[Webkit-unassigned] [Bug 34773] [GTK] Hits assertion on history back, with page cache enabled, in specific conditions

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=34773





--- Comment #4 from Gustavo Noronha (kov) <gns at gnome.org>  2010-02-10 08:17:57 PST ---
(In reply to comment #3)
> (In reply to comment #2)
> 
> Just to update this after some chat through jabber.

Too quick, because I was soo hungry =D

> > - I don't think the test really needs to open new windows? I certainly got the
> > crashes without doing that.
> 
> Apparently this was only done to organize the test, but it's not needed for it
> to crash. Guess we can try to get rid of it.

Yes, but I don't think it's worth the effort. The test works fine like it is,
and it uses the same pattern as fast/harness/use-page-cache.html, so unless you
have something ready I think we can leave it as it is.

> > - Do you really need the fourth page just for the PASS? Can't you just print
> > that  when you are back at the beginning or something? Or just get rid of it
> > completely, since the tests passes if you don't crash anyway.
> 
> Didn't discuss this :P

The fourth page was only needed to cause the first one to be evicted from the
page cache. I guess we could replace this with a first page that is never put
in the page cache instead. Would be a better solution overall.  I'll see if I
can do that.

> > - Do we really need the bit about saving the values in the adjustments?
> > Shouldn't that be set by the view when it's restored automatically through the
> > usual path? Or was that dependent on a new view being created?
> 
> This one seems to be needed because although the view indeed restores the
> values it does it before we are able to restore our adjustments, so it's
> useless. If this is the case there should be some FIXME/comment in the code
> about it.

Yeah, exactly. There could be other ways of triggering this, but I think this
is a good solution that has low risk of unwanted side effects. I'll add a
comment.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.