Wed, 11 Feb 2009 (23:05 +0900 UTC) Stephen J. Turnbull wrote:
robert delius royar writes:
> I do not see it on 10.5.6 (ppc G5) running X11.app
The server is not relevant to this, it's a libXt internal bug. I've
observed similar lockups in TTY-only instances of X11 builds, too.
The problem is the X event gathering mechanism, not the protocol side.
OK, this could be a useful bug for Jeremy if it is triggered reliably.
He is involved in a conversation on the x11-users [perhaps x11-dev?]
list with a few users who are finding CPU use to run up to 100%
apparently as part of an event-queue problem. He has not been able to
reproduce it (and others have also not been able), but some can create
it consistently on their machines.
I think it's a system, depending on all of the components to
manifest. However, XtAppPending is documented to *never* block, so
you're all at risk; XEmacs is entirely defenseless if XtAppPending
Also, I've gone to some effort to make sure that XEmacs picks up the
MacPorts versions of the libraries (among other things, I now have
unstripped X11 libraries available :-). If you're using the Apple
versions, they're probably somewhat older.
I do know that when I install Jeremy's alpha or beta versions, some of
the libraries in /usr/X11/lib get updated (in addition to the quartz
stuff and the server / starter). Also, I thought that he said the
macports builds would only be newer than the ones he rolls for
macosforge for those who are using OS X 10.4. I don't believe the
macosforge files run without 10.5. They did not used to, at least.
Dr. Robert Delius Royar Associate Professor of English
Morehead State University Morehead, Kentucky
XEmacs-Beta mailing list