Am 05.11.2014 17:31, schrieb Wietse Venema:
> Viktor Dukhovni:
>> On Wed, Nov 05, 2014 at 10:47:33AM -0500, Wietse Venema wrote:
>>
>>> postfix-us...@arcsin.de:
>>>> Since the tcp_table is consulted with a full address, it can produce wrong
>>>> answers even if it is placed at the last position of transport_maps.
>>> I have a revolutionary suggestion: DO NOT REPLY WITH INCORRECT INFORMATION.
>> Well, "wrong" is perhaps not the best way of putting it. Since
>> tables earlier in the list might only have partial keys, and the
>> full key is queried first, that "last" table may give a result
>> before the "higher priority tables" had a chance to give a "better"
>> answer.
> This should be addressed by making hard-coded table lookup
> behavior configurable. This could look like:
>
> mumble_maps = {maptype:mapname, flags={fixed,...}, ...} maptype:mapname, ...
>
> All this can be implemented in one place under the dict_open API,
> without needing changes in any other code. But it does of course
> require documentation changes with Postfix table-driven features.
>
> Likewise there is a need to specify the order of full and partial
> queries, but that could be specified via separate parameters:
>
> virtual_alias_maps_search = full, user, @domain
> access_maps_search = full, user@, domain, parent-domain
> transport_maps_search = full, domain, .parent-domain
>
> Wietse
Hi Wietse,
hi Viktor,
I would like to see this feature. :)
Regards
Damian
