Is anyone out there using the cyrus_sasl authenticator?
Please contact me, with your configuration.
Inspection shows that the testsuite check for it doesn't
function - and the implementation code has not changed
materially since first introduced in exim 4.43 (2004/10).
Specifically, there is no knowledge of the user-secret
(aka. password). I am forced to assume that the
library has some backdoor configuration (the docs
chapter mentions possible need for environment variables
but is not explicit).
The library provides for a callback during server-side auth, giving
the received name and password from the client and permitting the
calling code (the exim authenticator, here) to validate the
combination. The current coding does not request this callback,
and (at least for the PLAIN and ANONYMOUS auth methods) always
claims the authentication failed, as invoked by the testsuite.
Possibly this is because of some missing configuration as
mentioned above. Or possibly it has never worked.
Test hackery shows that with the callback I can do something
reasonable for PLAIN. My concern is that I break this
putative backdoor config mechanism - hence the call for
information.
- --
Cheers,
Jeremy jgh@???
GPG-encrypted mail welcome
