Dmitriy,
I spent a few minutes yesterday trying to track down the seg. fault -- there
are at least three issues here...
One appears to be a race condition inside of Tcl/Tk-to-Python call interface,
which is most apparent on multi-cpu systems, and for which I have no solution
yet other than disabling the external GUI altogether.
The second is a blunder on my part in the CmdReady function in layer4/Cmd.c:
static PyObject *CmdReady(PyObject *dummy, PyObject *args)
{
return(APIResultCode(TempPyMOLGlobals->Ready));
}
should be
static PyObject *CmdReady(PyObject *dummy, PyObject *args)
{
if(TempPyMOLGlobals) {
return(APIResultCode(TempPyMOLGlobals->Ready));
} else {
return(APIResultCode(0));
}
}
since TempPyMOLGlobals can't be derefrenced until it exists.
The third issue appears to be a Python initialization issue that also only
rears its ugly head on SMP machines:
finish_launching() in modules/pymol/__init__.py needs to be updated as follows:
def finish_launching():
e=threading.Event()
while not hasattr(__main__,'pymol'):
e.wait(0.01)
while not _cmd.ready():
e.wait(0.01)
while not hasattr(__main__.pymol,'xray'):
e.wait(0.01)
in order to allow time for the xray module to initialize before loading
structures.
These changes have been committed to CVS.
Cheers,
Warren
--
Warren L. DeLano, Ph.D.
Principal Scientist
. DeLano Scientific LLC
. 400 Oyster Point Blvd., Suite 213
. South San Francisco, CA 94080 USA
. Biz:(650)-872-0942 Tech:(650)-872-0834
. Fax:(650)-872-0273 Cell:(650)-346-1154
. mailto:war...@delsci.com
> -----Original Message-----
> From: Dmitriy Igor Bryndin [mailto:brynd...@msu.edu]
> Sent: Tuesday, October 11, 2005 5:06 AM
> To: Warren DeLano
> Cc: pymol-users@lists.sourceforge.net
> Subject: Re: [PyMOL] PyMol segmentation fault while starting
> from external python program
>
> Warren
>
> It seems like PyMol should have time to load after import
> pymol and before anything else can be done with it.
> If I´ll do
> ------------------------------
> pymol_argv = ['pymol', '-qx']
> import pymol
> # let`s give pymol some time to load
> for a in range(1000000):
> b=0
>
> pymol.finish_launching()
>
> from pymol import cmd
>
> cmd.load("$HOME/pept.pdb")
> cmd.show("sticks")
> ------------------------------
> everything is fine. It loads and shows the picture.
> ´for a in range(100)´ for example will produce segmentation fault.
>
> As far as I understand ´pymol.finish_launching()´ is meant to
> produce such delay. Correct me if I´m wrong. But this call
> produce segm fault by itself without a waiting cycle.
>
> Can you recommend some workaround for this problem? This
> stupid cycle may work on my machine, but may not work on a
> faster one. Is there any way to find out that PyMol window
> had loaded?
>
> Thanks
> Dmitriy Bryndin
>
>
>
>
>
>
>
>
> Warren DeLano writes:
>
> > Dmitriy,
> >
> > Due to problem with multithreading (especially with the Tcl/Tk
> > external GUI), we haven't been able to get the "import
> pymol" approach to work in
> > a stable robust manner across different OSes and environments.
> >
> > So sight now, as per comments in
> "modules/pymol/__init__.py", the only
> > supported way to launch PyMOL is to run the __init__.py script on
> > startup.
> >
> > %python modules/pymol/__init__.py
> >
> > Hoever, if you disable the external GUI, then you might be
> able to get
> > "import pymol" to work...
> >
> > pymol_argv = ['pymol', '-qx']
> > import pymol
> > pymol.finish_launching()
> >
> > from pymol import cmd
> >
> > cmd.load(...etc.
> >
> >
> > Cheers,
> > Warren
> >
> >
> > --
> > Warren L. DeLano, Ph.D.
> > Principal Scientist
> >
> > . DeLano Scientific LLC
> > . 400 Oyster Point Blvd., Suite 213
> > . South San Francisco, CA 94080 USA
> > . Biz:(650)-872-0942 Tech:(650)-872-0834
> > . Fax:(650)-872-0273 Cell:(650)-346-1154
> > . mailto:war...@delsci.com
> >
> >
> >> -----Original Message-----
> >> From: pymol-users-ad...@lists.sourceforge.net
> >> [mailto:pymol-users-ad...@lists.sourceforge.net] On Behalf
> Of Dmitriy
> >> Igor Bryndin
> >> Sent: Monday, October 10, 2005 2:33 PM
> >> To: pymol-users@lists.sourceforge.net
> >> Subject: [PyMOL] PyMol segmentation fault while starting from
> >> external python program
> >>
> >> Launching PyMol form external python script will produce
> >> segmaentation fault.
> >> For example starting "launch.py" from "/pymol/examples/launching"
> >> ----------------------------------------------
> >> $ python launch.py
> >> zsh: segmentation fault python launch.py
> >> ----------------------------------------------
> >> Crashes without even showing PyMol windows.
> >>
> >> Tried it on Fedora Core 1, Fedora Core 4, Mandriva 2005. The same
> >> story.
> >> Different pythons and compiling different versions of
> PyMol does not
> >> change anything.
> >>
> >> It will launch PyMol windows if there is only
> >> import pymol
> >> line.
> >>
> >> Adding
> >> pymol.finish_launching()
> >> or
> >> from pymol import cmd
> >> cmd.load("$PYMOL_PATH/test/dat/pept.pdb")
> >> will produce segmentation fault. With no windows shown.
> >>
> >> The same time if I'll try to debug step by step, let's say,
> >> "launch_demo.py"
> >> (from "/pymol/examples/launching")
> >> -------------------------------
> >> pymol.finish_launching()
> >> from pymol import cmd
> >> cmd.load("$PYMOL_PATH/test/dat/pept.pdb")
> >> cmd.show("sticks")
> >> -------------------------------
> >> using IDLE. Everything will work. It starts windows, loads a file,
> >> changes to sticks...
> >> Running it "python launch_demo.py" will wait for a second
> and return
> >> segmentation fault.
> >>
> >> If someone knows what's going on, please help me.
> >>
> >> Thanks
> >> Dmitriy Bryndin
> >>
> >>
> >>
> >>
> >>
> >> -------------------------------------------------------
> >> This SF.Net email is sponsored by:
> >> Power Architecture Resource Center: Free content, downloads,
> >> discussions, and more. http://solutions.newsforge.com/ibmarch.tmpl
> >> _______________________________________________
> >> PyMOL-users mailing list
> >> PyMOL-users@lists.sourceforge.net
> >> https://lists.sourceforge.net/lists/listinfo/pymol-users
> >>
> >>
> >>
> >>
> >
>
>
>
>
>
>
>
