2010/11/4 Platonides <Platonides(a)gmail.com>om>:
I think many of their changes would break MediaWiki
caching system.
We could reuse their image stripping module [1] on generating the
thumbnails, although the mention that it may produce a XSS make me wary
of such method.
They also say they do data URL inlining based on user agent. We use
aggressive caching so we can't do anything based on User-Agent. We
still do data URLs, though, with a fallback rule that happens to only
work in browsers that don't support data URLs. It's a dirty hack, but
it works.
Roan Kattouw (Catrope)