Bug#775194: RFS: mininet/2.2.0 ITP - process-based network emulator

[Tomasz Buchert <tomasz.buchert@inria.fr>]

On 12/01/15 15:03, Vincent Bernat wrote:
>  ❦ 12 janvier 2015 14:30 +0100, Tomasz Buchert <tomasz.buchert@inria.fr> :
>
> >   It builds those binary packages:
> >
> >     mininet - process-based network emulator
> >
> >   To access further information about this package, please visit the following URL:
> >
> >   http://mentors.debian.net/package/mininet
> >
> >   Alternatively, one can download the package with dget using this command:
> >
> >   dget -x http://mentors.debian.net/debian/pool/main/m/mininet/mininet_2.2.0-1.dsc
> >
> >   More information about hello can be obtained from
> >   http://www.mininet.org.
>
> In Ubuntu, the package is maintained by James Page. I pinged him a week
> ago about packaging it in Debian but got no answer. Your package seems
> an original work. Did you try to reach James about that? Did you look at
> how the problems you got have been solved in this version?

Hi Vincent,
I don't quite remember if I have met James "electronically" yet.
I know, however, that Ubuntu packaging of openvswitch provided
openvswitch-controller which solved the problem for them.
This is described in the bug #757761.

>
> About the package:
>
>  - in d/control, you recommend openvswitch-controller but no such
>    package exists in Debian.

Thanks, my bad.

>  - in d/copyright, you license debian/* under GPL-2+ but since the
>    original software is licensed as MIT, it would be "better"
>    to use the same license. This allows upstream to integrate your
>    changes more easily.

That makes sense. However, the license is *not* MIT literally
speaking (https://mailman.stanford.edu/pipermail/mininet-discuss/2014-August/004879.html).
I renamed the license to "mit-mininet-license" and used the same
license for debian/* as you proposed.

>  - in d/copyright, the license is MIT with a preface, just use MIT as
>    the keyword (but keep the whole license).

See above.

>  - in d/repack, why is this script here if not used?

Well spotted. I removed it (note, however, that there is a comment
at the beginning saying that it may be used one day).

>  - in d/rules, you use python_distutils as a build system, this will
>    call "python setup.py clean" in dh_auto_clean, not "make clean". This
>    explains why you have to use "make clean". As for CPPFLAGS, this is
>    because the Makefile don't include it. I think both of your fixes are
>    fine.

Ok, I removed todos.

>  - d/TODO should be removed if those problems are solved.
>

Done.

> I have not tested the result, but the package looks good otherwise.

I've just reuploaded the package to mentors. It should be visible in few minutes.

Thanks a lot,
Tomasz