gnus-group-get-new-news and topics

gnus-group-get-new-news and topics

[Jason Schroeder]

Should gnus-group-get-new-news and gnus-group-get-new-news-this-group work
with topics?

I think they should.

Additionally, if a topic is hidden, then gnus-group-get-new-news should
ignore the groups underneath that topic.   A topic could collect some state
indicating that it should gnus-group-get-new-news for its containing groups
when it is opened...

Jason

Re: gnus-group-get-new-news and topics

[Lars Magne Ingebrigtsen]

Jason Schroeder <shrode@wsc.com> writes:

> Should gnus-group-get-new-news and gnus-group-get-new-news-this-group work
> with topics?
> 
> I think they should.

Well, don't they?

> Additionally, if a topic is hidden, then gnus-group-get-new-news should
> ignore the groups underneath that topic.   A topic could collect some state
> indicating that it should gnus-group-get-new-news for its containing groups
> when it is opened...

Or perhaps one should be able to `M-g' on topic names to update all
groups in that topic?

-- 
Home is where the cat is.

Re: gnus-group-get-new-news and topics

[Steven L. Baur]

>>>>> "Lars" == Lars Magne Ingebrigtsen <larsi@ifi.uio.no> writes:

    Lars> Jason Schroeder <shrode@wsc.com> writes:
    >> Should gnus-group-get-new-news and
    >> gnus-group-get-new-news-this-group work with topics?
    >> 
    >> I think they should.

    Lars> Well, don't they?

Not really.

    >> Additionally, if a topic is hidden, then
    >> gnus-group-get-new-news should ignore the groups underneath
    >> that topic.  A topic could collect some state indicating that
    >> it should gnus-group-get-new-news for its containing groups
    >> when it is opened...

I'm not sure about this, but it would be consistent with the way
topics pick up state now (a folded topic will appear folded in the
next session, for example).

    Lars> Or perhaps one should be able to `M-g' on topic names to
    Lars> update all groups in that topic?

Yes, exactly.  The ability to do this would make the group visibility
parameter irrelevant for me, since I only need it for doing M-g on
specific groups.

-- 
steve@miranova.com baur

Re: gnus-group-get-new-news and topics

[Lars Magne Ingebrigtsen]

steve@miranova.com (Steven L. Baur) writes:

>     Lars> Or perhaps one should be able to `M-g' on topic names to
>     Lars> update all groups in that topic?
> 
> Yes, exactly.  The ability to do this would make the group visibility
> parameter irrelevant for me, since I only need it for doing M-g on
> specific groups.

One could imagine all (or many) of the group commands working on
topics.  (It's trivial to implement -- the commands would just
process-mark all the groups in the topic and execute the proper group
commands on the process-marked groups.)  

Is this desirable?  If you press `u' on a topic, should all groups in
the topic become unsubscribed?  

Anyways, `M-g' on a topic is obviously useful, but it seems to me that
it should work on all groups in the topic, not just the currently
visible ones, shouldn't it?  Which means that `M-g' on the topic would
do something slightly different that just doing `M-g' repeatedly on
all visible groups in the topic...

-- 
Home is where the cat is.

Re: gnus-group-get-new-news and topics

[Steven L. Baur]

>>>>> "Lars" == Lars Magne Ingebrigtsen <larsi@ifi.uio.no> writes:
    Lars> Is this desirable?  If you press `u' on a topic, should all
    Lars> groups in the topic become unsubscribed?

That wouldn't be used very often.  Other additions could be C-k for
killing, c and C for catchup, and C-c C-x for manual expiration.

    Lars> Anyways, `M-g' on a topic is obviously useful, but it seems
    Lars> to me that it should work on all groups in the topic, not
    Lars> just the currently visible ones, shouldn't it?

I'd prefer this.

    Lars> Which means
    Lars> that `M-g' on the topic would do something slightly
    Lars> different that just doing `M-g' repeatedly on all visible
    Lars> groups in the topic...

You can do that now.  It's somewhat inconvenient, IMHO.

-- 
steve@miranova.com baur

Re: gnus-group-get-new-news and topics

[Scott Blachowicz]

[note that I haven't used sgnus yet, so some comments may be
uninformed...]

>>>>> "Lars" == Lars Magne Ingebrigtsen <larsi@ifi.uio.no>

 Lars> One could imagine all (or many) of the group commands working on
 Lars> topics.  (It's trivial to implement -- the commands would just
 Lars> process-mark all the groups in the topic and execute the proper group
 Lars> commands on the process-marked groups.)  

Except that you wouldn't want the operation to necessarily query you to
confirm the operation on each individual group (like saving/piping a
message in the summary buffer does).

 Lars> Is this desirable?  If you press `u' on a topic, should all groups in
 Lars> the topic become unsubscribed?  

Depends...if you have groups in multiple topics, I'd say not.  I don't
want a `u' on my "editors" topic to unsubscribe me from gnu.emacs.all in
my "GNU" topic.

 Lars> Anyways, `M-g' on a topic is obviously useful, but it seems to me that
 Lars> it should work on all groups in the topic, not just the currently
 Lars> visible ones, shouldn't it?  Which means that `M-g' on the topic would
 Lars> do something slightly different that just doing `M-g' repeatedly on
 Lars> all visible groups in the topic...

Yes, I would think so.  I would view the group buffer as a tree.  Topics
are just ways of adding levels to the tree.  Since the unread article
count in a group/topic controls its visibility, what you SEE under a node
isn't necessarily what EXISTS in that node.  In general, an operation on
any point in the tree should apply to that point and anything that EXISTS
below it.

So, if I `u' a topic, I conceptually reduce a reference counter on each
subgroup by one.  If the count reaches zero, then the group is
unsubscribed.  Hmmm...now what happens if I `C-k' the topic?  For some
reason, I think of that as a visual manipulation (i.e. it doesn't kill the
underlying subgroups - just the tree structure at that point).  Maybe `u'
should work the same way - just change the "level" on the topic to affect
the visualization of the tree structure?  Oh, I don't know...maybe my
mental model needs little work. :-))

Scott Blachowicz    Ph: 206/283-8802x240    StatSci, a div of MathSoft, Inc.
                                            1700 Westlake Ave N #500
scott@statsci.com                           Seattle, WA USA   98109
Scott.Blachowicz@seaslug.org