[Webkit-unassigned] [Bug 168699] [GTK] Hangs when showing Google search results

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=168699

--- Comment #16 from Carlos Garcia Campos <cgarcia at igalia.com> ---
(In reply to comment #15)
> Comment on attachment 302650 [details]
> Updated patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=302650&action=review
> 
> > Source/WebKit2/Platform/IPC/unix/ConnectionUnix.cpp:477
> > +#if PLATFORM(GTK)
> 
> Why is there anything GTK-specific in ConnectionUnix.cpp? Can't we align our
> code with the other ports to avoid GTK-specific bugs?

There's nothing GTK+ specific, it's glib, but EFL didn't use our gsocket monitor, I don't know why, so we had to use PLATFORM(GTK) instead of USE(GLIB). I agree we could have used USE(GLIB) && !PLATFORM(EFL), though. This bug is not GTK specific at all, the problem was in the poll we used when sendmsg returns EWOULDBLOCK. The solution is GTK+ specific, well, GLib specific, because we use our socket monitor to properly poll the socket also for writing in case of EWOULBLOCK using the run loop.

> At the very very least, all of these should be changed to if USE(GLIB),
> right?

Right, now that EFL is gone, that's exactly my plan. We could also make it ifdef free by converting our socket monitor into a platform-specific class with platform implementations. However, now that there's no other unix ports using this code, it's not easy for me to do that refactoring without knowing the other implementations.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.webkit.org/pipermail/webkit-unassigned/attachments/20170228/4430093f/attachment.html>