Joshua Gatcomb <[EMAIL PROTECTED]> wrote: > --- Leopold Toetsch <[EMAIL PROTECTED]> wrote: >> $ make realclean >> $ perl Configure.pl --intval=long --opcode=long
> This was the ticket!!!! Thanks. Great. > For my own > edification, can you explain to me what in the code I > isolated that changed on the 15th accounts for this? It was the assumption that a PMC is not smaller then a Dead_PObj structure, which obviously is wrong. Other code paths that deal with Dead_PObj do check the size. But at the place, where you got the segfault, there wasn't such a check because of that assumption. The Dead_PObj structure had another problem, which is fixed now: I thought that pobj->buflen is an UINTVAL. Actually it's a size_t, which might not be the same length. BTW: STRINGs have an UINTVAL strlen and a size_t buflen, which might not be very useful. > ivtype='long long', ivsize=8 [ get rid of perl5 types in config ] > I think this is something I could do if I knew what it > *should be* using/getting the values from. Good question. But normally we should have: sizeof(void *) == sizeof(opcode) == sizeof(INTVAL) The sizeof(void *) is the "natural" size of the system. When we allow sizeof(INTVAL) == 8 on 32bit system, we need a big code cleanup. U?INTVALs are used all over the place just as plain ints. This was discussed earlier several times, but isn't really solved IMHO. > Thanks again Welcome. > Joshua Gatcomb leo
