On 26/01/05 12.47, Thomas Roessler wrote: > On 2005-01-26 12:40:48 +0100, Mads Martin Joergensen wrote: [snip] > > ... but the simple smtp client to enable > > mutt to use a relay-server is something closely tied together to > > the client itself IMO. > > We're on Unix, remember? "Do one thing, and do it well." > > There are command line SMTP clients around that you can integrate > with mutt by changing a simple configuration variable, and for any > kind of a complex setup (think queuing), you're better served with a > full-fledged MTA anyway. > > In short, an SMTP client is one of the things that, in my view, > *clearly* don't belong into mutt. Very good point indeed (and I for one agree that mutt should remain smtp-less). If we assume that this policy stands, how could the authentication problem be solved? How about looking at the pgp/gpg support, where external programs need a password. If mutt could be set to prompt for a password, and pass it securely to the $sendmail program, would that not afford the needed flexibility, without putting stuff in mutt that does not belong? /dossen PS: If anything I think that a future mutt could stand to have some of the features seperated into external programs, setup as defaults with a welldefined interface. And perhaps some way to leverage mutts input-features (prompts, lists of selectable options, etc.) for use as arguments for programs invoked from hooks, macros, pipes, etc. With good defaults (either builtin or external) and lots of interfacing options that could make mutt and even better mua. But that is just me being tired and dreaming out loud ;-)
Attachment:
pgpRLg1wGOsdE.pgp
Description: PGP signature