On Tue, 2003-01-28 at 20:06, Henrik Nordstrom wrote:
> I do not think refcounting is really needed. acls are only created > once and never destroyed.. A reconfigure only reloads the ACL data, > it does not destroy the actual acl. I hadn't dug in that much. My recollection though is that a new acl is created and the old one cbdatafreed(). There are checks for that condition in acl.cc. > Having the cache operate without change for the lifespan of the > request even during/after a reconfigure is probably preferable in my > opinion. Yes, that is the goal: Extant requests operate on the old data/results. New requests only operate on current at the time acl's. Rob -- GPG key available at: <http://users.bigpond.net.au/robertc/keys.txt>.
signature.asc
Description: This is a digitally signed message part
