headers
David Z Maze | 1 Jun 2005 18:36
X-Face
Face
Picon
Picon
Favicon

Article counts under nnimap

I'm using nnimap with total-expiry on Gnus 5.10.6 on GNU Emacs 21.2.1
(which is different from what I'm posting this from).  Several of my
groups have wrong article counts, apparently because Gnus believes
it's read articles that no longer exist.  For example, one of my
groups believes that it has 1939 articles in it, and if I run 'G E' on
the group the property list includes that Gnus has seen a sparse set
of articles (seen (6081 . 6100) (6102 . 6107) ... 8019 (8021 . 8022)).
But if I log into the IMAP server directly, a UID SEARCH ALL command
lists articles 7728 through 8022, including things like 8020 that
aren't in that seen list.  (It looks like 8019 and 8020 are
duplicates, and I have both gnus-summary-ignore-duplicates and
gnus-suppress-duplicates set to t.)

This all feels a little weird to me.  If nothing else, I'd like to
have the user-visible article count correct, but it also feels like
Gnus has a somewhat wrong feeling of what's going on on the IMAP
server?  Any hints?

Thanks,

  --dzm
Permalink | Reply | 2 comments |
headers
David Z Maze | 1 Jun 2005 21:41
X-Face
Face
Picon
Picon
Favicon

Duplicate suppression, other processing at split time

If I wind up with the same message delivered to me twice, can I cause
the second message to get dropped at split time?  (In particular using
nnimap-split-fancy, but I could probably adapt other techniques people
have for doing this.)  Is other processing possible?

The particular problem I have is people sending messages without
Message-Id to multiple lists that I'm on.  Then Gnus sorts each of the
copies into each of the lists in the To: line.  But, since there's no
Message-Id, duplicate suppression never trips, and I wind up seeing
n^2 copies of the message; if n is 3 or 4 this gets really annoying.
Another part of our local mail pipeline inserts a string that has the
right properties to be the Message-Id, so first I'd like to copy that
header value to Message-Id if there isn't already one, then do dup
suppression by Message-Id, *then* split.

(Procmail scares me -- I understand what Gnus does, and
nn*-split-fancy is kind of elegant in its own way -- but I think it is
an option, if this is something easy to do there.)

Thanks,

  --dzm
Permalink | Reply | 3 comments |
headers
David Z Maze | 1 Jun 2005 22:07
X-Face
Face
Picon
Picon
Favicon

Re: Duplicate suppression, other processing at split time

David Z Maze <dmaze <at> mit.edu> writes:

> If I wind up with the same message delivered to me twice, can I cause
> the second message to get dropped at split time?  (In particular using
> nnimap-split-fancy, but I could probably adapt other techniques people
> have for doing this.)  Is other processing possible?

Having Read The Fine Manual a little more, it looks like this is very
straightforward to do for nnmail, but not necessarily so
straightforward for nnimap.  Can I do info://gnus/Washing+Mail as
nnimap splitting happens?

  --dzm
Permalink | Reply | 1 comment |
headers
Marc | 2 Jun 2005 15:11

Re: Article counts under nnimap

David Z Maze <dmaze <at> mit.edu> writes:

> This all feels a little weird to me.  If nothing else, I'd like to
> have the user-visible article count correct, but it also feels like
> Gnus has a somewhat wrong feeling of what's going on on the IMAP
> server?  Any hints?

Hi,

I have no hints and I'm no gnus/emacs hacker but I can say that nnimap
does have a problem with article count. I'm splitting mail from INBOX
into several folders. After splitting is done, INBOX still has as count
number the total of incoming mails (seems that it also counts mails that
have been moved in another folder).

Sometimes it diplays a count != 0 but when you enter the group, it sees
that there is no new mail and does nothing...

Hope this is not too OT ;)

Cheers,
Marc
Permalink | Reply | 0 comments |
headers
Ted Zlatanov | 2 Jun 2005 19:46
X-Face
Favicon
Gravatar

Re: Duplicate suppression, other processing at split time

On Wed, 01 Jun 2005, dmaze <at> mit.edu wrote:

> David Z Maze <dmaze <at> mit.edu> writes:
> 
>> If I wind up with the same message delivered to me twice, can I cause
>> the second message to get dropped at split time?  (In particular using
>> nnimap-split-fancy, but I could probably adapt other techniques people
>> have for doing this.)  Is other processing possible?
> 
> Having Read The Fine Manual a little more, it looks like this is very
> straightforward to do for nnmail, but not necessarily so
> straightforward for nnimap.  Can I do info://gnus/Washing+Mail as
> nnimap splitting happens?

I don't think that will work.  You could try, however, to add
duplicate suppression to the gnus-registry and use that; I am thinking
of something like this in the fancy split:

(: gnus-registry-split-duplicates "duplicates") ; or 'delete

The problem is what to do when the message has no message-ID of its
own, as you mention.  It should be pretty easy to generate a unique
message-ID for the registry based on unique information such as the
subject and sender, while preventing collisions from multiple messages
with the same subject and sender.  Perhaps the time or the message's
backend information could be used.

Ted

----== Posted via Newsfeeds.Com - Unlimited-Uncensored-Secure Usenet News==----
(Continue reading)

Permalink | Reply | 0 comments |
headers
Sébastien Kirche | 3 Jun 2005 18:08
X-Face

get address of rss group

Hi,

maybe it is the  dumb question of the day, but how  could I get back the
url of an existing rss group ?

I tried several things like editing (G e and G E), editing parameters (G
p) or customizing (G c) but I only can get the description.

I looked also in the Gnus manual but it gives not much information about
rss (info "(Gnus)RSS") ...

Any clue ?

--

-- 
Sébastien Kirche
Permalink | Reply | 1 comment |
headers
Johan Bockgård | 3 Jun 2005 20:02
Picon
Picon

Re: get address of rss group

Sébastien Kirche <sebastien.kirche.no <at> spam.free.fr.invalid> writes:

> how could I get back the url of an existing rss group ?

C-h v nnrss-group-alist

--

-- 
Johan Bockgård
Permalink | Reply | 0 comments |
headers
bigfaceworm | 4 Jun 2005 00:43
Picon
Gravatar

imap articles come in wrapped

All,

I'm trying to track down if this is a Gnus problem, or an imap setting.
Some email I get (plain text, not HTML) has the lines already wrapped.

This is true whether I get the mail directly from the server using an
nnimap group, or when i use 'mail-sources and list 'imap as one of the
sources.

In Outlook, the email contains a line where all of this is on one line:

  element /vobs/myvob/src/Module/File.cpp
/main/OurBranch/bigfaceworm_development_branch_05ww21.5/3

and in Gnus, the line is always wrapped.  I do not have the auto-wrap
turned on, and the body is wrapped even when I look at the article
using    C-u g   or when I look at the nnml file (using the
mail-sources way to get the article).

Any help in how to track down how this wrapping occurs?

thanks,

BFW
Permalink | Reply | 0 comments |
headers
David Z Maze | 6 Jun 2005 21:54
Picon
Picon
Favicon

Re: Duplicate suppression, other processing at split time

David Z Maze <dmaze <at> mit.edu> writes:

> If I wind up with the same message delivered to me twice, can I cause
> the second message to get dropped at split time?
>
> (Procmail scares me -- I understand what Gnus does, and
> nn*-split-fancy is kind of elegant in its own way -- but I think it is
> an option, if this is something easy to do there.)

It turned out that dup suppression is something explicitly mentioned
in procmailex(5), so I'm doing it (along with adding a Message-Id if
need be) in procmail.

  --dzm
Permalink | Reply | 0 comments |
headers
David Z Maze | 6 Jun 2005 22:00
X-Face
Face
Picon
Picon
Favicon

Re: Article counts under nnimap

David Z Maze <dmaze <at> mit.edu> writes:

> I'm using nnimap with total-expiry on Gnus 5.10.6 on GNU Emacs 21.2.1
> (which is different from what I'm posting this from).  Several of my
> groups have wrong article counts, apparently because Gnus believes
> it's read articles that no longer exist.

The one weirdness I found was that some headers corresponding to
expired articles were living in
~/News/agent/nnimap/my.server/group/name/.overview and .agentview.
Those roughly corresponded to the lower bound on what Gnus was
counting as extant articles.  I ran gnus-agent-expire and it (looks
like it) got better.

Is this going to be an ongoing problem?  Why were there exactly two
articles in agent-space?  Why do I want the agent turned on in this
case?  The manual suggests that there's some interesting overlap
between nnimap and the agent -- there's that mysterious last paragraph
of info://gnus/Agent+as+Cache, for example -- but if I'm never
offline, and the only thing I read is nnimap (well, with a little
nnrss on the side) is it that useful?

  --dzm
Permalink | Reply | 0 comments |

Gmane