Re: RFC: casacore-2.1.0-1 debian package
thank you very much for the update! Especially for looking through the
files and their licenses. We are getting closer...
There is a Lintian warning
"copyright-refers-to-deprecated-bsd-license-file". This is connected to
the phrase in debian/copyright:
On Debian systems, the full text of the BSD-3-clause can be found in
the file `/usr/share/common-licenses/BSD'.
Since you already put the fill BSD-3-Clause text into debian/copright,
the simplest would be to just remove the phrase above.
Gijs, I didn't follow how the Python bindings work in casacore: I assume
that the libcasacore-python library is not enought to do an "import
casacore", and one needs your "python-casacore" package, right?
Could you also explain what is needed to support Python 3 here? Or maybe
(casacore is team-maintained at the end) just push the needed changes to
our git repository, if Benda agrees? I think this would speedup the
completion of the package.
In the moment, I have only limited internet access (being at the VLT in
Paranal), but maybe we can get it ready so that I can upload next week?
This would be great!
Gijs, would you write an ITP for the python-casacore package? I think
this would be the logical next step here. ;-)
On 09.05.2016 03:49, Gijs Molenaar wrote:
> Hi Benda,
> Thanks for maintaining this package. I noted that you use the name CASA
> sometimes. CASA is not CASACORE, CASA is a superset of CASACORE which is
> maintained by NRAO.
> Also, I reported 2 other issues a while ago by email, I think they where
> about python3 support and mixing spaces and tabs in the control file.
> Did these issues get lost? What is the right procedure for reporting
> issues with packages that are not in the official repositories yet?
> greetings, all the best,
> - gijs
> 2016-05-09 5:28 GMT+02:00 Benda Xu <email@example.com
> Hi Ole,
> I have pushed 2 commits to
> Ole Streicher <firstname.lastname@example.org <mailto:email@example.com>> writes:
> > Benda Xu <firstname.lastname@example.org <mailto:email@example.com>> writes:
> > So, here is my second part of the review:
> > debian/copyright
> > ================
> > * cmake/* has many different licenses and copyright holders, not
> > mentioned in debian/copyright yet, f.e.
> > FindCFITSIO.cmake: GPL-3, (c) 2009 ASTRON
> > FindBISON.cmake: BSD, (c) 2006, 2009 Kitware Inc, Tristan Carel
> > FindNUMPY.cmake: MIT, (c) 2012 Continuum Analytics, Inc.
> > and others
> > * "National Astronomical Observatory of Japan" is missing as copyright
> > holder for some files (first LGPL-2+ paragraph)
> > * Some files are not LGPL-2+, but GPL-2+, f.e.
> > fits/apps/fits2table/fits2table.cc or fits/FITS/test/tfits2.cc
> > please check this carefully.
> > * some files in d/copyright do not exist anymore (lintian -I)
> > images/Images/ImageReorderer.h, install/codedevl/glish-mode.el
> > * mirlib/* is a bit unclear for me. Can you confirm that they are
> > GPL-2+?
> They are LGPL-2+, as claimed in mirlib/README:
> > debian/control
> > ==============
> > * replace VCS-Git: with (lintian -I)
> > https://anonscm.debian.org/git/debian-astro/packages/casacore.git
> > * python --> Python in description (lintian -I)
> > Longer-term wishes
> > ==================
> > These are not needed for the initial upload. It would be however good to
> > keep them on the To-Do-List. Both also come from running `lintian -E -I`:
> > * discuss with upstream that a shared lib should not call
> > exit(). Calling exit() in the library makes it impossible to handle
> > the error in the calling program.
> > * Create symbols files for the shared libraries. This would allow for
> > dependent package to automatically determine the minimal required
> > version of the package.
> Yes, the libraries are not optimal. I will put those in my future plan.
> > I think the only real issue now is to fix debian/copyright. Could you
> > re-check it again carefully? The ftp-masters do this quite carefully
> > (you know already :-) ) and we would have another delay if they reject
> > the package. If you are ready, I will upload.
> Thank you very much for your review.
> Gijs Molenaar