[maemo-developers] Modest/TinyMail problems (continue from the blog comments)

[From: Philip Van Hoof]

On Wed, 2008-06-25 at 11:55 +0200, Luca Olivetti wrote:

> 
> I'm probably too dumb to find those options ;-) (btw, now polymer just 
> crashes)

No you are right, it's supported on your IMAP server. Cyrus is an
excellent choice too, by the way. That's a free software imapd that gets
most right. Which is nice, and exceptional. Another really good one is
Dovecot, by the way. 

There is one, that is regretfully being used far to often by people as a
'personal IMAP server', that gets quite a lot of IMAP core spec things
just wrong. I'll leave it as an exercise for the reading to figure out
which one that is.

That's extremely frustrating for E-mail client developers, of course.

It's "precisely" their mistakes at implementing the specification
correctly that makes developing a good E-mail client that consumes IMAP
unimaginably hard.

It's the fact that this is hard, that makes it more easy for a E-mail
services vendor to sell and develop a new E-mail server that has a
closed protocol, and glue a specialized E-mail client to it.

IMAP server developers have a responsibility in this, in my opinion. If
they want IMAP to succeed, they should NOT misinterpret the RFC.

 
> > Experimentally the client has a lot of value, since it shows how modern
> > IMAP usage would outperform any other offering (Blackberry or Windows
> > Mobile E-mail clients with Exchange) several times in a row.
> 
> I *do* hope he succeed: it would make me very happy if non-standard 
> proprietary protocols just disappeared (though many people seem happy 
> with their crap^H^H^H^Hblackberries and don't mind giving their 
> passwords to a 3rd party and let their mail pass through rim servers).

I'm not against a bit of competition for IMAP. It's that competition
that pushed IMAP developers to start thinking about enhancing the
protocol under the Lemonade umbrella.

Perhaps it would have happened without the competition too. I do think,
however, that competition inspires the passionate mind.

So I'm fine with the Blackberries and the Windows Mobile E-mail clients
and the Exchanges. We'll just compete. It's not that hard, they don't do
that many surprisingly difficult things. In contrary, actually.

Almost zero innovation comes from those factories. Doesn't sound to me
like a lot of passionate people work over there ... must be a really
boring job and it must suck realizing that you are actually blocking
innovation instead of fostering it.

Poor guys ...

> >> I'm currently tunneling imap through ssh with compression enabled.
> > 
> > 
> > That's a good temporary solution indeed. Note that your SSL layer might
> > already do compression, in which case the SSH tunnel will just add
> > latency to your connection.
> 
> Yes, my goal was to avoid ssl altogether (I can do that with 
> thunderbird) and at the same time not to expose the imap server to the 
> internet at large (after all it's my own personal server).

Not everybody has the same possibilities as you have. So most E-mail
clients default to encrypting the connection if possible.

>From a security perspective, that makes a lot of sense. How many people
care about that [x] Use STARTLS option ? Too few. Everybody should,
nobody is. So E-mail clients that just use it if they see it is
available, are right. E-mail clients that don't and yet make it
surprisingly difficult for the user to enable it, are wrong.

> Unfortunately it seems that claws cannot use cram-md5/digest-md5 and 
> it's forcing a starttls. Polymer can do cram-md5/digest-md5 but it's 
> forcing a starttls anyway.

Sounds like right behaviour.

The risk of exposing one person's private details (like passwords and
E-mails) on the naked Internet, outperforms the risk of serving a geek
who has made his own SSH tunnel at a very small extra latency hit.


-- 
Philip Van Hoof, freelance software developer
home: me at pvanhoof dot be 
gnome: pvanhoof at gnome dot org 
http://pvanhoof.be/blog
http://codeminded.be