Re: Proposed recompilation to remove dependency on libc6 2.0.7u
On Wed, 16 Dec 1998, Martin Schulze wrote:
> > dpkg *refuses* to even unpack a package when a Pre-dependency is not met,
> > which is not true for normal dependencies.
> So what? Run [I]nstall again and feel happy.
Oh, yes "Run [I]nstall again", but we should make this an old-fashioned
recipe. Why do you want to force the users to [I]nstall again if we can
avoid it so easily?
It seems to me that the other's people time has not any value for you.
> I don't mind having a 2nd install call, a 2nd configure run etc.
> Since system integrity does not break by doing so I don't see
> an urgent need to recompile these packages for slink,
If we want to make the upgrade hamm->slink as smooth as possible, we
should have to recompile them for slink. Otherwise it would be too late.
> hence forcing
> all affected ports to also compile them,
If the ports have also a bogus Pre-Dependency, they are also affected,
so they would not only recompile to be in sync with i386 but also to solve
this same problem.
> hence to wait for their
> installation, hence for their testing etc.
Since the source is the same, there is nothing additional to test.
The package will just have a sane Pre-Dependency.
> If there are serious problems with certain packages they need
> to be fixed,
No, in either case it is a bug anyway that a package has a dependency
which is stronger than it should be.
We are talking about considering them important bugs for the case the
package is essential.
I already know that you don't mind to [I]nstall twice, but I do, and most
people will not want to have to [I]nstall twice.
Do others value so poorly the userfriendlyness of the upgrade process?
"c800ee36e4a7ce2ef7c7690d20bac6cc" (a truly random sig)