On Jeu 8 mars 2007 01:00, Pierangelo Masarati wrote:
raphael.ouazana@linagora.com wrote:
I see at least 2 issues in this patch:
- the use of dnParent() in place on malloc'd data would lead to memory
corruption, because at some point the value of the set will be freed, but the bv_val of that value would no longer point to the beginning of a malloc()'ed block. You should rather use AC_MEMCPY() to move the parent's DN at the beginning of the malloc()'ed memory block.
- when trimming a set of DNs to some ancestor, DNs with common
ancestors could result in duplicate set values. I'm not sure this currently has any adverse effect on set evaluation, but I'd consider it bad anyway. You should eliminate duplicates, in case any arises.
Fixing these issues, I saw that set_parent will be very similar to set_chase. Do you think I should create a new gatherer instead of ?
Regards, Raphael Ouazana.