On Wednesday, 14 April 2010 at 12:32, Will Fiveash wrote: > On Wed, Apr 14, 2010 at 10:03:17AM -0700, Brendan Cully wrote: > > > IMO, changing the defaults is in line with the principle of least > > surprise, now that networked mail has become so much more > > important. So turning this around, what's the compelling reason to > > keep the defaults? > > Stability of the configure interfaces. In addition, will changing the > defaults increase the likelihood that either configure or the build will > fail because this introduces new default dependencies on code (libs) > that are external to the mutt source directory? We should avoid that. If an option hasn't been explicitly requested, mutt should only activate the feature if the support libraries are there. Otherwise it should print a message saying that feature x has been disabled because library y is missing, and move on. While ./configure interface stability is nice to have, I'd say unsurprising defaults for a majority of users takes precedence.
Attachment:
pgpyVDVPc8i1p.pgp
Description: PGP signature