Re: [exim] RSET command

Abhijit Menon-Sen wrote:
> At 2006-11-07 09:52:07 +0800, wbh@??? wrote:
>> The intent is to determine experimentally, the 'real', not
>> theoretical, response of a submitting client [1], if/as/when the
>> server with which it has estabished a connection were to issue a
>> 'RSET'. More accurately, an unexpected RSET.
>
> A server can't issue RSET, so your question is meaningless.
>
> -- ams
>

A 'server' - or client - can do whatever it is programmed to do.

RFC's and supposition as to what 'should be' are not the same thing as
executable code. The 'net is full of marginally incorrect clients, and more than
a few MTA implementations as well.

'meaningless' is what is "To Be Determined", most particularly w/r spambots -
the majority of which might reasonably be presumed to depart from the
specifications upon which we base conventional expectations.

By no means all spambots will abandon a connection in a fraction of that time,
(delay = 30s), but a useful percentage will do so. A very, very much smaller
number will hang on for more than ten hours, if you care to permit that.

Logged, not theoretical, in both cases.

Yet nothing in the RFC's or 'best current practice' would lead one to expect,
for example, an smtp timeout of much less than 4 wall-clock minutes nor much
more than 5 minutes.

Research is not the same thing as a recommendation or 'feature request', but
some things provide the odd surprise when tested.


Bill