[Webkit-unassigned] [Bug 67864] [Qt] Win32 build broken due to MachineStackMarker.cpp/.o failing to link against pthreads library

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Sun Sep 11 16:01:46 PDT 2011


https://bugs.webkit.org/show_bug.cgi?id=67864





--- Comment #8 from Jarred Nicholls <jarred at sencha.com>  2011-09-11 16:01:46 PST ---
(In reply to comment #7)
> > r+, but please add comment to the platform.h to make developers
> > not to try to remove it again until we find a proper way for
> > supporting multithreading on Qt-Win platform.
> 
> Guys, this is upside-down. We don't want to hold back JavaScriptCore development waiting for someone to figure out the threading story on QtWin32. This is especially true since, according to Andreas on IRC, nobody develops QtWin32, so we may be waiting forever.

If necessary, I'll work on it.  I have products that need Qt-Win32 support so breaking it perpetually is not going to be something I support.  I can keep a patch around to put the status quo back into place until I fix it all up.

> 
> We want to remove ENABLE_JSC_MULTIPLE_THREADS and all related code entirely (along with ENABLE_WTF_MULTIPLE_THREADS and others), so development can proceed with the baseline assumption that all platforms at a minimum support the concept of a thread. This is a critical baseline abstraction for a number of important algorithms. (See my email to webkit-dev.)

Before going further, wouldn't it be wise to then abstract the notion of a thread rather than going against pthread as if it's the only option, before proceeding with development?  I haven't analyzed this more than 10 minutes to know if that's even allowable without breaking the current interface, i.e., changing pthread specific signatures to library independent ones that support a client interface et al.

> 
> It would be nice if a QtWin32 developer would fix QtWin32 to support the baseline abstractions required by ENABLE_JSC_MULTIPLE_THREADS. But if that can't happen, the right solution isn't to hold JavaScriptCore at a standstill -- it's to remove QtWin32 from the core set of builders.

I wouldn't call that a solution to the problem, since the actual problem is JSC being married to pthread in the multi-threaded areas.  But it is what needs to happen if we have to remove those flags while the real solution is being implemented.

So, since there's no set timeline on getting JSC pthread-independent, let's either reopen this ticket or create a new one to do so, and keep Qt-Win32 out of the core build set.  When will you be removing these flags for good Geoff?  Are you ready to do it immediately after removing QtWin from core builders, or are you still doing other cleanups first?

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



More information about the webkit-unassigned mailing list