Gnus still doesn't count messages properly

Gnus still doesn't count messages properly

[Jérôme Marant]

Hi,

  I'm using oort gnus with nnmaildir at home.
  It seems that the number of messages within groups
  is not calculated properly: it usually gives a bigger
  number of incoming emails than there are really (sometimes,
  it tells me about new messages, although there aren't).

  It used to raise the issue and Paul Jarc told that nnmaildir
  counts emails correctly but Gnus doesn't.

  Does any gnus hacker know and intend how to fix this?
  Is there any workaround?

  Thanks in advance.

  PS: I'm not an elisp hacker at all but tell me if I can help
  in investigating.  

--
Jérôme Marant <jerome@marant.org>
              <jerome.marant@free.fr>

http://marant.org

Re: Gnus still doesn't count messages properly

[Lars Magne Ingebrigtsen]

Jérôme Marant <jerome.marant@free.fr> writes:

> It seems that the number of messages within groups
> is not calculated properly: it usually gives a bigger
> number of incoming emails than there are really (sometimes,
> it tells me about new messages, although there aren't).

The Gnus/backend protocol doesn't allow for the backend to really say
how many articles are in the group.  And even if it did, it wouldn't
be useful information -- if you have read messages 5-10 in a group,
and the backend says that there are 4 messages in the group, the
highest being number 13, the lowest 3 -- how many unread messages
should Gnus claim that there are in the group?

So the backend would have to return the actual range of articles in
the group during the activation process, which (for most backends) is
costly, and for some is impossible.

-- 
(domestic pets only, the antidote for overdose, milk.)
   larsi@gnus.org * Lars Magne Ingebrigtsen

Re: Gnus still doesn't count messages properly

[Kai Großjohann]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> So the backend would have to return the actual range of articles in
> the group during the activation process, which (for most backends) is
> costly, and for some is impossible.

But for some backends, it's possible and easy, so the backend
interface could be extended.

Isn't there already a way, similar to extra headers in NOV data, to
have the backend return extra information?  Then Gnus would just have
to make use of it.

But I've said before it's easy, and then it turned out to be hard...
-- 
Ambibibentists unite!

Re: Gnus still doesn't count messages properly

[Lars Magne Ingebrigtsen]

kai.grossjohann@uni-duisburg.de (Kai Großjohann) writes:

> Isn't there already a way, similar to extra headers in NOV data, to
> have the backend return extra information?  Then Gnus would just have
> to make use of it.

I think Simon did some work in this area related to nnimap...

-- 
(domestic pets only, the antidote for overdose, milk.)
   larsi@gnus.org * Lars Magne Ingebrigtsen

Re: Gnus still doesn't count messages properly

[Simon Josefsson]

Lars Magne Ingebrigtsen <larsi@gnus.org> writes:

> kai.grossjohann@uni-duisburg.de (Kai Großjohann) writes:
>
>> Isn't there already a way, similar to extra headers in NOV data, to
>> have the backend return extra information?  Then Gnus would just have
>> to make use of it.
>
> I think Simon did some work in this area related to nnimap...

Never integrated though, I think...  Since I keep saying that it
should be simple to fix this unread-count problem for backends that
can provide the necessary information, I will try to put my foot where
my mouth is, or however the saying goes.