scp@acl.lanl.gov (Stephen C. Pope) (03/15/90)
After bringing my perl up to PL14 and working all day on an old script (esp. making use of the new (LIST)[SLICE] trick), I'm finding some very irregular, non-deterministic behavoir that I'm loath to debug: I'm getting spurious complaints from Larry's malloc that I've got a ``bad free() ignored''; it happens pretty randomly and in no repeatable (that *I* can figure...) fashion but always in statements involving arrays. Also, I have a subroutine which does a system "cd /etc/yp; make". I'm doing nothing to set or alter the environment or path, but about 75% of the time when this subroutine is invoked from one point in the program, various things like ``makedbm'' spawned by the ``make'' die a horrible and unexplained death, but this happens only about 5% of the time when the subroutine is invoked from another point; again, very stochastic... Thing is, (cd /etc/yp; make) from the command line always works 110% of the time. I'm not *sure* I can blaim this on perl, but I'm kinda suspicious... Anyone on the same or similar system seen such behavoir? Anything I might be doing wrong to provoke the malloc free() problems? Any tips on tracking this down? stephen pope advanced computing lab, lanl scp@acl.lanl.gov
lwall@jpl-devvax.JPL.NASA.GOV (Larry Wall) (03/16/90)
In article <SCP.90Mar14172125@blanche.acl.lanl.gov> scp@acl.lanl.gov (Stephen C. Pope) writes:
: After bringing my perl up to PL14 and working all day on an old script
: (esp. making use of the new (LIST)[SLICE] trick), I'm finding some
: very irregular, non-deterministic behavoir that I'm loath to debug:
Especially since it's already been debugged. Patchlevel 14 is known to
be unstable on many machines, so please install patch 15.
Larry