Re: How to stop Agent from fetching headers of uninteresting Groups?

[Eric Abrahamsen <eric@ericabrahamsen.net>]

nljlistbox2@gmail.com (N. Jackson) writes:

> I use the Agent for reading news offline (because I don't have an
> Internet connection for many hours of the day). I only read a
> handful of news groups offline, but the Agent downloads the
> headers for every single group on my agentised servers. This takes
> about fifteen or twenty minutes every time I do `J s' because I
> have very many groups on those servers (most of which I only read
> about once a year). Downloading the actual news for the groups I'm
> interested in reading offline only takes about a minute, so it
> seems insane to download all these headers that I don't want.
>
> How can I tell the Agent to not download even the headers for a
> group?
>
> Currently I have one Agent category for each of my agentised
> groups. That is the "default" category and it is set to 0 (by the
> Agent when it first set itself up). For the groups I want to read
> offline I have group parameters that tell the Agent to retrieve
> all messages (I have a `true' predicate).
>
> It seems I need an Agent predicate that is more strongly negative
> than `false', something like `not-even-headers'.
>
> I suppose I might be asking the wrong question, since the Agent
> predicates are to tell the Agent which articles to get, it must
> have already retrieved the unwanted headers before it even starts
> applying the predicates. So I need a way of telling the Agent at a
> higher level not to even get the headers. But I don't see anything
> useful in the documentation.
>
> What am I missing?

I don't think it's possible to do this presently -- looking at
`gnus-agent-fetch-group-1', the headers are indeed fetched before the
predicate is examined. In principle, however, there's no reason why this
wouldn't be possible: we could retrieve the predicate higher up in the
function, and ignore the group altogether (unless some articles had been
marked specifically for downloading).

I suggest using `gnus-bug' to report this and describe the situation;
hopefully someone will get to it!

Eric