[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: libc6 dependency generation



> > Is it possible to have libc6 not list its dependency information with
> > so much accuracy?  That is, could it leave off the "-devrev" or maybe
> > even the most minor revision number (".x-debrev")?
> >
> > As it is, it's impossible to do any kind of mixing between stable and
> > unstable even when the difference in libc6 is only 2.2.5-4 to 2.2.5-13.
> >
> > apt-get absolutely refuses to ignore a package that I've forced the
> > installation of when a dependency like this is violated and yet
> > the newer package from "unstable" runs just fine.
> 
> If only you knew what you were talking about before you decided to say
> it was wrong.

Be fair now; I simply asked if it was possible.  I never said it was
wrong.  I said it made things difficult.  Starting a discussion by
telling me not act stupid (even if it were true) does not make me open
to your ideas.


> The fact is that we pull in some changes from CVS quite often. A lot of
> times these are ABI breaking changes. It might not affect most
> applications, but it's impossible to do "if this uses the new foo()
> interface, then we need libc6 >> 2.2.5-13, otherwise we just need libc6
> >> 2.2.5-1".

It seems to me that changing the ABI with every release of the package is
a bad idea to begin with.  After all, there are some systems using
binaries that are _not_ debian packages (with all the exact dependencies)
and it would be nice to have them continue to run even after a upgrade,
espectially one as apparently minor as 2.2.5-4 to 2.2.5-13.


> libc6 is a fairly tricky package, especially since ABI changes can
> change across several platforms or even just one. Better to err on the
> side of safety, than to create a situation where ppl's system can go
> poof.

Of course it is.  But erring on the side of safety would really mean not
changing the ABIs.  Changing that is what makes peoples' systems "go
poof"; having detailed dependencies just helps hide the problem better.
It's not that different from dressing up a bug and calling it a feature.

                                          Brian
                                  ( bcwhite@pobox.com )

-------------------------------------------------------------------------------
   Tired of spam?  See what you can do to fight it at: http://www.cauce.org/



Reply to: