Re: [exim] Not frozing on unexistent user

On 21 Feb 2006, at 01:18, John W. Baxter wrote:

> On 2/20/06 2:04 PM, "David Woodhouse" <dwmw2@???> wrote:
>
>> On Fri, 2006-02-17 at 20:58 +0000, Peter Bowyer wrote:
>>> verify = recipient/callout
>>
>> One should never say the above without explicitly including the
>> 'use_sender' option -- since recipient verification without
>> use_sender
>> is fairly much broken.
>>
>> verify = recipient/callout,use_sender
>
> "Never" seems a bit strong. If one controls the server being
> called out to
> and knows that the sender doesn't matter to that server's response,
> then it
> shouldn't matter. On the other hand, in that situation, use_sender
> doesn't
> hurt much, either (it may enlarge caches, or I would just say "doesn't
> hurt").

Quite correct. I would say that the use_postmaster option is good
too, as it would help the caching and would work also in cases where
VERP is used (unless the receiver accepts <> for regular local parts
as far as RCPT TO is concerned). Of course this does not work in case
of blacklists on senders at the receiving end, but this would be a
design fault, as the blocking must always be done on the front end.

Giuliano