει ἐμοί 04:15, 11 July 2010 (UTC)