Re: APT broken ?
On Sat, 4 Apr 1998, Dale Scheetz wrote:
> On Sat, 4 Apr 1998, Jason Gunthorpe wrote:
>
> >
> > On Sat, 4 Apr 1998, Raul Miller wrote:
> >
> > > Jason Gunthorpe <jgg@gpu.srv.ualberta.ca> wrote:
> > > > APT will -never- allow you to do something that results in broken
> > > > dependancies, it does very complete checking of this at all stages of
> > > > the installation.
> > >
> > > How do you mark a package as "outside the scope of APT"?
> >
> > There is not a way presently. That might be a good solution to the problem
> > with desired unmet dependancies... Will think about it.
> What happened to Hold, from dselect? Isn't APT satisfying such
> compatibility issues?
APT interprits hold exactly as hold means, do not install a new version of
this package [and it only does that on upgrade or dselect-upgrade
commands].
Hold does not mean that the package's broken dependancies are okay, and it
probably never should mean that.
> I would suggest that packages marked "on hold" should be considered to
> satisfy any dependencies on them. This would provide a mechanism for
On them?? I think that is a serious abuse of what hold means. Their own
dependancies I could relucantly accept, but not dependancies on them.
My current feeling is to make a flag that completely removes a package
from APT's view of the system, as far as APT is concerned it does not
exist [and all dependancies on it will fail]. That is probably the best
solution to the orignal problem.
What you are talking about is something completely different and might be
handled by another flag - I would have to consider that some more -
it would certainly be a very dangerous flag to use.
Hold is certianly not the correct thing for either sitatuation.
Jason
--
To UNSUBSCRIBE, email to debian-devel-request@lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster@lists.debian.org
Reply to: