completion for gnus-group-jump-to-group

[Julien Cubizolles]

I don't really know if this is a helm of gnus question, but here goes:
where does helm get its completion list when running
gnus-group-jump-to-group ? I end up with a lot of very old and obsolete
group names dangerously close to some real ones.

Julien.

[Eric Abrahamsen]

This is Gnus question (helm doesn't make it's own list of groups), and
the answer -- in `gnus-group-completing-read' is it uses the variable
`gnus-active-hashtb', which is an obarray holding the active groups. All
those groups ought to be subscribed and in play: if you hit "L" in the
Group buffer, you see them all listed, right?

E

[Julien Cubizolles]

Eric Abrahamsen <er...@ericabrahamsen.net>
writes:

Not all of them. There are some groups in gnus-group-completing-read
that aren't displayed in the list of groups displayed by
gnus-group-list-all-groups, that I can't find in any "active" file and
that can't be found when browsing a giver server in the *Server*
buffer. They are for example groups associated to some old nnimap
methods that I don't use anymore. Is there any way to prune this hash
table ?

Julien.

[Eric Abrahamsen]

The hash table is constructed at load time -- the entries are probably
in your newsrc.eld file, in the `gnus-newsrc-alist' variable. Take a
look in there and see if that's the case, maybe? You can always shut
down Gnus and then (very carefully) manually remove those entries.

E

[Julien Cubizolles]

Eric Abrahamsen <er...@ericabrahamsen.net>
writes:

> The hash table is constructed at load time -- the entries are probably
> in your newsrc.eld file, in the `gnus-newsrc-alist' variable. Take a
> look in there and see if that's the case, maybe? You can always shut
> down Gnus and then (very carefully) manually remove those entries.

Thanks, that was easy. I had to do it manually since I was getting stack
overflow errors with the regexp matching but I could trim it down quite
a lot.

Julien.