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

Re: New user requests and etch-bpo



Frank Küster <frank@debian.org> wrote:

> I think it would be strange to have to install tex-common from etch with
> pinning, when texlive which depends on it is on backports.org.

ACK. In fact, I'm uncomfortable with this new rule that "packages which
can be installed directly from testing shouldn't be uploaded to bpo".

The reason is that it's been a loooooong time ago since I've given up
looking at .debs from testing or unstable trying to see if they can be
installed on stable. First, it almost never works (at least for arch:
any packages, due to dependencies on shared libraries such as libc6), so
you're losing your time downloading the .debs and checking their Depends
and Conflicts. Second, in the rare cases where it works, it results in
an ugly setup, because the .debs directly installed via "dpkg -i" aren't
in the available file and thus, dselect displays them as
"Obsolete/Local". Eeek.

For these reasons, I've long ago resorted to having a deb-src line for
_unstable_ in my sources.list on the _stable_ system I use daily, and
whenever I'm looking for a package from testing or unstable, I first
check whether it's already on bpo (which is much quicker than
downloading the right set of .debs and checking their dependencies,
maybe leading to recursively download other .debs they depend on until
one of them can *not* be installed---ugh!), and if not, I can usually
make my own backport with four simple steps:
  - apt-get source <pkg>;
  - edit the debian/changelog file;
  - debuild;
  - <insert here you favorite incantation for uploading the package to
     a local repository>

So, I wouldn't be hostile to this rule being reconsidered.

-- 
Florent

Reply to: