[webkit-reviews] review granted: [Bug 50758] Defer loading print stylesheets : [Attachment 76406] updated based on ap's comments

[bugzilla-daemon at webkit.org]

Alexey Proskuryakov <ap at webkit.org> has granted Antti Koivisto
<koivisto at iki.fi>'s request for review:
Bug 50758: Defer loading print stylesheets
https://bugs.webkit.org/show_bug.cgi?id=50758

Attachment 76406: updated based on ap's comments
https://bugs.webkit.org/attachment.cgi?id=76406&action=review

------- Additional Comments from Alexey Proskuryakov <ap at webkit.org>
View in context: https://bugs.webkit.org/attachment.cgi?id=76406&action=review

>> Please modify other project files.
> Done.

Please also update the ChangeLog.

> > Please remove this FIXME. There is no such thing as "unnecessary
stylesheet".
> Stylesheets that are never used are unnecessary to load. This is major
performance concern. I have bug open to fix this too.

How do we know that they are never used? The only thing we could do is block on
CSSOM access and load the stylesheet on demand, but that's even worse, as that
would use a synchronous load. And even that could have compatibility
implications - servers heavily rely on getting exactly one request for a Web
font, for example.

Please do remove the FIXME - it's not really necessary if there's a bug, and we
can continue the discussion in the bug.

> > Please consider also implementing this at least for WebKit2.
> Later.

OK. Generally, we should treat WebKit2 as first class citizen now, and if I
were making this patch, implementing a WebKit2 DRT hook while I was still fully
in the context of this work would be easiest for me.

> LayoutTests/platform/win/Skipped:1098
> +# For https://bugs.webkit.org/show_bug.cgi?id=50758
> +# These require DRT setSerializeHTTPLoads implementation to be reliable.

As a courtesy, we try to modify other platforms' Skipped lists, too (when it's
straightforward).