Re: Re: [evolution-patches] Possible fix for #322016

[Jeffrey Stedfast <fejj novell com>]

On Tue, 2005-12-06 at 16:57 +0000, karllinuxtest relton ntlworld com
wrote:
[snip]
> > 
> 
> Right - using Thunderbird at work as an imap client gives
> the new mail 'beep' when I transfer an unread mail from
> one folder back to the Inbox: obviously a false notification for the very reasons you express.
> 
> 
> So the best we can do is limit the cases of false notifications. We can clearly do alot better than the current situation in Evolution.
> 
> My patch suggestion, once suitably tidied, should limit
> false notifications to when a message is moved to a folder
> that (then) contains unread messages.
> 
> I wonder if we could easily go further, by
> a) seeing if the messages just added to the folder all have
> unread status

unfortunately this could require a full scan of the folder in question
in order to ascertain that information. Currently we simply request
STATUS (MESSAGES UNSEEN RECENT) or some such and so there's no way to
find out if the unseen/recent counts overlap at all because all we have
are counts (same problem we have in the code).

Querying for a full list of message flags is just not viable.

If the folder in question is the folder in the SELECTED state in the
IMAP provider (not necessarily the same as the selected folder in the
UI, mind you), then querying the server for a full list of flags is not
necessarily needed (the flags in the CamelFolderSummary for said folder
will contain the info we need I think, unless the \Recent message(s)
haven't yet had a chance to have their info's fetched... not sure), but
sadly this does not help us with the generic API.

> b) limiting the notifications to only be issued if the
> folder affected is an appropriate folder for brand new
> mail (i.e. an 'Inbox' type folder).

that's how we originally did things but people wanted notification if
new mail arrived in any folder.

-- 
Jeffrey Stedfast
Evolution Hacker - Novell, Inc.
fejj ximian com  - www.novell.com