Re: Seg fault with XQuartz 2.6.0 (xorg-server 1.9.3)
Re: Seg fault with XQuartz 2.6.0 (xorg-server 1.9.3)
- Subject: Re: Seg fault with XQuartz 2.6.0 (xorg-server 1.9.3)
- From: Jeremy Huddleston <email@hidden>
- Date: Thu, 6 Jan 2011 10:39:29 -0800
On Jan 6, 2011, at 02:38, Giles Chamberlin wrote:
>
> Can you explain "xcb backed" vs "xtrans backed"? I don't recognise the
> terms.
libX11 needs to communicate with the server. At the base level, this can be done with sockets, pipes, etc. The way that libX11 actually opens and deals with these underlying methods is through wither xtrans or libxcb. xtrans is legacy code that goes back to the original X11 days at MIT and as such was created in a much different world than we have now. libxcb on the other hand allows for better support of extensions, easy binding to other languages, and employs more modern techniques. It has allowed for great performance improvements in the generic cases and huge/massive improvements in edge cases with tons of client-server communication.
The libX11 shipped with Snow Leopard still used the legacy xtrans support. Over the past year, I've squashed the remaining known regressions when using libxcb, so XQuartz 2.6.0 now uses the libxcb support in libX11. Many of the Linux distributions, including Ubuntu, have been using libxcb for a few years now.
Since the crash is occurring in _X11TransWrite (which is part of xtrans), I'm curious what will happen if you instead use an xcb-backed libX11 (like the one in /opt/X11). That might explain why it works fine on Ubuntu but not XQuartz 2.3.6.
>
>> Also, a full crash log would be helpful. You can copy/paste the crash log or
>> just find it in ~/Library/Logs/DiagnosticReports
>>
>
> Attached - and thank you for your help.
>
> Giles
>
>
>
> <emacs_2011-01-05-141921_gilecham-mac.crash>
_______________________________________________
Do not post admin requests to the list. They will be ignored.
X11-users mailing list (email@hidden)
This email sent to email@hidden