[Debichem-devel] [Debichem-commits] r612 - in /unstable/openbabel/debian: changelog patches/python_setup_fix.patch

[Michael Banck]

On Sat, Jul 21, 2007 at 03:17:36AM +0200, Daniel Leidert wrote:
> Am Freitag, den 20.07.2007, 18:02 +0200 schrieb Michael Banck:
> [..]
> > Hrm, another worry: In principle, I am not against shipping libinchi
> > >from the openbabel source package, but long-term I hope inchi will get
> > its own source package and all.
> > 
> > Now, the current package is libinchi0_2.1.1-1, and I'm worried that this
> > calls for an immediate epoch as soon as inchi releases on its own.
> > 
> > Alternative would be linking statically to libinchi inside openbabel and
> > ignoring it for now (which we've done for openbabel-2.0 I think).
> 
> I would use an epoch, when it becomes necessary. This is IMHO the
> natural way. Is there anything bad with epochs?

Epochs (unconsitently) expose an implementation detail of our versioning
policy to the users which should really not be bothered with it.

But thinking about it some more, I am less convinced that we should
build inchi packages from the openbabel source - openbabel itself just
keeps the source in the tree because there was no good external source
to rely on (and there still really isn't).  The OpenBabel maintainers do
not maintain the inchi code inside openbabel, and there's no particular
reason why the openbabel source package should have it.

Are there other Debian packages (or prospective packages) which would
profit from libinchi packages?  If no, I would rather opt for not
introducing those libraries now and rather package them seperately.


Michael