r36486 - in /packages/R/r-cran-xtable/trunk/debian: ./ README.Debian changelog control copyright rules

Charles Plessy plessy at debian.org
Tue May 18 08:52:02 UTC 2010


Hi Andreas,

Le Tue, May 18, 2010 at 10:04:57AM +0200, Andreas Tille a écrit :
> On Tue, May 18, 2010 at 12:46:30AM -0000, plessy at users.alioth.debian.org wrote:
> > Author: plessy
> > Date: Tue May 18 00:46:29 2010
> > New Revision: 36486
> > 
> > URL: http://svn.debian.org/wsvn/debian-science/?sc=1&rev=36486
> > Log:
> > Uploaded the updated package prepared by Andreas, after adding a couple of modifications.
> 
> Great!  Thanks.

You're welcome :)

> > +  * Added notes to explain how this package is tested (debian/README.Debian).
> 
> While this is interesting I'm afraid I might simply forget to do this
> procedure

I am afraid of this too, and actually the testing is done only on the architecture
where the DD uploads (and this will not hold true in the future…). Also, in case
of binNMUs, no testing is made at all.

Nevertheless, I like to document how to simply test the package. How about
keeping the same notice, but calling it ‘notes to explain how this package can
be tested’?

By the way, in some cases the notice always true: when there are regression
tests that are done at build time, I (try to) indicate this in README.Debian
with a URL to the build logs. For these packages (typically Perl modules) I
usually do not do more testing than trying to upgrade the package.



> > Modified: packages/R/r-cran-xtable/trunk/debian/copyright
> > URL: http://svn.debian.org/wsvn/debian-science/packages/R/r-cran-xtable/trunk/debian/copyright?rev=36486&op=diff
> > ==============================================================================
> > --- packages/R/r-cran-xtable/trunk/debian/copyright (original)
> > +++ packages/R/r-cran-xtable/trunk/debian/copyright Tue May 18 00:46:29 2010
> > @@ -1,7 +1,7 @@
> >  Format-Specification: http://wiki.debian.org/Proposals/CopyrightFormat
> >  Upstream-Name: xtable
> >  Upstream-Maintainer: David B. Dahl <dahl at stat.tamu.edu>
> > -Upstream-Source: http://cran.r-project.org/src/contrib/
> > +Upstream-Source: http://cran.r-project.org/src/contrib/xtable_1.5-6.tar.gz
> 
> Perhaps we should find a common agreement about the Upstream-Source
> field: From a maintenance point of view it makes sense to not touch to
> many files in the debian/ directory and thus I try to avoid touching
> debian/copyright for every new version.  I admit that the download
> directory contains a lot of files and thus just specifying the directory
> is somehow indirect.  But do we really win so much when specifying the
> exact file by having at the same time the risk to have an incorrect
> copyright file if a maintainer forgets to update this field.  I'm
> definitely a candidate for forgetting this.

I admit it is a bit of overdoing. My motivation is that I strongly dislike when
the download like is obsolete (404), so I test it at each upload (uscan
--force-download,…). As a result I get the full URL for cheap. A Lintian test
could be better, but I do not have much time for proposing one with a patch,
and the DEP5 is not yet accepted anyway.

Have a nice day,

-- 
Charles



More information about the debian-science-maintainers mailing list