Re: New mailbox interface and gmime messages

[Emmanuel <e allaud wanadoo fr>]

Le 2003.02.26 19:49, Pawel Salek a écrit :
> 
> On 2003.02.26 21:34 Magick wrote:
>> Hi all,
>> 
>> I've been slowly working on a new mailbox interface, using gmime as 
>> a message-container.
>> Currently it is still using libmutt, but it is pretty easy to switch 
>> it for something else.
>> 
>> The patch is at:
>> http://demands.xs4all.nl/~magick/libmailbox-4.patch.gz or
>> http://demands.xs4all.nl/~magick/libmailbox-4.patch
>> 
>> I would really appreciate some/any feedback...
> 
> I spotted one mistake: UIDVALIDITY in case of IMAP determines "age" 
> of UIDs and whether, eg, cached messages from previous session can be 
> reused.
> 
> I am not yet sure how the changes you suggest fit in the whole 
> picture. We plan to replace libmutt's imap code (and possibly entire 
> libmutt, if we have enough work force) with our own code that is more 
> bandwidth friendly (and if we manage to replace libmutt by gmime: 
> more flexible).
> 
> I started coding the imap library but it is not ready for release 
> yet. Current code is available at 
> http://balsa.gnome.org/libimap-1.0.0.tar.gz (the number has nothing 
> to do with the the completness of this code) under GPL. There is one 
> or two half-working examples in test/ directory. In particular, 
> browse-tui demonstrates primitive access to an imap server, listing 
> folders, browsing the mailbox index in a bandwidth-friendly way. The 
> crucial part is that the code fetches only headers for the visible 
> messages: try using another approach on a message with several 
> thousands messages.
> 
> This is partially related to another change we would like to 
> implement soon: a GTkTreeModel interface for LibBalsaMailbox. 
> Currently, introducing the changes described above would not be 
> enough because displaying messages via GtkTreeStore would fetch all 
> headers anyway. So, these two changes go really together - but can be 
> implemented separately. I have a preleminary patch for this interface 
> - sorting and threading parts are missing. Interested people can 
> download the patch at 
> http://balsa.gnome.org/balsaGtkTreeModel.diff.bz2 - be warned: it may 
> even not apply cleanly to current cvs.

Hmm, what you guys do you think about opening the new branch soon? This 
could help us to try out the new features instead of having patches 
floating around.
Note that I think we could do that if everyone agrees that HEAD is ok 
in view of stability.
Bye
Manu