On Mon, Nov 21, 2005 at 04:52:59PM +0100, Enrico Zini wrote: > On Sun, Nov 20, 2005 at 12:31:22AM +0100, Hervé Eychenne wrote: > > On Sat, Nov 19, 2005 at 01:57:08PM +0100, Enrico Zini wrote: > > > Implies is currently totally ignored by libtagcoll. The reason is that > > > I feel that it's better to do all this automatic tag inference in the > > > autodebtag script, and to provide clients with a nice and complete tag > > > database. > > If so, I think this is a design mistake. > > Implications in inherent to tagging, so it definitely must be handled > > by the tag library. > > Programs using libtagcoll should not have to reinvent autodebtag's > > wheel.
> I don't understand: the idea is that the result of autodebtag is to be > fed back to the package archive, and regularly. In that respect, > programs using libtagcoll will have no need to reinvent autodebtag's > weel because they would find a complete set of tags anyway. Maybe I didn't understand exactly what it is all about... All that I wanted to say is that tag dependencies and implications is something that is related to the concept of tagging itself (and not to *_deb_tags, which is "only" a practical application of tags for Debian packages). So implications must be dealt with by libtagcoll as much as possible, and libtagcoll must contain as much code as possible related to generic operations about tagging (such as the handling of implications, here). So, if you should agree with this basic principle (factorize as much as possible in the low layers), what makes you feel that tag inference should be done by autodebtag, and that implications should be ignored by libtagcoll? Hervé -- _ (°= Hervé Eychenne //) Homepage: http://www.eychenne.org/ v_/_ WallFire project: http://www.wallfire.org/ _______________________________________________ Debtags-devel mailing list [email protected] http://lists.alioth.debian.org/mailman/listinfo/debtags-devel
