Re: .py endings or no .py endings for scientific packages
Le Sat, Jun 16, 2018 at 08:14:36PM +0200, Steffen Möller a écrit :
>
> In a recent email exchange with Chris I had explained the situation
> with downstream packages depending on those .py files and he just
> suggested to properly override the lintian warning with an explanation.
> I found that a very reasonable thing to do.
Hi Steffen,
indeed, it is a Policy "should", meaning that one should only derogate
with a good reason, and I think that our reason is good. Lintian
overrides will be a good way to communicate and keep track of our
exceptions.
> How about a paragraph in the Debian Med policy? Maybe like
> "Many packages created for scientific purposes have executables with
> language-characteristic suffices like .pl or .py. This is not appropriate
> for public interfaces of a software that should be implementation-agnostic.
> It is a Debian Policy requirement to correct this for packages uploaded
> to the main distribution. Such distribution-specific changes however
> are discouraging the exchange of protocols between scientists. The
> package maintainer is encouraged to work with upstream to correct
> the naming of these files, but may decide to locally only override
> the lintian warning for the time upstream needs to distribute a corrected
> version of their software."
Good idea. And the Lintian overrides can link to the Debian Med policy.
Regarding the text above, I would go even further and not suggest to
contact upstream with a rename requests, unless the script is a "public
interface", which is almost never the case in our packages.
Have a nice Sunday,
--
Charles
Reply to: