On Thu, 12 Jul 2001, Artur Bergman wrote:
> 01-07-12 19.11, skrev Doug MacEachern p� [EMAIL PROTECTED] f�ljande:
>
> > this patch fixes the problem for me, all tests pass now (except b-deparse
> > #15 which has been failing for quite a while)
> >
> > --- op.c 2001/07/12 06:35:45 1.1.1.49
> > +++ op.c 2001/07/12 17:03:48
> > @@ -2951,7 +2951,7 @@
> > #ifdef USE_ITHREADS
> > {
> > SV* repointer = newSViv(0);
> > - av_push(PL_regex_padav,repointer);
> > + av_push(PL_regex_padav,SvREFCNT_inc(repointer));
> > pmop->op_pmoffset = av_len(PL_regex_padav);
> > PL_regex_pad = AvARRAY(PL_regex_padav);
> > }
> >
> Huh? The segfault problem?
>
> I fixed it with another patch allocating a new entry in regcomp.h. Doesn't
> this create a leak?
looking at it again, our patches are not related. your patch to regcomp.c
just creates the PL_regex_padav entry for PL_reg_curpm, mine applies to
all the rest in newPMOP. as for leaks, i don't see any running the leak
program i posted a while back (the ones i saw then were cured by #11188)
i'm curious, how does PL_regex_padav get released? i don't see a
SvREFCNT_dec(PL_regex_padav) anywhere.
