Difference between revisions of "Submissions:2015/Why MediaWiki's skins suck and we will never have anything nice"

From WikiConference North America
Jump to navigation Jump to search
(Whee?)
 
(→‎Interested attendees: yay, I'm the first! +1)
Line 43: Line 43:
 
'''If you are interested in attending this session, please sign with your username below. This will help reviewers to decide which sessions are of high interest. Sign with four tildes. (<nowiki>~~~~</nowiki>).'''
 
'''If you are interested in attending this session, please sign with your username below. This will help reviewers to decide which sessions are of high interest. Sign with four tildes. (<nowiki>~~~~</nowiki>).'''
   
  +
# [[User:Protonk|Protonk]] ([[User talk:Protonk|talk]]) 09:03, 3 September 2015 (EDT)
 
# ''Add your username here.''
 
# ''Add your username here.''
   

Revision as of 13:03, 3 September 2015

Title
Why MediaWiki's skins suck and we will never have anything nice
Theme
tech
Type of submission
presentation
Author
Isarra
E-mail address
zhorishna at gmail
Username
Isarra
Affiliation
MediaWiki developer/sysadmin
Abstract
MediaWiki's skinning system is kind of not good, in quite a few different ways. In this I will talk about how and why this is, run through some of the history, and talk about anything I notice and/or can make sense of trying to actually make it better (Jack Phoenix's attempts to get some stuff from Wikia into core, Jon Robson's thing to abstract it all to use markdown instead of php and crap).
Why it's horrible:
  • The skinning system is basically just a pile of hacks that's been extended and cleaned up a bit over the years (may go into examples if I'm not already too depressed)
  • Skins have high technical requirements to create, since they are basically just a specific extension type written in php, LESS, and horror
  • Lack of flexibility (only a few canned options in core, mostly centred around Vector)
  • Changes have a tendency to break existing skins
  • Starting out, people often don't know there's an example skin and wind up usually using Vector as an example and forking that instead, even though Vector is one of the most complicated, over-engineered skins in existence
Why it isn't becoming less horrible:
  • It needs a complete architectural overhaul, but nobody can agree on what that would look like, has the resources to actually do it, or wants to face the consequences of breaking every existing skin ever
  • The requirements for a new system are huge - it needs to do everything the current one does, but better - in order to justify the change
  • The system has a long history of needless breaking changes and skin developers can get a bit cranky
  • Most people who actually use the skinning system to make new stuff are third party developers/sysadmins/whatever who have very little say in mw core or development as a whole
  • Vector's history
Will probably also run everyone through the process to actually make a skin as an example.
Length of presentation
15-30 min? I dunno.
Will you attend WikiConference USA if your submission is not accepted?
If I can

Interested attendees

If you are interested in attending this session, please sign with your username below. This will help reviewers to decide which sessions are of high interest. Sign with four tildes. (~~~~).

  1. Protonk (talk) 09:03, 3 September 2015 (EDT)
  2. Add your username here.