Re: Oort, agent - fetching read articles

[Harry Putnam <reader@newsguy.com>]

Simon Josefsson <jas@extundo.com> writes:

> Paul Stodghill <stodghil@cs.cornell.edu> writes:
> 
> > I've noticed that the agent in Oort gnus does not fetch all of the
> > articles that I expect it to, and it seems to be that the articles that
> > it doesn't fetch have been read before I try to fetch them.
> >
> > The newsgroups in question are in a category whose predicate is "true".
> > I am not explicitly marking articles for downloading. I'm just doing a
> > "J s" in the *Group* buffer.
> >
> > Is the agent supposed to fetch articles that have already been read? Is
> > not, there some way to get it to do so?
> 
> No, the behaviour is hard coded.  I think I posted a patch here a
> while ago that introduced a new agent predicate `read'.  I'm not sure
> if it still applies (or works) correctly, maybe you could try.

Not being code literate, I haven't really tried to devine out what the
code really does, but it seems that marking such messages unread
should then have the desired affect, but that doesn't always work
either.  At the very least one would expect that using the explict %
mark should have the desired affect but that doesn't always work
either.

It would be nice if at least the % mark forced the agent to download a
message, regardless of read/unread or anything else, even overwriting
it if already on disk if need be.  Can that be managed?

Far as I can see, the only really sure fire way to get a message
downloaded if there is a question, is to go `plugged' and mark it for
cache, and then of course it isn't under the agents care.