[PPL-devel] Segmentation faults with Ciao 1.9#44

[Roberto Bagnara]

Dear Ciao people,

one of the problems we faced in our attempt to interface the
Parma Polyhedra Library with Ciao Prolog, was a segmentation
fault that we temporarily cured, as suggested by Manuel Carro,
by inserting `GLOBALSTKSIZE=1000000' into the environment.

About one week ago, as a consequence of some improvements we
made to the PPL-Ciao interface, another segmentation fault
showed up.  We thought it could be cured by increasing
`GLOBALSTKSIZE' even further, but we were wrong.  So we started
thinking we had introduced a bug.
After a significant amount of debugging work, we have discovered
that the new segmentation fault goes away if one leaves
`GLOBALSTKSIZE' undefined.

In summary, we have two testcases:

1) one bombing if `GLOBALSTKSIZE' is not defined;
2) another one bombing if `GLOBALSTKSIZE' is defined.

Does this ring a bell?
Do you want us to set up a directory here in Parma
so that you can log in and play with the testcases?
Cheers,

      Roberto

-- 
Prof. Roberto Bagnara
Computer Science Group
Department of Mathematics, University of Parma, Italy
http://www.cs.unipr.it/~bagnara/
mailto:bagnara at cs.unipr.it