On Thu, April 21, 2011 7:11 pm, Wietse Venema wrote: > Dominik Schulz: >> > > because my transport is special crafted to return a default value >> for >> > > non-existing users, i.e. it can never fail. >> > Your table does not satisfy Postfix requirements. >> I'm not trying to be mean, I'd just like to use Postfix in way that's >> not the >> regular use-case. > Neither am I. I just explain how Postfix works, to avoid disappointments > due do invalid expectations. Thank you for the clarification. You were right, as always ;), from the beginning. My proposal would have changed the way Postfix handles those lookups. I just didn't realize it. > You're welcome to design, implement and document a different query > format, but you can't change the way that Postfix handles the "not > found" condition. I'm not sure if I am up to that task, but perhaps someone will provide some additional lookup macros some time. You can probably live without those most of the time but it would at least allow cleaner queries and add bit flexiblity.
On Fri, April 22, 2011 1:55 am, Jeroen Geilman wrote: > As documented, "user" is tried if "user+extension" does not yield a > result. > You could easily add a static transport_maps entry AFTER your mysql map, > like so: > transport_maps = mysql:/etc/postfix/transport.cf, > static:smtp:[nexthop] > First, all normal map lookup keys will be tried, and then, if there was > no result, the static value is returned. That looks like a good idea, too. But in my case this behaviour is only desired for one specific domain, not all domains handled by this server. So I can not use this solution. On Fri, April 22, 2011 2:42 am, Noel Jones wrote: > The best solution is to craft an sql lookup that doesn't > respond to addresses with extensions, or at least so the > catchall doesn't match extensions. That is exactly what I did in the end. Btw. if anyone is interested in my setup, see [1]. -- Best Regards, Dominik Schulz [1] - http://blog.gauner.org/blog/2011/03/10/postfix-partial-relaying-to-exchange/
