Last modified: 2014-11-10 19:16:08 UTC
It flipped the y-offset for some reason: ltr: text-shadow: #333333 0px -1px 1px, #FFFFFF 0px 2px 1px; rtl: text-shadow: #333333 0px 1px 1px, #FFFFFF 0px 2px 1px; That was weird. Now of course in that case it needn't flip at all and I can just tell it not to, but this still ain't proper behaviour.
(In reply to comment #0) > It flipped the y-offset for some reason: > > ltr: text-shadow: #333333 0px -1px 1px, #FFFFFF 0px 2px 1px; > rtl: text-shadow: #333333 0px 1px 1px, #FFFFFF 0px 2px 1px; When you say "it flipped," what is "it"? Is this a ResourceLoader bug?
See the numbers in the rule? The -1 became a 1. It flipped. Is resourceloader what flips the interface for rtl languages?
(In reply to comment #2) > See the numbers in the rule? The -1 became a 1. It flipped. Yes, the CSS changed (flipped), but what changed it? Where did you see this? What were you doing when this happened? Are there exact steps to reproduce?
I don't know; where does mw flip the interface css for rtl languages? I was juggling ostriches on the moon at the time. I suppose to reproduce it you would have to go to the moon and juggle ostriches, then. Oh, and put something like what I pasted above in a file that gets flipped, probably. Unless it really was the ostriches what did it? Now there's a thought.
URL and screenshot welcome.
(In reply to comment #0) > ltr: text-shadow: #333333 0px -1px 1px, #FFFFFF 0px 2px 1px; > rtl: text-shadow: #333333 0px 1px 1px, #FFFFFF 0px 2px 1px; There are two shadows defined here, and ResourceLoader (or actually CSSJanus) only supports flipping the first one in a rule right now. :( I added the half-baked feature in I97ee7431 to fix bug 45677. (Rephrasing the summary and moving to a better component.)
Ah, thanks. That explains it (and why this seemed so familiar, d'oh).
