Re: another silly question
* On 2005.05.05, in <20050505112905.GC8472@xxxxxxxxxxxxx>,
* "Vincent Lefevre" <vincent@xxxxxxxxxx> wrote:
>
> This is not a problem, as mailboxes created by Mutt (e.g. for postpone
> or user archives) will normally be read only by Mutt. Even if maildir
In my experience that's not true. Thousands of my users use at least
two mail applications. We've counted. We've made charts, tables, and
graphs. Of those mailers, only Mutt understands Maildir. Virtually
everything understands mbox.
When people leave the University, many take their e-mail with them. It
needs to be in a portable format, unless they know what they're doing.
Requiring a conversion to a non-portable format to be a conscious act
saves a lot of trouble. Most users are not extremely knowledgeable of
details like their mail storage format.
In the world of SourceForge and FreshMeat and OtherSuchThings there
are dozens of converters and programs of various other categories
that understand either. But that doesn't reflect the world that the
typical uninformed user lives in. I think Maildir is great for some
purposees, but when you stop limiting yourself to open source users and
*ix nerds, it drifts toward nichedom. That becomes a large support
burden. Maildir users, in general, already know how to handle format
conversion; or they're already motivated to learn.
Maildir provides better resilience under NFS, but how many people really
encounter NFS problems with their mail? It's a problem, but I doubt
that most users of UNIX mail clients use NFS, and probably not even most
people who are using mbox over NFS have issues. (And until recently, we
used NFS here.) It doesn't alone make a persuasive argument to change
something that potentially affects many people, especially when the
alternative is a one-line config directive.
You mentioned that Maildir is often recommended, and mbox is not. I've
seen mbox recommended, but besides that you're choosing both sides of
this issue for your argument. Maildir is often recommended when people
are having specific problems with which mbox is a known weakness --
which is exactly parallel with when people recommend mbox over maildir.
Maildir is also often recommended when the speaker doesn't really know
what the problem is, but "Maildir works for me", so maybe it'll work
around the other problem too. There's an element of underdog advocacy,
and everyone loves to roll over the established dinosaur. I'd wager at
least half of all recommendations of postfix or qmail over sendmail fall
into the same category, but that doesn't mean sendmail is an unsuitable
MTA. I admit that this is a bit of an unfair comparison, since sendmail
and UNIXv7 (mbox) are so intertwined. But you can't really count these
recommendations against mbox.
What's actually *wrong* with mbox as a default, other than that it's not
your preference?
--
-D. dgc@xxxxxxxxxxxx NSIT University of Chicago