Difference between revisions of "User talk:Chat"

Revision as of 21:57, 3 June 2011

Mmm instead of changing all the templates again from {{Weapon data}} to {{Weapon data var}} couldn't we just detect if the new params are there in {{Weapon data}} and use them?

I mean I changed them all because I had to add the onlyinclude bits but it was much of a pain.

--Frazyl 21:15, 3 June 2011 (UTC)

We could, but I don't want to bloat the weapon data templates more than I have to - fixing {{ratebar}} bought us some time, but we're still going to run into expansion limits sooner or later if we don't avoid growing large template trees.

In any case, the conversion from {{Weapon data}} to {{Weapon data var}} should happen at a much slower rate (see Talk:Main Page#Parametric judge), so it'll hopefully end up less of a painful process.

Note that if you need to change huge numbers of pages en-masse, bots are a good idea. --Chat 21:28, 3 June 2011 (UTC)

Oh yeah bots would have been a good idea I guess. I did lots of cleanups though.

Not sure it would bloat the {{Weapon data}}... Surely it's not parsing it every time but just going in the switch? Any objections if I check how changing it affects loading the Swords page for example? --Frazyl 21:43, 3 June 2011 (UTC)

Forgot to say the issue is every change to {{Weapon data}} has to be reflected to {{Weapon data var}}. --Frazyl 21:44, 3 June 2011 (UTC)

Be my guest - the main thing to check is the 'post-expand include size' - see here for how to get at it. That's the thing we were hitting the limit on before.

--Chat 22:21, 3 June 2011 (UTC)

Before change to make both judge work:

 
NewPP limit report
Preprocessor node count: 35758/1000000
Post-expand include size: 294000/2097152 bytes
Template argument size: 32946/2097152 bytes
Expensive parser function count: 0/100

After change to make both judge work:

 
NewPP limit report
Preprocessor node count: 36190/1000000
Post-expand include size: 294000/2097152 bytes
Template argument size: 32946/2097152 bytes
Expensive parser function count: 0/100

So on the swords page it only changed Preprocessor of 0.0432 % of the limit. I think that's acceptable. --Frazyl 23:09, 3 June 2011 (UTC) With rbkey

NewPP limit report
Preprocessor node count: 38404/1000000
Post-expand include size: 520932/2097152 bytes
Template argument size: 33624/2097152 bytes
Expensive parser function count: 0/100

Adding rbkey seems to roughly increase by 10% of limit the post-expand and add a bit to the others. --Frazyl 23:26, 3 June 2011 (UTC)

Bugger. I don't think we can really do without rbkey, though - without it the tables just weren't sorting correctly, and sorting's the main thing people will want to do with the weapon pages. Maybe there's a way to reduce its size? --Chat 23:30, 3 June 2011 (UTC)

Well do we need padding? Maybe use a switch to convert 0 to 9 with 00 to 09. Also checking if {{{1}}} before doing anything is present is going to help until the table is all filled which would be a false positive. --Frazyl 23:36, 3 June 2011 (UTC)

Err well for Post-expand it's the size so pad to 2 I guess. --Frazyl 23:38, 3 June 2011 (UTC)

As a much more substantial proposal we could put class names for the td and span and use css to assign the style bits automagically instead of outputting it every line. --Frazyl 23:49, 3 June 2011 (UTC)

Done.

NewPP limit report
Preprocessor node count: 38404/1000000
Post-expand include size: 361986/2097152 bytes
Template argument size: 33624/2097152 bytes
Expensive parser function count: 0/100

So for Post-expand that's 158946 less or 7.6% of max less. --Frazyl 01:57, 4 June 2011 (UTC)