Stas Bekman wrote: > Nicholas Clark wrote: > > On IRC Rafael said that he believes that his patch solves the problem > > without needing changes to perl. Is this the consensus view? > > Rafael's patch solves the problem in mod_perl 2.0. The problem still exists > for any other application that embeds perl and does the fork on its own (not > using pp_fork). So 5.8.2 (same as 5.8.1) may break applications that ran OK > under 5.8.0. I've suggested a patch to pp_getppid that fixes this problem in perl.
On the other hand, the problem exists for $$ (there is no getpid function in perl) in all perls. So all applications that embed perl and that fork need at least to save and restore the PID (mod_perl 1 & 2 do this.) It's up to Nicholas to see if he wants to restore the 5.8.0 behavior in 5.8.1 but I think that PL_ppid will stay in 5.10. I don't think your proposed patch works in the case where perl does a getppid(), so it marks it as cached, then the embedding application does a fork, and the PL_ppid_cached flag stays to "true". (Oh, and sorry for my weak mod_perl patch-fu :) Next time I'll grab the CVS version.) --------------------------------------------------------------------- To unsubscribe, e-mail: [EMAIL PROTECTED] For additional commands, e-mail: [EMAIL PROTECTED]
