[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: [workaround] Recent upgrade killed matlab [and everything java related]

On Mon, 7 May 2007 00:01:13 +0300
Micha Feigin <michf@post.tau.ac.il> wrote:

> On Sun, 6 May 2007 10:24:09 -0600
> "Wesley J. Landaker" <wjl@icecavern.net> wrote:
> > On Tuesday 01 May 2007 01:58:49 Micha Feigin wrote:
> > > Recent upgrades of sid killed matlab 7 (sp2) for me. When I try to start
> > > it with the jvm active (gui) I get the error:
> > >
> > > MATLAB: xcb_xlib.c:50: xcb_xlib_unlock: Assertion `c->xlib.lock' failed.
> > > Aborted
> > >

Turns out that the new xcb support is not compatible with java's awt. The only
solution I found was to downgrade libx11 to pre xcb support and uninstall all
xcb packages.

Turns out this problem attacks anything java on any xcb enabled libx11

> > > Any idea what this is, is it solvable and whether it is a common problem
> > > or only me ?
> > 
> > No idea about your matlab issue (looks like a possible xlib 
> > incompatibility--you could always run matlab in a etch chroot or 
> > something), but I can't pass up the opportunity to recommend using octave 
> > <http://www.octave.org/> instead of matlab if at all posible. =)
> > 
> Afraid it's not even remotely possible.
> I tried octave at the time but it was so far behind it was on the verge of
> unusable for the real world. Scilib seemed much better but both fail on the
> colaboration issue when everyone around you uses matlab (or as I just started
> doing, c/c++  and fortran).
> I do computer vision and medical imaging, which means huge PDE and (non)linear
> systems. Also one of octave's big drawbacks (besides being about on par with
> matlab 5 without the toolboxes) is that it has terrible graphic support
> (plotting/displaying data, including 3d)
> Add to that the availability of existing code ...
> Anyway, it turns out that the problem is not matlab, its something between
> Java's awt and x.org (I didn't quite understand who's at fault, seems that awt
> uses something that is in the specks but not properly implemented). I just
> found out that jabref (open source bibref manager), freemind (open source mind
> mapping tool) and maple are also dead.
> I saw some pointer towards libxrandr but it seems that it is at it's latest
> version.

Reply to: