[1830] in athena10
Re: mutt config
daemon@ATHENA.MIT.EDU (Evan Broder)
Tue Mar 24 03:05:48 2009
Message-ID: <49C88608.10404@mit.edu>
Date: Tue, 24 Mar 2009 03:04:40 -0400
From: Evan Broder <broder@MIT.EDU>
MIME-Version: 1.0
To: Greg Price <price@mit.edu>
CC: debathena@mit.edu
In-Reply-To: <20090324063948.GD15690@mit.edu>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
This all sounds mostly good - a few comments inline (with lots of stuff
snipped).
Greg Price wrote:
> This setting of $record is dangerous -- it means that mail you send is
> forgotten, with no local copy. This bit me for my first few days at
> MIT, using sipb-locker mutt. The upstream default is "~/sent", which
> at least means the mail is saved, though not in an optimal place.
> I set it to +INBOX.sent-mail, which is my own idiosyncratic choice.
> Recommendation: leave default, unless there's a justifiable "MIT
> mail" convention; does webmail pick a sent-mail folder?
>
I'm not really sure there is a convention. I don't believe webmail saves
sent messages by default. I also don't think we should be looking to
webmail for any inspiration of any sort.
Looking at the "official Athena mail clients":
- alpine saves mail into the sent-mail IMAP folder.
- Evolution saves mail into a local Sent folder (stored in ~/.evolution)
On the other hand, the Evolution wrapper takes steps to make sure that
~/.evolution isn't world readable. I'd like to not have to go to that
trouble, so I think I'd be in favor of following alpine's lead here.
> Using MIT LDAP
> --------------
> set query_command = "athrun sipb mutt_ldap_query.pl %s"
>
> See http://www.mutt.org/doc/devel/manual.html#query .
>
> In its present form this can run something from the sipb locker, which
> we won't do. The script it invokes is from upstream's contrib, with
> chiefly these configuration lines:
>
> my $ldap_server = "ldap.mit.edu";
> my $search_base = "dc=mit,dc=edu";
> my $LDAPSEARCH="/usr/athena/bin/ldapsearch -Q";
>
> In principle we could depend on lbdb, which packages this script in
> Debian, and the script naturally can be configured from the command
> line or with a config file. But that seems like a lot of added
> complexity for fairly little gain -- this is a feature I've never used.
>
> If someone sees a simpler way so that we're only supplying roughly the
> strings "ldap.mit.edu" and "dc=mit,dc=edu", and not extra software,
> on the other hand, then probably we should do that.
>
> Recommendation: probably drop.
>
Hmm...if this script is using ldapsearch, debathena-ldap-config should
automatically set a host of ldap.mit.edu and a base of dc=mit,dc=edu,
meaning that wouldn't even be providing the strings. It would probably
be sufficient to pull in the package and point at it, which seems much
less annoying.
We don't currently configure LDAP by default for alpine or evolution,
but they both do have the capability built-in, so users could configure
it themselves. My inclination is to add lbdb to -extra-software-nox or
something so that users who want it can enable LDAP with what should be
a single config line.
> Comments welcome. I'll commit a version of this later this week if
> the discussion has settled. There is a .d mechanism, so it only means
> dropping a file in place.
>
All of our releases but Etch and Dapper actually create a /etc/Muttrc.d
directory when they install. You should check to see if Etch's and
Dapper's mutts will respect a Muttrc.d directory if you create it
instead of them. If not...this gets uglier, although it wouldn't be too
hard to transform on Etch/Dapper and just drop the file in everywhere
else. If it was just Dapper, I'd say screw it, but I think we do still
care about Etch...for now, at least.
- Evan
