want zombie nnimap groups!

want zombie nnimap groups!

[Kai Großjohann]

I'm using the default value of gnus-subscribe-newsgroup-method,
`gnus-subscribe-zombies'.  New native groups are added to the zombie
list for convenient browsing via `A z'.  But new nnimap groups don't
appear there.  Is this normal, or have I messed up things?  Is there a
way to make Gnus subscribe the new nnimap groups as zombies, too?

I have added nnimap to gnus-secondary-select-methods, if that matters.

kai
-- 
Beware of flying birch trees.

Re: want zombie nnimap groups!

[Simon Josefsson]

Kai.Grossjohann@CS.Uni-Dortmund.DE (Kai Großjohann) writes:

> I'm using the default value of gnus-subscribe-newsgroup-method,
> `gnus-subscribe-zombies'.  New native groups are added to the zombie
> list for convenient browsing via `A z'.  But new nnimap groups don't
> appear there.  Is this normal, or have I messed up things?  Is there a
> way to make Gnus subscribe the new nnimap groups as zombies, too?

nnimap can't tell if a group is "new" or not, since IMAP mailboxes
does not have creation dates attached to them, as nntp have.  Listing
all mailboxes and comparing with a old list wouldn't work either,
mailbox lists can be infinite in IMAP.

Mailboxes that are created by Gnus/nnimap itself might be returned,
but that only solve a small part of the problem so I haven't bothered
doing so.

Non-generic solutions (such as returning all mailboxes available,
which could work for small-size Cyrus IMAPD installations) could
perhaps be added to solve the problem for some people?

Re: want zombie nnimap groups!

[Kai Großjohann]

Simon Josefsson <jas@pdc.kth.se> writes:

> nnimap can't tell if a group is "new" or not, since IMAP mailboxes
> does not have creation dates attached to them, as nntp have.  Listing
> all mailboxes and comparing with a old list wouldn't work either,
> mailbox lists can be infinite in IMAP.

Hm.  If gnus-save-killed-list is non-nil, Gnus would in principle be
able to find that out, wouldn't it?  I know nothing about the guts of
Gnus, though, so maybe something there prevents the functionality from
working...

FWIW, I do use Cyrus, so I'll be happy with something which works
there :-)

kai
-- 
Beware of flying birch trees.

Re: want zombie nnimap groups!

[Lars Magne Ingebrigtsen]

Simon Josefsson <jas@pdc.kth.se> writes:

> nnimap can't tell if a group is "new" or not, since IMAP mailboxes
> does not have creation dates attached to them, as nntp have.  Listing
> all mailboxes and comparing with a old list wouldn't work either,
> mailbox lists can be infinite in IMAP.
> 
> Mailboxes that are created by Gnus/nnimap itself might be returned,
> but that only solve a small part of the problem so I haven't bothered
> doing so.
> 
> Non-generic solutions (such as returning all mailboxes available,
> which could work for small-size Cyrus IMAPD installations) could
> perhaps be added to solve the problem for some people?

The other backends (that have the same problem as IMAP) return the
same list with nn*-request-newgroups as with nn*-request-list; that
is, all the groups, and then Gnus sorts out the newness issue.

If the number of groups is (potentially) extremely large, this is
probably not a good idea for nnimap, but is it likely that the number
of imap mail groups is extremely large?

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

Re: want zombie nnimap groups!

[simon]

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

> The other backends (that have the same problem as IMAP) return the
> same list with nn*-request-newgroups as with nn*-request-list; that
> is, all the groups, and then Gnus sorts out the newness issue.
> 
> If the number of groups is (potentially) extremely large, this is
> probably not a good idea for nnimap, but is it likely that the number
> of imap mail groups is extremely large?

I think so, yes.  UoW's imap server (I'd guess it's the most widely
deployed server) export home directories.  Issuing a command to list
all files in even a quite small home directory easily kill performance
on a server (and the client too, of course).

...and listing the groups isn't enough -- nnimap enter each of theese
groups to find out total number of articles and number of unread
articles.  Ideally, this shouldn't be done but fixing it would require
changing the backend interface or making Gnus handle bogus data.

(And we haven't started considering cyclic paths, or when the server
ends up listing / recursively, which normally contain cyclic symlinks,
or if the imap server export the usenet hierachy etc, if the server
contain lots of public groups that many Exchange and Cyrus server
does, etc...)

Re: want zombie nnimap groups!

[Kai Großjohann]

On 13 Aug 2000, simon@josefsson.org wrote:

> Lars Magne Ingebrigtsen <larsi@gnus.org> writes:
> 
>> If the number of groups is (potentially) extremely large, this is
>> probably not a good idea for nnimap, but is it likely that the
>> number of imap mail groups is extremely large?
> 
> I think so, yes.  UoW's imap server (I'd guess it's the most widely
> deployed server) export home directories.  Issuing a command to list
> all files in even a quite small home directory easily kill
> performance on a server (and the client too, of course).

Well, nnimap does offer the nnimap-list-pattern workaround for this.

kai
-- 
I like BOTH kinds of music.