On Sunday 17 July 2005 03:09, OpenMacNews wrote:> hi all,
>
> "uncle"! i've thrown up my hands with cyrus, and am looking
for a
> 'better alternative' ...
>
> i'm brand new to dovecot (haven't even built it yet ...), and have
> spent the last while reading through the LIST archives here.
>
> apart from the TECHNICAL aspects, which seem _more_ than
> adequate/competitive,
>
> i'm most attracted by a comment abt this LIST/community:
> > awesome technical support with patient developer-dudes on hand
>
> which, in a nutshell, is what i seem to be missing.
>
> so, i'm preparing to "take the plunge" w/ dovecot.
>
> two preliminary questions, if i might:
>
> (1) what sort of *interest* is there here (Timo? others?) in Dovecot
> on OSX? i'm usually able to get most anything up-n-runnig on my OSX
> boxes, but i'd like to join a community where every issue-on-Mac is
> not simply ignored/slammed ... and there might be folks willing to
> work with me on issues
There is a Darwinport for 0.99.14 so that seems like a good sign.
Lots of us, me included, run Dovecot on FreeBSD which is (in this
context) about as similar to OS X as you will find.
> (2) here-we-go-again with the Maildir/Mbox decision :-)
>
> what's the latest thinking/recommendation here wrt mailbox/mdir? my
> specifics are -- i intend to backend Dovecot with Exim ... and will
> run my server 'closed box' ... i.e., no file-system-related User
> accounts. rather, just 'virtual' spaces.
>
> i'm leaning towards Maildir's one-flat-file-per-message format,
with
> an eventual/possible move to storage in PostgreSQL if/when Dovecot
> adds the capability ...
I think Maildir is better. I don't think storing mailboxes in
an SQL database is a good idea, a network filesystem will perform
better with large mailboxes and messages, plus if you can use
Maildirs with lots of existing tools to manipulate the mailboxes.
If you keep it in an SQL database, you will have to write all
your own tools for processing the data. I suppose it depends on
what your goals are.
Maildir and mbox have different trade offs so picking the right
one is about knowing the problem you need to solve.
> thoughts? (yes, i know, y'all must've had the discussion a
g'zillion
> times ... a pointer to a good/relevant thread is great too!)
>
> cheers,
>
> richard
HTH,
--
Dominic Marks