[webkit-reviews] review granted: [Bug 27474] Crash if the selection is set in a textarea/text input immediately after setting display:none : [Attachment 33303] Fixes crashes

[bugzilla-daemon at webkit.org]

Darin Adler <darin at apple.com> has granted Ojan Vafai <ojan at chromium.org>'s
request for review:
Bug 27474: Crash if the selection is set in a textarea/text input immediately
after setting display:none
https://bugs.webkit.org/show_bug.cgi?id=27474

Attachment 33303: Fixes crashes
https://bugs.webkit.org/attachment.cgi?id=33303&action=review

------- Additional Comments from Darin Adler <darin at apple.com>
> +	   Unfortunately, this seems to be
> +	   untestable. Loading an external stylesheet and then having an inline

> +	   script hit this code did not result in an pending stylesheets.

You should talk to Hyatt. I'm sure he can give you a way to test this.

If we put too many of these calls in we get "flashes of unstyled content"
(FOUC).

We need to put these into calls that need to function even when called early,
while stylesheets are still loading.

But we should not put it into calls where it's OK if they silently do nothing
when called early, while stylesheets are still loading.

There's a risk that by following my request to do the stronger form of
updateLayout you will introduce FOUC.

> +static bool isTextFieldWithRendererAfterUpdateLayout(HTMLInputElement*
element)

I think the AfterUpdateLayout here is slightly too explicit.

> +    ASSERT(!document()->childNeedsStyleRecalc() ||
document()->inStyleRecalc());

If the assertion is going to take this form, then maybe we should expose this
expression instead of the raw inStyleRecalc value.

r=me