Testing whether Xlib threading support works in XORG-RELEASE-1branch...

[Roland Mainz]

Jim Gettys wrote:
> While I don't doubt there could be problems, you should
> be aware in the past that Mozilla's threading itself was badly
> broken (IIRC, at the time wasn't even actually calling Xlib's thread
> initialization at all; this was in late Netscape days).  I
> remember a very frustrating debugging session on the
> phone with Keithp where we ended up having to conclude we
> couldn't fix it (it being Netscape, without source available),
> and that the fault was 100% Netscape's.

Erm... is it possible that you're mixing problems with NS4.x with issues
in NS7.x/Mozilla1.x ? Mozilla (that one from http://www.mozilla.org/ -
which is OpenSource (see http://lxr.mozilla.org/)) shouldn't be a big
problem since AFAIK all rendering is single-threaded.

> So before going down this (potential) rat-hole, what makes
> you believe that the tests themselves are not buggy?  Chasing
> threading bugs through mozilla isn't something we'd like
> to do, given its size.

The idea of testing the Xlib threading support with Mozilla has a
different intention here: I'd like to test whether the Xlib+X extension
threading code works with a real-world application which is used outside
a lab environment. I've seen lots of problems and rants about the Xlib
thread support code... Eclipse people ran into it and had to work around
it, JRE people hit it and had to work around it etc. ... and I'd like to
end this mess. That's why I added some test code to the Mozilla codebase
that the OS vendors can test their Xlib implementations against that.

BTW: A quick test with Xfree4.3.0 shows that even an application which
only has one thread hangs in libGTK initalisation thanks to a deadlock
in the Xinput extension library... it seems that the original authors of
the threading support in Xlib assumed that a mutex can be recursively
entered... ;-(

----

Bye,
Roland

-- 
  __ .  . __
 (o.\ \/ /.o) roland.mainz at nrubsig.org
  \__\/\/__/  MPEG specialist, C&&JAVA&&Sun&&Unix programmer
  /O /==\ O\  TEL +49 2426 901568 FAX +49 2426 901569
 (;O/ \/ \O;)