Code that modifies these properties dynamically need to do a little song and dance to get the properties to take effect: http://help.dottoro.com/ljkpxqht.php
The simplest solution here is to implement nsSubDocumentFrame::AttributeChanged (since nsSubDocumentFrame::GetMarginAttributes is where the attributes are handled). Probably both of them should also check that the element in question is an HTML iframe element, though (or perhaps also xul iframe element, if needed).
Bobby, you want to work on this? If not, I'll steal.
(In reply to comment #2) > Bobby, you want to work on this? If not, I'll steal. I was hoping to get around to it when midterm craziness dies down, but we'll see. So I'll take it for now, and also take the responsibility for pinging you back in a week or two if I decide not to. If there are release timetable reasons why this should happen sooner let me know. ;-)
There is no great rush here. It's not like it ever worked....
I've got the new marginWidth and marginHeight onto nsFrameLoader::mDocShell, but I'm having trouble getting it to take effect. What should I be doing to trigger the restyle here? nsIPresShell::StyleChangeReflow() doesn't seem to be cutting it.
You probably want to call RebuildAllStyleData() on the prescontext (and not do any manual reflowing; the style data rebuild should do the trick).
I don't think that'll work, since I think the marginwidth and marginheight attributes never actually end up in style data. (Maybe they should, though, but we'd need to figure out as what...)
> since I think the marginwidth and marginheight attributes never actually end up > in style data They do. See BodyRule::MapRuleInfoInto where it grabs those from the docshell.
Should we consider making BodyRule follow the nsIStyleRule contract correctly?
Sure. I don't think we should make Bobby do that, though....
Created attachment 530533 [details] [diff] [review] Bug 653930 - Tests. v1
Created attachment 530534 [details] [diff] [review] Bug 653930 - v1 Adding tests and fix. Flagging bz for review.
Comment on attachment 530534 [details] [diff] [review] Bug 653930 - v1 Just nix the comment in AttributeChanged and don't leave a blank line before the else, please. r=me with that. We should clean up the FrameLoader() thing; I'm not sure it can actually be null here. But good enough for now.
Comment on attachment 530533 [details] [diff] [review] Bug 653930 - Tests. v1 r=me
Created attachment 530667 [details] [diff] [review] Bug 653930 - v2 Addressed bz's comments. Carrying over review.
Pushed to mc: http://hg.mozilla.org/mozilla-central/rev/cf9650c9429a http://hg.mozilla.org/mozilla-central/rev/a7e15c159eec Resolving fixed.