Dear Andreas,
I understand your argument.
I decided to keep it in science. This is actually also how the
NeuroDebian package it build and how Yaroslav (from NeuroDebian)
proposed it originally . I included an override of the Lintian
warning. (Easy, since Raphael did it already).
Due to all this changes, I think it is also required to "debchange
-i", right?
Oliver
On 02/04/2014 09:36, Andreas Tille
wrote:
Hi Oliver,
On Tue, Apr 01, 2014 at 09:14:33PM +0200, Rafael Laboissiere wrote:
* Oliver Lindemann <oliver.lindemann@uni-potsdam.de> [2014-04-01 19:02]:
Thanks. I applied all patches and kept, in line with Andreas
comment, the section as "science".
You might then apply the patch attached below for overriding the
Lintian warning.
Well, I might have choosen a wrong wording: I tried to suggest
following lintian in using "Section: python". The thing is that this
Section field has more or less no real meaning and it finally does not
really matter what you put there. But if lintian *tells* you that
policy would expect some section I'm personally following its advise
rather than using some lintian overrides. If a package clearly falls
into *both* sections there is so few reason to worry about the section
field that there is no point in adding any extra line of code (like an
lintian override).
I'm trying to rephrase my arguing a bit:
In a Section field the package maintainer tries to guess what a
user might do with the package - in this specific case either
Python programming *(exclusive) or* scienctific research. This
exclusive or is obviously nonsense.
In Blends tasks different users declare in what field they are
using a certain package and for sure a package can show up in
various differnet tasks. So this package could show up in
*different* sciences and - just to stretch the example - in
certain games programming or education.
So the tasks concept is to some extend fixing the Section concept of
debian/control which is weak in several ways but we carry it for
historic reasons. Finding a value which keeps lintian happy is a
perfectly valid way to deal with this in my opinion (as long as lintian
is not really wrong - which I can not see in this example).
Since this is my personal opinion I'm perfectly fine with every choice
you might draw.
Kind regards
Andreas.
--
Dr. Oliver Lindemann
Division of Cognitive Science
University of Potsdam
Karl - Liebknecht Str. 24/25, Building 14, 14476 Potsdam,
Germany
Tel: +49 - 331 - 977 2915, Fax: +49 - 331 - 977 2794
Room: 6.24, Building 14, http://www.cognitive-psychology.eu/lindemann
|