[MMTK] Runtime errors

Julien Michel J.MICHEL at soton.ac.uk
Wed Oct 15 10:51:00 CEST 2003


        I was actually able to trace the error to a few lines of C code. 
The script protein.py is useful as an example.
    In that script, setup and equilibration works fine but the program 
crash during the production run and this is caused by the option 
'Trajectory output(...)'.  If I comment this option the script pass the 
production (and crash later because the trajectory file is empty).
    The line that cause the 'Memory error' seems to be
     if (PyTrajectory_Output(output, i, data_descriptors, &this_thread) 
== -1) {
     PyUniverseSpec_StateLock(universe_spec, -2);
      PyEval_RestoreThread(this_thread);
      goto error;
    }
    in MMT_dynamics.c aorund line 700, just before the comment /*First 
Part of integration step*/
   
    Sadly I don't understand why PyTrajectory_Output() is equal to -1 in 
this particular case.

    The water example in the manual crashes for the same reasons as well.

    Does this ring a bell to somebody ?

    Julien Michel

>
> I list below the error message associated with each failing script.
>Does somebody know if this is can be linked to an installation problem ?
>  
>
>
>Unless you really have very little memory, this looks like a problem with something outside MMTK, as MMTK delegates all memory handling to Python.
>
>Is anyone else using RedHat 9? I wonder if they did something to their Python installation.
>
>Something you could try is to install your own Python from the sources, e.g. in /usr/local.
>
>Konrad.
>
>
>.
>
>  
>






More information about the mmtk mailing list