> > Specifically, can the *.pmc file be in a different directory than the > > *.pm file that was intended to be used?
If it were implemented exactly as Audrey said, it would -- if checking for $Module.pmc first, then for $Module.pm , which doubles the number of fstats to (upto) 2*scalar @INC. Possibly to avoid this security issue, possibly to buy back the overhead, it appears to have been implemented in a more efficient fashion : > A quick test seems to say no: That's hopeful. That implies it fstats PMC only after finding a PM but before reading the PM, and reduces the suggested 2xfstat overhead greatly. This could be verified with strace/truss. I wonder if this more secure and efficient behavior is an emergent behavior or a specified feature? I was worried about checksum verification of the PM being delegated to the PMC, but any other means of enforcing Inline::C-like "oops, time to rebuild" would be forgeable anyway (at least in open-source languages). -- Bill [EMAIL PROTECTED] [EMAIL PROTECTED] _______________________________________________ Boston-pm mailing list [email protected] http://mail.pm.org/mailman/listinfo/boston-pm
