On Mon, Oct 18, 2010 at 05:44:45PM +0200, Martin Baehr wrote:
> the question is, which behaviour do we want for foresight?
> do we want to add the installLabelPath to a default foresight install?
For normal users, I feel fairly strongly that sticking to the
groups will give them a better experience -- you were trying
to install packages from outside the groups, which is a
reasonable thing for a developer.
> if not then i think we would want "conary rq" to not search the
> installLabelPath either.
>
> the issue here is not whether installLabelPath is in the system-model
> but rather how helpful or confusing it is that conary rq and conary
> update/install do different kind of searches. in my eyes, the point of
> conary rq is to tell me what is available for install, unless you want
> to introduce another verb here: 'conary search' for example.
conary rq has always been non-symmetrical with conary q; it does not
take the system flavor into account, so "conary rq" has never been a
substitute for "conary update --info".
I think it would be extremely counterintuitive to have "repquery"
("rq" is short for "repquery") be querying the local system model
rather than the repository.
I would think that when packagekit has a system model implementation,
having it search the available packages from the system model makes
a great deal of sense.
"conary install --info foo" in the system model is really effectively
"search for foo in all available packages in my system model". The
question is whether we need some different behavior that warrants a
different verb at the command line. Does that make sense?
_______________________________________________
Foresight-devel mailing list
Foresight-devel@lists.rpath.org
http://lists.rpath.org/mailman/listinfo/foresight-devel
