On Wed, 2 May 2001, David Saez Padros wrote:
> Not the first time, it should retry the usual way until it reaches the
> maximum retry times, that's when it should (in my opinion) try without TLS.
This is fine in theory; in practice the implementation makes it very
hard to do this. The checking for timeouts happens at an entirely
different place, after Exim has tried to deliver to all the possible
hosts for the domain.
> Maybe you can think that a temporary error that repeats forever is just
> a permanent error.
It is, when the retry time for the host times out. But there is no
apparatus for "retry this particular host (possibly one of several
hosts) again using a different set of rules".
In order to implement what you want, Exim would have to remember each
host for which it had a temporary STARTTLS error, and then back out of
timing it out and set up some way of retrying with different options.
I'm afraid the code is not designed to facilitate this; it's cutting
across several different "phases" of the way Exim processes remote
deliveries.
--
Philip Hazel University of Cambridge Computing Service,
ph10@??? Cambridge, England. Phone: +44 1223 334714.
