[Debichem-devel] cclib

[Michael Banck]

On Sun, Apr 17, 2011 at 09:58:07AM +0200, Karol M. Langner wrote:
> On Sat, Apr 16, 2011 at 07:49:58PM +0200, Daniel Leidert wrote:
> > > > > W: cclib: binary-without-manpage usr/bin/ccget
> > > > > W: cclib: binary-without-manpage usr/bin/cda
> > > > 
> > > > If you provide the information about switches, environment variables,
> > > > etc. I can write the manual page in GROFF for you. Check e.g.
> > > > apbs/debian/apbs.1 or pymol/debian/pymol.1 (call `man -l ...file...') to
> > > > see, which information is usually provided by a manual page. Or check
> > > > this howto:
> > > > http://www.schweikhardt.net/man_page_howto.html
> > > > 
> > > > I also wrote a small howto for writing a manual page for Debian
> > > > packages: http://www.wgdd.de/?p=65
> > > 
> > > Thanks for the links. I wrote up minimal manpages for both ccget and ccda.
> > 
> > I will check that later. JFTR: ccget and ccda are not very specific.
> > There is a high chance, that we will see a file conflict sooner or
> > later. Any chance to rename them to cclib-cda and cclib-get?
> 
> Dobe in r2699.

You should rename the manpages accordingly I think, the lintian warning
is back.

I further see this one again:

W: cclib source: build-depends-on-python-dev-with-no-arch-any

Probably my fault for suggesting to Build-Depend on that, apparantly
"python" would be sufficient (but not a big deal anyway).

Finally, I think it would be prudent to tighten the Depends of cclib on
python-cclib, currently it is unversioned.  Not sure how tightly the
scripts (and tests I guess) are coupled to python-cclib, but as the
scripts appear non-trivial. I would say a (= ${source:Version}) would
be ok, but what do the others think?


Michael