Jump to content

Topic on Talk:New skins system

Dantman (talkcontribs)

Some of the assertions here sound incorrect.

installing a new skin is complicated and requires shell access, which excludes / unnecessarily makes life harder for some of our re-users; and

Skins don't require shell access like upgrades used to. They only require filesystem access. Which is basically the same thing needed for extensions. This assertion looks wrong.

contributing a new skin is essentially impossible except for those of us in the ivory tower of the Wikimedia Foundation.

Modern was introduced by river. He's a root but he is not part of the WMF. Additionally our old standard default skin for almost a decade MonoBook, was created by Gabriel Wicke who back then was just a volunteer.

So this assertion is wrong too. The problem here is not the WMF. It's that to create a good core-level skin you need two completely different skill sets; The ability to design a really good and at the same time really functional design. And a complete understanding of MediaWiki that allows you to make designs that perfectly fit with MediaWiki's features rather than the features of other CMS systems. And sadly, finding someone with those kind of skill sets is very rare. So it's very rarely that a skin good enough to be considered for core shows up. Since being a core skin means that the skin should both be generic enough to be useful to every wiki and also make absolute perfect use of the different things we output, with no room for hacks.

Jdforrester (WMF) (talkcontribs)

As I'm sure you can see, this is a draft. Feel free to correct it. :-)

Skins don't require shell access like upgrades used to. They only require filesystem access. Which is basically the same thing needed for extensions. This assertion looks wrong.

Sure. Fixed. But the comment still stands; requiring filesystem access is still inappropriate for a basic system administration task like skins. That extensions also need fixing is a known issue.

Modern was introduced by river. He's a root but he is not part of the WMF. Additionally our old standard default skin for almost a decade MonoBook, was created by Gabriel Wicke who back then was just a volunteer.

I am aware of our history; I was there. :-)

However, it is true that contributing a new skin to core has not been achieved in the past 5 years except by the Vector team, who were employed by WMF, and that when someone proposed on wikitech-l the inclusion of a new skin there was no process for them to follow that was acceptable.

Dantman (talkcontribs)

Could you show me any topics where someone actually suggested a skin for core?

Jdforrester (WMF) (talkcontribs)
Dantman (talkcontribs)

Erudite couldn't possibly qualify as a core skin. I was just to polite to actually start listing the issues.

Jdforrester (WMF) (talkcontribs)

It's "polite" to deliberately fail to answer someone when they ask what issues with their skin would need work before you would deign to let them include it? Your dictionary differs from mine.

Dantman (talkcontribs)

It's not fixable.

Erudite is just a random WordPress skin duck taped up to work on MediaWiki. The base css itself comes from a WordPress theme which means a few things:

  • The css itself doesn't even follow our coding guidelines. That's fixable but minor.
  • The css is not based on our standard styles. They're not included at all so none of the standard styles you would expect to work on a standard skin are available.
  • The design is css reset based. Usually a reset based design will completely break if you try to remove the reset. It takes a lot of work to fix it. But using a reset on a wiki skin has the potential to break site and user styles in the content.

Besides the css:

  • The markup is problematic for a standard skin. Standard ids expected are not used so you can't target things with the same css. The navigation uses it's complete own classes and ids and discards other attributes we usually output. And you can see how it's not a MW skin. Content actions/tabs are entry-meta. Post and styles are scattered around. The sitename is a "blog-title". The tagline which is aimed at print output but sometimes stylized is a "blog-description" and always visible in an ugly-by-standard way.
  • The design's typography and even overall design and layout is not generic. Fancy serif fonts are used. The first-letter style nor the fancy <hr> fit a generic theme. The placement of search and personal tools are also bad for a standard theme, but fitting of a theme for some random MW based site that likes the theme enough to use it as their default. And the placement of tabs like watch, history, edit, etc... is not fitting for a community edited wiki. The design carelessly uses <br>'s too.

But fundamentally there are two issues:

  • The theme was not designed from the start as a MediaWiki skin. You cannot put MediaWiki content made to work on one of our standard skins and expect it to work as you would expect any standard skin to work. Even if you managed to fix this things fundamental to the design like the restricted page width would interfere with lots of existing wiki content. This is something detrimental to a standard skin but useful to a standalone skin.
  • The theme by necessity is a navigation hack. It uses a horizontal bar instead of a sidebar menu. And to do this it hacks around with the sidebar navigation in a way that discards navigation that users have added and won't even work at all on some wikis. We do not have a standard way to handle horizonatal navigation. So until I rewrite the skin system and eliminate our dependence on MediaWiki:Sidebar for site navigation we can't accept any theme using horizontal navigation as a standard theme.
Reply to "Incorrect assertions"