Re: [imap5] Feature set? - was Re: Designing a new replacement protocol for IMAP

[Bron Gondwana <[e-mail]>]

On Thu, Feb 16, 2012, at 09:57 AM, Dave Cridland wrote:
> On Thu Feb 16 09:20:44 2012, Adrien de Croy wrote:
> > SMTP: specification of server, choice of authentication method,  
> > choice of security (SSL vs STARTTLS vs none), username and password.
> > IMAP: specification of server, choice of authentication method,  
> > choice of security (SSL vs STARTTLS vs none), username and password.
> > LDAP: specification of server(s), choice of authentication method,  
> > choice of security (SSL vs STARTTLS vs none), username and password.
> 
> Well, of course, I'd argue that you could use a combination of SRV,  
> common options, discovery, and ACAP to fix all that.
> 
> The problem with Thunderbird isn't that it has all these options,  
> it's that it requires the user to enter them, and fails to do  
> discovery properly - Tony Finch wrote a particularly good blog post  
> on why it's so awful several years ago, and provided solutions, too.
> 
> Interestingly, XMPP has generally gone the discovery route, and the  
> result is that you only enter a jid and a password.
> 
> For Polymer, you enter a username, ACAP server, and password - ACAP  
> doesn't have the SRV option, and maybe I should just add that in -  
> not that anyone but me cares anymore.
> 
> Having written multiprotocol clients, I just don't think they're as  
> hard as people make them out to be.

Having seen support tickets for many years, those of us in a "service
provider" role know how bad the multiple-credential, multiple service
with different uptime issues and intermediate unreliability issues is.

The autoconfiguration space is getting better, many services "just work"
with Thunderbird now, for example.

As for the ACAP server - the world is moving, for the better I think, to
"you enter your email address and password, and it just works".  That's
the minimum short of a single-signon solution.  One thing that uniquely
identifies you and one thing that's secret.  Entering the ACAP server
address as a third datapoint is 50% more overhead, and that's the 50%
that you need to look up somewhere, because you already know your
email address and password.

Which is a vote from me for "yes, SRV option makes sense".

Bron.

-- 
  Bron Gondwana
  [e-mail]