On Mon, May 23, 2005, [EMAIL PROTECTED] said: > A BUGNOTE has been added to this bug. > ====================================================================== > http://www.dbmail.org/mantis/bug_view_advanced_page.php?bug_id=0000057 > ======================================================================
> ---------------------------------------------------------------------- > niblettda - 23-May-05 14:02 CEST > ---------------------------------------------------------------------- > I think my patch is mostly compliant with what you are saying. (I still > don't completely get the ACL and anyone user, but I'm working on that.) > > The only difference is that my "ACL" specification happens in the > dbmail.conf file rather than the ACL table. Being able to specify which folders can be created is good. > I agree with Aaron, and understand what he was saying about where folder > creation should happen, however that doesn't exist at this time, so I > worked with what I had. Right now I set the folder when the LMTP RCPT > line is processed, which still seems logical to me. The problem is that deeper code is unaware of who or how this mailbox was set in the dsn_user structure, and that is critical for determining if the mailbox should be created or not. > I expected that RFC wouldn't allow creation of the box unless it exists. Why not? > I also think that it would be best to create the "deliverable" boxes when > the user is created. Maybe it would be better for me to modify the > dbmail-user code and create everything in the CREATE_FOLDERS list and have > LMTP only accept delivery for boxes that exist. Can't be done for LDAP users, among other things. > This could certainly help with one problem in that I can't figure out how > to subscribe the user to the mailbox when it's created. Right now, I > determine if the box needs to be created, and add that to the dsn_user > structure. > I can't subscribe the user to the box because the box and Email don't > exist yet. The function find_create_mailbox() is where the hard work should be. It needs to take an additional argument telling it where the mailbox spec came from. > I'll take my direction from the developers because that's who is going to > have to deal with the legacy if it ever gets added to the code. Very appreciated, thanks for putting the effort into this project! Aaron
