Re: Announcing cvm version 0.50

Bruce Guenter <[email protected]> writes:

> On Wed, May 11, 2005 at 04:14:58PM -0400, Dale Woolridge wrote:
>>     Citing only one example (cvm-vmailmgr chained with cvm-qmail), if the
>>     cvm-vmailmgr module gets to a point where qmail_lookup_cvm() was
>>     successful, but the virtuser doesn't exist, that's authoritative;
>>     cvm-qmail shouldn't even be consulted.
>
> Not necessarily.  If there is a .qmail-someuser-default file, vmailmgr
> won't know about it, but messages will still be deliverable by qmail.
> Next?

What do you think is a better approach, then, Bruce?  A custom module
that combines qmail and vmailmgr?

I've been working a bit with cvm-chain, and I ended up doing almost
exactly what Dale described: provide cvm-vmailmgr a way to say "I
don't know, that's not my domain", and have those requests fall
through to qmail.  I deal with the potential problems you describe
with simple policy: don't create .qmail files in a virtual domain.  So
far it's worked fine, though I can imagine other systems where
something more complex might be needed.

----ScottG.