Wiktionary:Votes/pl-2013-03/Romanization and definition line

• Voting on: Requiring that each romanization entry contain at least one definition line starting with "#" in the wikitext. This revision of chūkei does not contain a definition line in the wiki text and thus fails to meet this requirement. By contrast, this revision of bàndǎotǐ and this revision of afdomjanda do contain a definition line starting with "#", and thus do meet the requirement.

An example of wikitext that does not meet the requirement:

==Japanese==
===Romanization===
{{ja-romaji|ちゅうけい}}

Examples of wikitext that meet the proposed requirement:

==Gothic==

===Romanization===
{{got-rom}}

# {{got-romanization of|𐌲𐌰𐌱𐌰𐌹𐍂𐌰𐌽}}

==Mandarin==

===Romanization===
{{cmn-pinyin}}

# {{pinyin reading of|半導體|半导体}} semiconductor

• Vote starts: 00:01, 1 July 2013 (UTC)
• Vote ends: 23:59, 31 July 2013 (UTC)

• Vote created: Dan Polansky (talk) 08:14, 30 March 2013 (UTC)
• Discussion:

  Wiktionary:Beer_parlour/2013/February#Stripping_extra_info_from_Japanese_romaji

  Wiktionary:Grease_pit/2013/March#Simplification_of_romaji_entries

  Wiktionary talk:Votes/pl-2013-03/Romanization and definition line

  Wiktionary:Beer_parlour/2013/March#Romanization and definition line

  Wiktionary:Requests_for_deletion/Others#Template:ja-romaji

• Guidelines:

  Wiktionary:About Japanese#Romaji entries

  Template:ja-romaji/documentation

• Related vote:

  Wiktionary:Votes/pl-2013-03/Japanese Romaji romanization - format and content

1. Support -- Liliana • 19:11, 1 July 2013 (UTC)
2. Support —CodeCat 19:24, 1 July 2013 (UTC)
3. Support Matthias Buchmeier (talk) 20:24, 1 July 2013 (UTC)
4. Support. --Dan Polansky (talk) 20:25, 1 July 2013 (UTC) My rationale: To make the job of spotting missing definitions easier for humans and bots; to achieve uniformity of formatting of entries that direct the reader to other entries for definitions, as is the case with inflected form entries and alternative form entries. --Dan Polansky (talk) 20:25, 1 July 2013 (UTC)

   @Dan, as has been noted before, JA romanization entries have no definition, so that point is wholly moot with regard to {{ja-romaji}}. -- Eiríkr Útlendi │ Tala við mig 20:54, 1 July 2013 (UTC)

   In my rationale, by definition I mean whatever apears on the definition line, which may be a {{non-gloss definition}}, which, strictly speaking, is not a definition. Thus, an entry that has "# Romanization of ..." on the definition line has a definition, and the bot can spot its presence. If this broadened use of the term "definition" troubles you, you may propose another term. Until then, I have clarified what I meant. --Dan Polansky (talk) 21:00, 1 July 2013 (UTC)

5. Support - -sche (discuss) 21:35, 1 July 2013 (UTC)
6. Support.​—msh210℠ (talk) 00:32, 2 July 2013 (UTC)

1. Oppose , quite strongly. This would require 1) longer wikicode, 2) multiple template calls instead of one call, and 3) splitting things like {{ja-romaji}} into multiple templates. All for no appreciable benefit. -- Eiríkr Útlendi │ Tala við mig 19:17, 1 July 2013 (UTC)

   As an addendum, Chinese is categorically different from Japanese and Gothic in this example -- Chinese romanization entries are disambig entries, one-to-possibly-very-many, with each line possibly providing a gloss. Gothic entries are one-to-one, AFAIK. Japanese entries are one-to-one in many (most?) cases. Even in those instances where Japanese entries are one-to-many (up to a maximum of six, if I recall correctly), each link is itself to a disambig entry, and that entry would list the relevant lemmata and glosses, not the romanization entry.

   Bots are quite important to keeping Wiktionary working well and we should not underestimate that. There is quite a substantial benefit in the parseability of the page to supporting this proposal. A bot cannot tell which definitions a page contains if the # is hidden inside a template. Yes, bots can expand templates, but this is somewhat limited (I think they can expand either completely or not at all, not just one "level") and it also slows the bot down substantially. And it is pretty much impossible to do expansion on a wiki dump, mainly because expansion requires the bot to send a request to the wiki, which would slow processing down by like 99% if not more (dump processing is very fast otherwise; it might take only a few minutes to process the whole thing). Furthermore, the template used in the dump might not match the one on the current wiki, so the expansion would give incorrect results. Not to mention that it becomes harder for human editors as well to see if an entry is missing a definition. The templates that people are now using for this have certainly tripped me up plenty of times. I really don't see how longer wikicode or having multiple templates really makes any reasonable difference whatsoever. Longer code or more template calls in the code is not necessarily a bad thing and to assume this is short-sighted. —CodeCat 19:33, 1 July 2013 (UTC)

   • Yes, bots are important. They can be programmed to handle lots of things automatically.

   So why cannot a bot be programmed to recognize a template like {{ja-romaji}} and ignore that accordingly? Or if templates themselves are the trouble, then why cannot a bot be programmed to recognize that XYZ language only contains header ===Romanization===, and ignore that accordingly?

   • About human editors, I completely disagree about this instance -- romanization entries are, by our very definition, purely stubs. For Japanese entries, simply noticing that {{ja-romaji}} is present is enough. Moreover, anyone working with {{ja-romaji}} would be working with Japanese, in which case they would either already know the convention, or would be expected to learn about it in order to continue working effectively with Japanese entries.
   • I admit to some annoyance about this vote. I had understood this to have been entirely superseded by Wiktionary:Votes/pl-2013-03/Japanese Romaji romanization - format and content. Moreover, I think the only language that currently uses the templated format that you and Liliana are opposed to is Japanese -- which neither of you work with. I'm honestly confused -- why can't you two just program your bots to handle these cases? Instead of doing that, you appear to be demanding that human editors completely rework a set of entries in a language that is otherwise entirely outside of your scope. I really don't understand this. -- Eiríkr Útlendi │ Tala við mig 19:55, 1 July 2013 (UTC)
     • In an office of 1,000 people, if 999 people use Windows, you can't seriously tell all of them to learn Linux just so they can work with your computer. This is exactly what you're doing here. -- Liliana • 20:06, 1 July 2013 (UTC)
       • I'd argue the opposite. The only language impacted by this template is Japanese. Neither you nor CodeCat deal with Japanese.

       Moreover, if you make the change in your bot once, then everyone goes on about their business. The wikicode is already settled, so you have your spec effectively in place. If instead we change the entry format and templating, then everyone dealing with JA has to change what we're doing by 1) halting all work on JA romanizations while we figure out what the heck we're doing, again, 2) come up with new templates, 3) relearn the templating conventions for JA romanizations when creating new entries, 4) struggle to bring other less-involved editors up to speed with the conventions for JA romanizations, again, and 5) get bots (presumably) to rework the however-many existing JA romanization entries. It doesn't make any sense. -- Eiríkr Útlendi │ Tala við mig 20:17, 1 July 2013 (UTC)

   Since this template violates WT:ELE (which is implied by the fact that it is tagged as an error by bots that enforce it), sanctioning the use of this template essentially implies that it becomes a part of WT:ELE. Since all significant changes to that page require a vote, this does too. Until then, I can't support its use. —CodeCat 20:20, 1 July 2013 (UTC)

   • Your logic is spurious when you say that bots tagging this as an error means that it's in violation of WT:ELE. Any bot could tag anything as an error -- that depends entirely upon the bot's programming, which may or may not be correct. As has been pointed out before (and in response to you directly, CodeCat), WT:ELE makes no such cut-and-dried statement of requirement. As such, I argue that it is the bots that are incorrect. -- Eiríkr Útlendi │ Tala við mig 20:33, 1 July 2013 (UTC)

   So basically this is an issue of the interpretation of WT:ELE? In that case, support this vote in favour of the interpretation that WT:ELE requires the listed elements to be present in the wikitext of the page itself, and not in a template. (After all, your interpretation would mean we could hide other things behind templates too, like headers. But such a move would break a lot more bots than just Autoformat...) —CodeCat 20:38, 1 July 2013 (UTC)

   FWIW, I would support a move to rework WT:ELE to remove ambiguity, and to make it absolutely plain what elements are required in the wikitext itself. ‑‑ Eiríkr Útlendi │ Tala við mig 22:39, 20 July 2013 (UTC)

2. Oppose, per Eirikr. I don't there are serious technical obstacles here. It seems, the opposition to the use of {{ja-romaji}} and the simplified format is political and personal. --Anatoli ^{(обсудить}/^вклад) 22:39, 1 July 2013 (UTC)
3. Oppose I'd also support "stubbing" of all such useless redirects (Gothic romanizations listed above, as well as inflected forms, come to mind). The less space that kind of data occupies in the database dump, the better. Bots should skip these kind of "entries" because they are not dictionary headwords but a workaround for primitive search facilities. --Ivan Štambuk (talk) 14:03, 20 July 2013 (UTC)

1. Abstain My objection was to having full definitions given for Romanizations, and I wanted to replace them with a link to another page. Where or how that link is added is not a big concern to me. I've never tried to parse this site or build a bot for it, so if someone like Codecat who has a great deal of experience in that area says that not using # is a problem, then I'll take their word for it. --Haplology (talk) 23:45, 1 July 2013 (UTC)
2. Abstain. Both sides are convincing. — Ungoliant ^(Falai) 23:22, 20 July 2013 (UTC)

• Vote passes 6–3–2. Fortunately, {{ja-romaji}} is already compatible with being used only as a headword template (it doesn't cause any ill formatting, nor add any cleanup templates or whatnot, when no parameters are passed), so this can be fixed rather straightforwardly, via the following steps:
  1. Create a definition-line template, e.g. {{ja-romanization of}} or whatnot.
  2. Alter existing uses of {{ja-romaji}} to be compatible with this vote, using # and the new template to generate definitions, leaving {{ja-romaji}} just as a headword-template. (This seems very bottable, and I'm happy to help with it.)
  3. Modify {{ja-romaji}} to be a pure headword template.
  (We'll also want to notify existing users of that template, and to unblock KassadBot (talk • contribs); those steps do not require any particular order.)
  Incidentally, I do not believe any modification to Wiktionary:Entry layout explained is absolutely required, since despite claims above, it already states that definitions "are simply added in one big block, line after line, each beginning with a number sign (#)"; but if others disagree, please comment here and we can figure out how to improve the wording.
  —Ruakh_TALK 18:30, 2 September 2013 (UTC)

• Now done, except that I left {{ja-romaji}} able to create definition lines for now. (It'll just add {{attention|ja}} as well.) After a while, if we find that no one is still using it in the old style, that support can be removed. —Ruakh_TALK 16:57, 8 September 2013 (UTC)

• Thank you for all the work on this. --Dan Polansky (talk) 18:43, 9 September 2013 (UTC)