Wikipedia

Wikipedia:Templates for discussion/Log/2013 March 29

< Wikipedia:Templates for discussion | Log
< March 28 March 30 >

March 29

edit

Template:Cite quick

edit
The following discussion is an archived debate of the proposed deletion of the template below. Please do not modify it. Subsequent comments should be made on the appropriate discussion page (such as the template's talk page or in a deletion review). No further edits should be made to this section.

The result of the discussion was mark as historical Plastikspork ―Œ(talk) 22:58, 6 April 2013 (UTC)[reply]

Template:Cite quick (talk · history · transclusions · logs · subpages)

This template was created as a 'fast-citation template' to improve on the performance of existing templates by removing some little used fields. It is now no longer necessary, as the older templates have been converted to Lua and no longer have performance problems. This could be similarly improved but as it used in only a few dozen articles it makes more sense to replace it with the more common templates and delete it, to eliminate the now unnecessary duplication. JohnBlackburnewordsdeeds 18:46, 29 March 2013 (UTC)[reply]

  • note a previous deletion discussion for this template is here.
  • note I used twinkle and it was unable to add the TfD notice to the template as it's protected (not sure why with only a few dozen uses). I had the 'noinclude' option set so it would not show up on pages anyway, and few people will visit the template directly, but it should probably be added so it's properly categorised.--JohnBlackburnewordsdeeds 18:52, 29 March 2013 (UTC)[reply]
  • Strong keep (by author): The simple Template:Cite_quick has been shown to run as fast as the highly complex Lua versions (for the same parameters), and hence no need to rewrite with Lua script, but also, {Cite_quick} allows more instances per page, before hitting template limits, and thereby still outperforms Template:Cite_web or Template:Cite_book, even after they were transitioned to use Lua. Also, this TfD is severely premature, as {cite_web} was just transitioned, to use Lua, only 18 hours prior to this TfD. Hence, the "paint is not even dry" on the Lua-based cite templates, and already, User:JohnBlackburne is trying again to debate deletion, as done just 5 months earlier when the eventual consensus was a strong keep due to technical performance advantages. Anyway, as a small, traditional template, {cite_quick} can be ported to the other-language wikipedias to rapidly display any citations within translated articles copied from English Wikipedia, with the cite templates inside. This TfD might have been more appropriate several months later, such as 6 months after full transition of the wp:CS1 cite templates to Lua, rather than just 18 hours after initial deployment of {cite_web} with Lua. These continual TfD's are exhausting the community's patience. -Wikid77 (talk) 19:43, 29 March 2013 (UTC)[reply]
Old Citations {{cite quick}} Lua Citations
Execution Time (seconds) 18.2 4.1 3.2
Preprocessor visited node count 155512 34144 7638
Preprocessor generated node count 52232 36891 32518
Post-expand include size (bytes) 947553 424847 565834
Template argument size (bytes) 515827 177747 0
Highest expansion depth 14 7 2
Expensive parser function count 0 0 0
  • Comment. Using a test bed of 310 citation templates taken from Barack Obama that I've been using as a benchmark, I find that Lua-based citations are about 20% faster than {{cite quick}} while making no compromises about the supported function set. Hence, for most uses, I can see no reason to prefer cite quick. The only tracked parameter where cite quick outperforms Lua is in the post-expand include size, where the cite quick template benefits from the lack of COinS metadata and the omission of various parameters, which allows cite quick output to be 25% smaller than the Lua citations. It is possible that there may be articles on the cusp of violating the include size limit (2 MB) where that 25% size difference is enough to matter, though I don't know of any examples. Personally, I think we probably should begin removing cite quick from articles in favor of the Lua citations, but it might be premature to say that it could never be useful, so I'm not really in favor of deleting it right away. Restore the old templates in article space first, and make sure the transition goes smoothly, and then we can decide whether we want to preserve cite quick or not. Dragons flight (talk) 21:00, 29 March 2013 (UTC)[reply]
"Bibliography of South America" Old Citations  Cite_quick  Lua Citations
Execution Time (sec.) [480 cites] ~24.3 7.2 7.6 (+6%)
Preprocessor visited node count 259930 61831 32307
Preprocessor generated node count    158749 149821 149534
Post-expand include size (bytes) 1374493 452373 909871
Template argument size (bytes) 756729 168154 0
Highest expansion depth 12 8 4
Lua time usage -- -- ~0.540s
  • Cite_quick faster for mix of citations: Timing tests with several other articles reveal that {cite_quick} can be slightly faster (6%), than the Lua-based wp:CS1 cite templates, such as when using numerous blank parameters, as in "Bibliography of South America" compared during 20 trials. When running the benchmarking comparisons, it is important to test a wide variety of cites, in actual article pages (but without infoboxes or other extra templates). To magnify the speed differences, the cites were repeated 3 times, as a total 480 cites in the page. The faster run times of {cite_quick}, over {cite_web}, might be due to faster handling of many blank parameters, or merely an artefact of background server load, because the speed is so close, the repeated timing tests could slightly favor one, or the other. The times shown in the table, are the average of the 3 lowest times (but both trended near ~8 seconds), while {cite_web/old} reached 30 sec. Various tests with using parameter "archiveurl=" have shown a slight slowing of {cite_quick}, so the speed depends on which parameters are used. Therefore, I characterize the comparison as being, essentially, {cite_quick} can "run as fast" as the Lua-based templates (although in some cases 6% faster), in actual article pages. -Wikid77 (talk) 14:53, 30 March 2013 (UTC)[reply]
  • comment I agree, there's no need to delete it, just move it or add a notice to it so it's not used; it may be of interest to someone, if only historical when looking at the development of the templates. Now the Lua versions are available there's no need to use it, it's just a slower and less useful variant. It's a little shorter but as noted above (something I had forgotten from the previous discussion) it only gets this through omitting metadata, and on the few if any pages where this is still an issue it's best to fix the pages by editing them as they must be too long by any reasonable metric. So there's no reason to use it. Replacing it will make pages faster, useful on the pages it's on which include some long and slow pages. As for delaying for six months, why? If the Lua based templates are rolled out to millions of pages they must be finished and working 100%. They won't be any more so in six months.--JohnBlackburnewordsdeeds 23:41, 29 March 2013 (UTC)[reply]
  • Lua cites fix clerical errors and run 6x faster not 100%: The major reason the Lua-based templates have been deployed, now, is to improve the overall performance in 1.8 million pages, as running 6x faster and fixing many thousands of clerical errors, such as double-dot "inc.." or "Doe, J.." or "pp.7" (as singular "p."), plus inserting separator dots in 4,900 pages, among nearly 2 million pages using {cite_*}. The Lua cites are not working exactly "100%" because some minor parameters have a different format, such as for editor names, and people are still discussing alternatives, while also considering new features. The current deployment is a tradeoff of fixing prior bugs yet shifting some parameter locations, this week, in favor of fixing many thousands of typos (over 100,000) and reformatting major articles as 6x-faster cites, but still expecting some slight problems in rare parameters. It is a question of the overall priorities, for 1.8 million pages. See: Module_talk:Citation/CS1 for numerous discussions about the Lua cites. Meanwhile, {Cite_quick} has been completely stable, with no change in format during the prior months. -Wikid77 (talk) 14:53, 30 March 2013 (UTC)[reply]
The above discussion is preserved as an archive of the debate. Please do not modify it. Subsequent comments should be made on the appropriate discussion page (such as the template's talk page or in a deletion review). No further edits should be made to this section.

Non-standard film infoboxes

edit
The following discussion is an archived debate of the proposed deletion of the template below. Please do not modify it. Subsequent comments should be made on the appropriate discussion page (such as the template's talk page or in a deletion review). No further edits should be made to this section.

The result of the discussion was merge Plastikspork ―Œ(talk) 23:01, 6 April 2013 (UTC)[reply]

Template:Infobox Chinese film (talk · history · transclusions · logs · subpages)
Template:Infobox Japanese film (talk · history · transclusions · logs · subpages)
Template:Infobox Korean film (talk · history · transclusions · logs · subpages)
Template:Infobox film (talk · history · transclusions · logs · subpages)

Propose merging Template:Infobox Chinese film, Template:Infobox Japanese film and Template:Infobox Korean film with Template:Infobox film.
Following on from discussions at WT:FILM and on the standard infobox page. These variants are not needed and should be merged into the one standard infobox. Any language parameters currently used should be in the body of the article and the data could be moved to the talkpage for future use. Lugnuts Dick Laurent is dead 17:35, 29 March 2013 (UTC)[reply]

Clarification, are you proposing a 'merge' which means the unique fields from each special template are brought into the main template, or are you talking about 'delete' as in all templates except the main one are simply removed? ₪RicknAsia₪ 18:18, 29 March 2013 (UTC)[reply]
Being useful is not a valid reason for keeping the template. This is about standardisation - there is simply no need to have variant templates that have 3 or 4 additional parameters that are used on a small minority of film articles. Lugnuts Dick Laurent is dead 18:16, 29 March 2013 (UTC)[reply]
When I said "useful" I was thinking: These templates remove clutter from the article, they give multiple fields for information which are not provided in the general template and they are well used. They serve a purpose and should not be tossed away. Please clarify above (where I originally asked) what a merge would be as "merge" in the previous debate was used for calls of deletion. ₪RicknAsia₪ 18:35, 29 March 2013 (UTC)[reply]
WP:ITSUSEFUL is an essay, and doesn't necessarily represent consensus. It certainly isn't a guideline or policy. ···日本穣? · 投稿 · Talk to Nihonjoe · Join WP Japan! 16:12, 3 April 2013 (UTC)[reply]
The above discussion is preserved as an archive of the debate. Please do not modify it. Subsequent comments should be made on the appropriate discussion page (such as the template's talk page or in a deletion review). No further edits should be made to this section.
Retrieved from "https://en.wikipedia.org/w/index.php?title=Wikipedia:Templates_for_discussion/Log/2013_March_29&oldid=1138535001"