From: Keean Schupke (keeanfry-it.com)
Date: Sat Dec 01 2007 - 10:41:33 CST

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

Found what I think is a mistake in the documentation. "dict_del" in
"util/dict_open.c" is documented as returning non-zero on success.
This seems to be wrong, it returns greater than zero if the key does
not exist, zero if the key does exist and is deleted okay, and less
than zero on any other error.

Regards,
Keean Schupke, Fry-IT Ltd.

On 30/11/2007, Wietse Venema <wietseporcupine.org> wrote:
> Victor Duchovni:
> > On Fri, Nov 30, 2007 at 06:36:49PM +0000, Keean Schupke wrote:
> >
> > > Okay, a question about the verify service.
> > >
> > > If I do this, it seems that there will be all sorts of strange
> > > interactions with the normal verify service. You may have conflicting
> > > goals...
> >
> > No, none at all. The verify(8) service is just a database.
>
> I would not use verify(8), because that solves a different problem:
>
> did a probe for a specific email address get bounced or deferred?
>
> The problem at hand is this:
>
> did x allow me to log in with this particular username?
>
> But that is the less important issue.
>
> The reason for a dedicated daemon is that the typical UNIX DB/DBM
> library is not suitable for concurrent updates by multiple processes.
> This is why Postfix has tlsmgr(8) for TLS cache management, and
> verify(8) for address verification management.
>
> Now, this dedicated daemon could be competely outside of Postfix,
> if the client side of the protocol is simple enough.
>
> Wietse
>

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]