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

Re: alternative dependency ordering - with respect of packages in main

On Tue, Sep 20, 2011 at 07:41:11PM +0200, Luk Claes wrote:
> On 09/20/2011 01:12 PM, Gerfried Fuchs wrote:
> >  Policy is clear on packages in main aren't allowed to depend on
> > packages outside of main.  Now in a fair amount of cases this has been
> > worked around by having the package outside of main as alternative
> > dependency and a package in main offer basic functionality for the
> > package to still be able to work.
> > 
> >  I know that the buildd system likes to pull in the first package in
> > such an alternative dependency chain.  And now I start to wonder:
> That being said, it might be that having a predictable set of build
> dependencies on the buildds is not wanted anymore?

I'm fairly certain that we do want predictability and consistency.

> >  tl;dr - what do you think, is a "Depends: foo-contrib | foo" acceptable
> > for packages in main or should it be "Depends: foo | foo-contrib"
> > instead?
> Personally I think we should stay with a predictable set of build
> dependencies on the buildds (only have them look at the first
> alternative) and we should stay with having a package in main as the
> first alternative of a build dependency of a package in main.

Just to comment on the build dependencies: first-only alternatives
are only implemented for Build-Depends and Build-Depends-Indep
(for all resolvers).  They get stripped when computing the set of
packages to install.

Alternatives in regular package Depends are delegated to
apt-get/aptitude, and what is installed will depend upon what is
available and installable.  For packages being build for main, the
contrib and non-free archives should not be available, so such
alternatives should not be considered.


  .''`.  Roger Leigh
 : :' :  Debian GNU/Linux             http://people.debian.org/~rleigh/
 `. `'   Printing on GNU/Linux?       http://gutenprint.sourceforge.net/
   `-    GPG Public Key: 0x25BFB848   Please GPG sign your mail.

Reply to: