On Mon, Oct 18, 2010 at 07:36:38PM +0200, Martin Baehr wrote: > On Mon, Oct 18, 2010 at 01:31:27PM -0400, Michael K. Johnson wrote: > > $ conary rq --components --all-troves --trove-flags > > uvcvideo-kernel=/foresight.rpath....@fl:devel//2-qa-kernel/0-0.2-1 > > uvcvideo-kernel=0-0.2-1 [Redirect -> Nothing] > > uvcvideo-kernel:debuginfo=0-0.2-1 [Redirect -> Nothing] > > uvcvideo-kernel:runtime=0-0.2-1 [Redirect -> Nothing] > > What should it have done? > > it should have told me that the package is removed. > conary rq should have told me that actually.
It just did, see above. The simpler query is whether a trove exists. It does. Conary has never before told you about resolving these redirects; if you install that package with the old model it resolves the redirect and does the same thing. It's not clear to me that this behavior needs to change. If we later implement model tracking and simplification, I'd think that we would want to consider the possibility of reporting redirects. However, it is reasonable for a trove to sometimes be one redirect, then be replaced with a different redirect, then be replaced by a normal trove, then be a redirect again -- think of it like a symlink. You don't think of it as an error that the shell does not report a warning for every symlink that the kernel traverses when you try to cat a file, for instance. So I see this in the context of generally reporting information that allows you to refine your system model, rather than as a specific case that needs exceptional handling. Does this make sense? If not, what's the context that I'm missing? (Thanks again, as always!) _______________________________________________ Foresight-devel mailing list Foresight-devel@lists.rpath.org http://lists.rpath.org/mailman/listinfo/foresight-devel
