Template talk:GAN link

Improvements

This template is mainly used for WPVG project banners via the inclusion of Template:WPVG announcements.

The use of the {{Exists}} template caused the talk pages of the nominated articles to be transcluded, invisibly.

This pollutes Special:MostTranscludedPages currently, for example:

542. Talk:Music of Final Fantasy XIV/GA1‏‎ (used on 57,492 pages)
543. Talk:Devil Survivor 2: The Animation/GA1‏‎ (used on 57,490 pages)
544. Talk:Makoto Yuki (Shin Megami Tensei: Persona)/GA1‏‎ (used on 57,490 pages)
545. Talk:Drakengard (series)/GA1‏‎ (used on 57,473 pages)
546. Talk:F-Zero Climax/GA1‏‎ (used on 57,396 pages)

for example. The template also called "{{Replace}}" about eight times, when twice is sufficient.

I have resolved these issues. Clearly if a large list (>500) of GANs are needed, a version using {{Exists}} can be created specially for that page.

Parser profiling data for {{WPVG announcements}}

Old version of GAN link

CPU time usage 1.764 seconds
Real time usage 1.837 seconds
Preprocessor visited node count 4377/1000000
Preprocessor generated node count 6950/1500000
Post-expand include size 66721/2048000 bytes
Template argument size 16962/2048000 bytes
Highest expansion depth 16/40
Expensive parser function count 15/500
Lua time usage 0.308/10.000 seconds
Lua memory usage 1.29 MB/50 MB

Cite error: The opening <ref> tag is malformed or has a bad name (see the help page).===New version of GAN link=== CPU time usage 0.896 seconds
Real time usage 0.958 seconds
Preprocessor visited node count 3427/1000000
Preprocessor generated node count 4758/1500000
Post-expand include size 41593/2048000 bytes
Template argument size 15632/2048000 bytes
Highest expansion depth 13/40
Expensive parser function count 26/500
Lua time usage 0.139/10.000 seconds
Lua memory usage 1.28 MB/50 MB

All the best: Rich Farmbrough, 18:15, 5 July 2014 (UTC).

@Rich Farmbrough, I'm not sure what exactly you did here yet, but as you can see from {{WPVG announcements}} screenshot, it isn't at all cleared up. I'm reverting your edit for now—let's work this out in the sandboxes first before pushing it live, please. czar ♔ 18:19, 5 July 2014 (UTC)[reply]

Thanks for that. For some reason it is failing to parse 1 the way I would expect. I think I know the problem. All the best: Rich Farmbrough, 18:43, 5 July 2014 (UTC).

Take a peek at the test cases. I added {{WPVG announcements}}. All the best: Rich Farmbrough, 20:05, 5 July 2014 (UTC).

@Rich Farmbrough, the WPVG announcements testcases were not using GAN link/sandbox, but I made a section that does and there are no obvious errors. Thanks for helping spruce this up. Do you still have the same memory benefits? What was the benefit of splitting out the guts into /core? czar ♔ 20:38, 5 July 2014 (UTC)[reply]

Yes they (the {{GAN link/testcases}}) were

:



|A-Class_reviews=

|good_article_nominees= 
* {{GAN link/sandbox|Makoto Yuki (''Shin Megami Tensei: Persona'')}}
* {{GAN link/sandbox|Music of ''Final Fantasy XIV''}}
* {{GAN link/sandbox|Devil Survivor 2: The Animation|i=y}}
* {{GAN link/sandbox|''F-Zero Climax''}}
* {{GAN link/sandbox|''Silent Hill: Origins''}}
* {{GAN link/sandbox|''QuackShot''}}
* {{GAN link/sandbox|''Infamous Second Son''}}
* {{GAN link/sandbox|''Sonic X''}}
* {{GAN link/sandbox|''Skeet Shoot''}}
* {{GAN link/sandbox|''Space Cavern''}}
* {{GAN link/sandbox|''Dragon Ball Z: Attack of the Saiyans''}}

Splitting the guts into the core means we don't have to call the "Replace" template and module twice every time we want to see the unornamented name.

The figures are pretty good:

Parser profiling data:
CPU time usage	0.700 seconds
Real time usage	0.756 seconds
Preprocessor visited node count	3318/1000000
Preprocessor generated node count	3946/1500000
Post-expand include size	36329/2048000 bytes
Template argument size	13306/2048000 bytes
Highest expansion depth	13/40
Expensive parser function count	25/500
Lua time usage	0.120/10.000 seconds
Lua memory usage	1.24 MB/50 MB

It's really the time that I think is the benefit here. Up to a second on rendering each of 57k pages, nearly 16 hours across the wiki.

All the best: Rich Farmbrough, 23:17, 5 July 2014 (UTC).

It's not really about speed though. It's about post-include template size and calls to expensive parser functions. Include size went down, and that is great. Expensive parser function count went up, and that's bad. I'll take another look at this when I get back to a computer. :) — {{U|Technical 13}} ^{(e • t • c)} 10:47, 6 July 2014 (UTC)[reply]

It's not really bad. Because although #ifexists is classed as an "expensive" function it is less expensive than transclusion, which clearly has to answer the "if exists" question as well as doo a lot of other stuff. {{Exists}} exists only to get around the limitation in suitable circumstances. All the best: Rich Farmbrough, 23:41, 11 February 2016 (UTC).[reply]

Weird transclusion issue

@Technical 13 and Rich Farmbrough:, could you take a look at Wikipedia_talk:WikiProject_Video_games#Strange? There was an incident where a use of this template was transcluding incomplete refs from a GA talk page. czar ♔ 14:28, 26 September 2014 (UTC)[reply]

Looks like it was a ref issue and it has been fixed already? Is this correct? — {{U|Technical 13}} ^{(e • t • c)} 19:59, 26 September 2014 (UTC)[reply]

This is prone to happen again and it is a known problem: {{Exists}} transcludes the linked page. 84.127.80.114 (talk) 03:26, 27 September 2014 (UTC)[reply]