Re: Balsa GUI stalls during mail check

[Brian Stafford <brian stafford uklinux net>]

On Tue, 12 February 19:55 Magick wrote:
> 
> On Tue, 12 Feb 2002 10:03:03 Brian Stafford wrote:
>> On Mon, 11 February 21:14 Pawel Salek wrote:
>>> On 2002.02.11 20:40 Bart Visscher wrote:

>>> Well, it will be easy provided you remember about proper structure. I 
>>> actually think that doing converse might be easier, i.e implementing imap 
>>> first. But it is just an impression.
> You mean implementing a imap lib first and then change Balsa over to that 
> one? Getting Balsa to use a different mailbox implementation is quite hard, 
> it assumes a lot of things that libmutt handles.

IMAP is a protocol for manipulating mailboxes and messages (unlike POP which 
just fetches messages from a respository).  It defines a model for mailboxes 
and messages and how to manipulate them.  The protocol is really just an 
implementation of and secondary to that model.  If a mailbox API were to be 
designed that did not closely match the IMAP mailbox concept, it would be very 
difficult to integrate later.

>> My own feeling is that unless a new IMAP client offered significant 
>> enhancement over the one in libmutt, implementing one is probably wasted 
>> effort.  Unlike SMTP and POP-3 which are orthogonal to MIME and can stand 
>> alone from it, IMAP is intimately bound to MIME.  I suspect that IMAP could 
>> not be added to a mailbox library that did not understand MIME in some way.
> Well take a look at http://www.geocities.com/haleykd/api/index.html
> Kenneth Haley has been trying to define a mailbox API that should work with 
> IMAP.

Have done.  It largely seems to stop at the level of the RFC 2822 message and 
sketches over the access to the RFC 2045-9 structure of a message.  
Alternatively, it defers to Gmime.  Either way, I feel that more work needs to 
be done on the message access model befoire it is ready for IMAP.  I also feel 
that any message access methodology should operate at the parsed hierarchical 
MIME structure level and not the message level.  An important point - the API 
should be able to access and manipulate the message structure and parts 
independently.

> I'm working on changing Balsa to use this new interface, and even for the 
> local access it is a major rewrite of the mailbox handling.

That doesn't really surprise me that much.

Brian Stafford