Re: agent/fetch only articles from certain levels

[Richard Riley <rileyrg@googlemail.com>]

Eric S Fraga <e.fraga@ucl.ac.uk> writes:

> asjo@koldfront.dk (Adam Sjøgren) writes:
>
>> On Fri, 25 Mar 2011 21:05:42 +0000, Eric wrote:
>>
>>> Adam, thanks for the pointer to the relevant place in the manual.  My
>>> guess was basically completely wrong.  Oh well...
>>
>>> However, I am not sure the documentation really explains anything at
>>> all.
>>
>> How did you then get to the understanding that you were wrong, then?
>
> Ummmm, I guess the fact that there are three types of groups whereas I
> was thinking of only two types...
>
>>> It's the kind of documentation that makes sense if you already know
>>> what it is telling you and otherwise just sounds like a circular
>>> definition... and, unfortunately, there are quite a few sections like
>>> this in the gnus manual.
>>
>> I am sure suggestions for improvements are very welcome!
>
> I would be more than happy to not only make suggestions but also add
> text etc.  However, the problem is I don't really understand the text
> well enough to even begin adding or changing text!  Vicious cycle here
> (for me).

Add me to that list ;)

>
>> What I got from reading it is that groups from servers in select-method
>> and secondary-select-methods are called native and other groups are
>> called foreign. Seems quite straightforward?
>
> Okay, I didn't get that from the text but I guess that makes sense.  It
> still doesn't really tell me how or why the distinction is made,
> especially between secondary and foreign?  How/where are the latter
> actually defined?
>
> Anyway, in my case, this distinction does not appear to be anything I
> need to worry about.  And I think the OP has solved his problem, in the
> sense of identifying the boundary conditions for gnus for his use case.
>

Just to re-iterate : I feel it makes sense to activate ALL subscribed
groups which are "agentised" since there is no network access
required. Remembering the "C u 5 g" type thing is inconvenient AND
initiates re-reads on non agent-ised groups (eg my imap groups since
agent doesnt work properly with multiple imap groups) since all non
agentised groups are considered "plugged".

Agent/Plugged/Unplugged is incredibly useful but also incredibly
complicated - even the notion of "foreign groups" means you need to know
the depths of Gnus to even begin to understand the activation
triggers. But I do feel the suggestion above makes for a more common
sense default.