🚀 go-pugleaf

In article <1993Jan1.154059.22364@klaava.Helsinki.FI> kankkune@klaava.Helsinki.FI (Risto Kankkunen) writes:

   user. You could have a subject menu showing the group name and subject
   line for each article. Or you could have a sub-group selection menu

i like this, though i've imagined just a single character to
disambiguate, then maybe a new header or something for the full
information if the user needs it, to conserve subject buffer space.

   presented to you when you enter the pseudo group.

your latter idea is implemented as gnus-topic.el.  i am not in favor
of it personally, since i'd rather have my fake groups act real, but
you might like to give it a try.

   I see pseudo groups completely orthogonal to virtual groups etc. You can
   have virtual groups, pseudo groups, pseudo groups containing virtual
   groups, pseudo groups containing pseudo groups etc. Pseudo groups are
   just another way to partition your news universe.

i'm afraid i don't know what you mean by pseudo groups yet.  wish i
did.

   An example of one application of pseudo groups: I would have pseudo
   groups called work, programming, sources and hobbies. Work group would
   contain all the groups related to my work, e.g. comp.sys.sun,

that's what i want too.  but maybe you're thinking of the gnus-topic
type idea where you get a new buffer with the constituent newsgroups,
from which list you select a subject buffer, from which list you
select articles to read.  i'd rather have newsreaders have the
capability of having virtual groups display in the subject buffer just
like regular groups do.  (we can be pluralistic about it too!  you can
have your topics; i'll have my virtuals....)

   As you see there would be different kinds of groups in one pseudo group,
   like sources and discussion groups. I think the virtual group idea to

same with virtual groups.... wish i knew what you meant; it sounds
interesting.

   merge all the articles of its component groups doesn't fit into this
   kind of grouping very well. Another thing to note is that a groups can

???  you can always kill, and even kill based on the original
newsgroup name.  if you mean by component groups things like
comp.sources as a component of comp, that has nothing to do with
virtual groups.... in fact, virtual groups are a way to subvert the
hierarchy.  you choose the groups and articles (via kill, perhaps)
that go into one.

   be in many pseudo groups (which is the case with virtual groups, too, I
   understand).

right.  you can put, say comp.sources.whatever in both your work and
your sources virtual groups if you like.  and you can have separate
kill files for them too, so in your work group you get widget articles
while in your source groups you get doohickey articles.  note that
this essentially implements complete virtual newsgroups while fitting
cleanly into the kill file model.  and it's easy to upgrade from
simple virtual newsgroups (the set-theoretic idea).

   Do you think my definition of pseudo groups matches any of the terms
   mentioned in previous articles? Do you think pseudo groups would be of
   any use? Or do you prefer to browse you hundreds of plain, virtual or
   other groups as a simple linear list?

well, i think it's a good idea with lots of use and i encourage you to
pursue it if you think you'd like to, but i wouldn't like it or use it
myself.  i like my linear list :-).  (i already have groups ->
subjects -> articles; it's too confusing to go up and down a menu
hierarchy deeper than that all the time for news.)

   I don't think there is One True Universal Concept of accessing news, so
   I would like to see different grouping methods to proliferate. Each one

yeah, yeah, yeah!  let a thousand interfaces bloom!

   could then choose the method they like. Even if the gnus-virtual-groups
   were a subset of vgrs, the former may be preferred, if it is easier to
   control what groups go to what virtual groups with it, for example.

right; not only that, but virtual groups are a good stepping stone to
anything else we'd like to implement, including your topic idea and
the previous poster's ultra flexible algol like language idea (n.b. he
might want to look at newsclip).  and imho, with judicious integration
with kill files and some speed hacks, virtual newsgroups are enough to
keep me very, very happy.

do i have any takers yet?  how about the gentleman who's editing the
data structures of gnus to make it faster?  he's got to have some
ideas on implementing virtual newsgroups, and also the perfect
opportunity if he's changing things anyway.

comments?
-- 
i'm open to comparing research career notes with other cogsci/cogneuro people.