So actually, a bit more careful grepping and only user.c is a consumer of the db...alias functions! Everyone else just uses auth_check_user(), which does resolve aliases but also does so recursively, until either an address is found to be a forward, exec, or a numeric user account.
I do still think that the alias functions should have auth_ equivalents, but I'm also thinking that it might be very useful to have some things still in the database. I can think of at least a few situations where I would want to be able to enter addresses into the database rather than rely on LDAP. Postmaster is a favorite example, because we'd want to be able to deliver to postmaster even if the LDAP has gone down, and perhaps in fact BECAUSE it goes down, send an error message over to postmaster. But, then again, isn't postmaster even better off as a direct alias from Postfix, or any other MTA? What if DBMail goes down and... Aaron On Fri, 10 Jan 2003, Aaron Stone wrote: > Here's another idea that builds upon this, with fewer changes... since I > can think of at least one potential auth mechanism that won't work well > with aliases - PAM - and so perhaps we should do a combination effort: > > Each db implements the alias function as db_...alias... > > All calls to the ...alias... function MUST be to auth_...alias... and > unless the auth mechanism can provide its own aliasing, just pass along > the call to the underlying db_...alias... function. > > In this way, an auth module could also use both its own aliases and > aliases from the database to create a more configurable environment. > > Or maybe it's just needless complexity, and I'm scratching my head hoping > that someone is listening and thinking about alternative auth's... > > Aaron > > > On Tue, 7 Jan 2003, Aaron Stone wrote: > > > Hi, > > > > As I get further along on the LDAP auth module, I've noticed that the > > alias table functions are all in the db.h and not in the auth.h > > abstraction. Since the "native LDAP" posixAccount and MS's Active > > Directory "user" classes both include the possibility for multiple mail > > and mailAlternate attributes, it makes sense to use these fields as the > > aliases for the given account. > > > > Perhaps we could move these functions into auth.h... > > > > u64_t db_get_user_from_alias(const char *alias); > > char* db_get_deliver_from_alias(const char *alias); > > > > int db_get_user_aliases(u64_t userid, struct list *aliases); > > int db_addalias(u64_t useridnr, char *alias, u64_t clientid); > > int db_addalias_ext(char *alias, char *deliver_to, u64_t clientid); > > int db_removealias(u64_t useridnr, const char *alias); > > int db_removealias_ext(const char *alias, const char *deliver_to); > > > > > > Aaron > > > > _______________________________________________ > > Dbmail mailing list > > Dbmail@dbmail.org > > https://mailman.fastxs.nl/mailman/listinfo/dbmail > > > > _______________________________________________ > Dbmail mailing list > Dbmail@dbmail.org > https://mailman.fastxs.nl/mailman/listinfo/dbmail >
