Mailinglist Archive: zypp-devel (227 mails)

[Michael Matz <matz@xxxxxxx>]

Hi,

On Mon, 14 Apr 2008, Martin Vidner wrote:

> On Mon, Apr 14, 2008 at 02:18:23PM +0200, Michael Matz wrote:
> > On Mon, 14 Apr 2008, Duncan Mac-Vicar Prett wrote:
> > > - Use the solver, and during parsing inject something like
> > > provides:fix(bugzilla:23444) when filling the repodata. Then applications
> > > could inject a requires in the solver fix(bugzilla:23444) and just solve. 
> > > I
> > > dont't know the consequences of this.
> >
> > I don't think this would be a good idea.
>
> Why?

Conceptual inelegance always hints at some real problem hiding :)

> Because of flooding the solver with unnecessary data?

For instance.  And because it is no provide :)  I also would fear that, 
should we implement this as provides, someone goes and thinks it would be 
a good idea to add heaps and heaps of such virtual provides also to normal 
packages, whenever they fix some bug.  Then there's the problem how to 
_really_ model this as provides in the solver sense.  E.g. what if a 
bugfix needs two packages?  Making them both provide fix(bla) would still 
require only one of them to be fulfilled.

No, I don't think we should try to force this piece of information into 
the solver corset.  Make it easily searchable (we have that already), and 
expose this to the user.

> We do that all the time with autoprovides don't we?

In a way, yes.  But we shouldn't follow bad examples, should we?


Ciao,
Michael.
-- 
To unsubscribe, e-mail: zypp-devel+unsubscribe@xxxxxxxxxxxx
For additional commands, e-mail: zypp-devel+help@xxxxxxxxxxxx