Re: [fetchmail-users] fetchmail 6.3.6-rc4 release candidate

[Matthias A. <mat...@gm...>]

Jakob Hirsch schrieb am 2006-11-29:

> But why the superfluous CAPA? I know the server does not support SSL or
> TLS, and I have no "ssl" for it in my fetchmailrc, so why look for it at
> all? It probably does not harm much, but I think we should not bother
> the server unnecessarily, so the extra connect should be avoided. We
> could instead try to login despite of the failed CAPA, and only
> reconnect after this login failed (which will surely happen with some
> servers).

1. You did not ask for SSL, but you did probably not prohibit TLS
either, so fetchmail will - that's its default - look if it has the
opportunity to use TLS.  Using 'sslproto ""' should defeat this CAPA
probe. This goes along with <http://www.fetchmail.info/fetchmail-FAQ.html#K6>

2. ESR got reports that some servers dropped the connection later on
when probed for capabilities, and a reconnect was chosen as a solution.
I don't know many of the details though, if someone knows URLs off-hand,
let me know. I'm a bit loathe to change this part of the behavior at
this time though.

Given that CAPA doesn't cause state change, servers that change behavior
after "reconnect without CAPA" are broken anyways.

3. Does fetchmail re-probe capabilities in daemon mode for the 2nd and
subsequent polls? It shouldn't.

-- 
Matthias Andree