[patch] doc typo

[patch] doc typo

[Jens-Ulrik.Petersen]

Just a little thing.  Should I make a ChangeLog entry for something
this small?  (This is relative to pgnus-0.96.)

Jens
who's trying to get nnir to work over nnimap.


--- gnus-group.el~	Wed Oct 13 10:01:42 1999
+++ gnus-group.el	Wed Oct 13 10:01:42 1999
@@ -1600,7 +1600,7 @@
 If REQUEST-ONLY, don't actually read the group; just request it.
 If SELECT-ARTICLES, only select those articles.
 
-Return the name of the group is selection was successful."
+Return the name of the group if selection was successful."
   ;; Transform the select method into a unique server.
   (when (stringp method)
     (setq method (gnus-server-to-method method)))

Re: [patch] doc typo

[Simon Josefsson]

Jens-Ulrik.Petersen@nokia.com writes:

> who's trying to get nnir to work over nnimap.

Is there some problem?  I think the latest nnir should work right
away.

Re: nnir UI

[Jens-Ulrik Petersen]

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

> Jens-Ulrik.Petersen@nokia.com writes:
> 
> > who's trying to get nnir to work over nnimap.
> 
> Is there some problem?  I think the latest nnir should work right
> away.

No, no, probably I'm just being stupid.  

Well actually yes, I get the following error when I try to use nnir
(v1.45) with nnimap-0.130.

  error("Must specify groups for IMAP searching.")
  nnir-run-imap(((query . "haskell")) nil)
  nnir-run-query("((query . \"haskell\"))")
  nnir-request-group("((query . \"haskell\"))" "-ephemeral" nil)
  gnus-request-group("nnir:((query . \"haskell\"))")
  gnus-group-read-ephemeral-group("((query . \"haskell\"))" (nnir "") t (#<buffer "*Group*"> . reply-yank) nil)
  gnus-group-make-nnir-group(nil "haskell")
  call-interactively(gnus-group-make-nnir-group)

To be find the UI/setup of nnir a bit unfriendly.  Why not just use
the backend on the current line for the search by default?  Why do I
need to change a variable to search a different backend?

Ok, I probably missing something.  Can something enlighten me?

Cheers, Jens

ps I use the following bindings:

      nnir-mail-backend '(nnimap "")
      nnir-search-engine 'imap

Re: nnir UI

[Simon Josefsson]

Jens-Ulrik Petersen <jens-ulrik.petersen@nokia.com> writes:

> Well actually yes, I get the following error when I try to use nnir
> (v1.45) with nnimap-0.130.
> 
>   error("Must specify groups for IMAP searching.")

With IMAP groups you need to process mark (`#') groups you want to
search -- searching all IMAP groups is too slow.

Perhaps the error message could be more helpful.

> To be find the UI/setup of nnir a bit unfriendly.  Why not just use
> the backend on the current line for the search by default?  Why do I
> need to change a variable to search a different backend?

I remember adding code so one wouldn't need to configure nnir at all
for nnimap groups -- if you process mark nnimap groups they are
searched with the imap search-function.

Perhaps it could search the backend on current line -- like most gnus
functions do when there are no process marked groups -- what do you
think Kai?

Re: nnir UI

[Kai Großjohann]

Jens-Ulrik Petersen <jens-ulrik.petersen@nokia.com> writes:

> No, no, probably I'm just being stupid.  
> 
> Well actually yes, I get the following error when I try to use nnir
> (v1.45) with nnimap-0.130.
> 
>   error("Must specify groups for IMAP searching.")

This means that you should use `#' to process-mark a couple of groups
to search in.  This is required because searching in all groups is way
too slow with most IMAP servers.

> To be find the UI/setup of nnir a bit unfriendly.  Why not just use
> the backend on the current line for the search by default?  Why do I
> need to change a variable to search a different backend?

Simon has sent me a patch for choosing the right backend
automatically.  Hm.  Maybe I forgot to include it in nnir.el -- sorry,
Simon.

There is a historical reason for nnir.el not choosing the right
backend automatically.  I designed it to work with search engines
(freeWAIS-sf and Glimpse, at first), and for them you have to create
an index over your mail files first.  Obviously, the index only covers
a certain backend, so I required users to choose the backend manually.

I did come across the problem that this method does not (easily) allow
people to use two different indexes, possibly even over two different
backends.  Hm.  I thought that most people have stored all their mail
in one backend, anyway, so such a feature would not be necessary.  And
this choice was before there was the feature of process-marking some
groups to search in.  Back then, I was using nnml exclusively (with a
freeWAIS-sf index).  Now, I use nnimap and I find that the need to
process-mark groups to search in is rather inconvenient.  But with
IMAP, searching should be implemented on the server side, it seems.  I
intend to do some work on Cyrus for better integration with a search
engine, but I'm not promising anything, right now!  And the query
language specified in the IMAP protocol is not what is desirable from
an IR point of view.  Oh, well...  (I guess that Lotus Notes already
provides nice searching capabilities -- any Notes users out there?
Didn't the Notes people also violate the IMAP standard in order to
provide for better searching?)

I do intend to get back to work on nnir some time in the future, so
don't hesitate to contact me about new ideas, and I'll try to file
them away for later looking at them, when I have time.

kai
-- 
Life is hard and then you die.

Re: nnir UI

[Kai Großjohann]

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

> Perhaps it could search the backend on current line -- like most gnus
> functions do when there are no process marked groups -- what do you
> think Kai?

I'm not sure.  Probably something needs to be done, though.  Hm.

kai
-- 
Life is hard and then you die.