[linux-dvb] em28xx merge process issues with linuxtv and upstream kernel
Jelle de Jong
jelledejong at powercraft.nl
Fri Oct 31 22:55:49 CET 2008
-----BEGIN PGP SIGNED MESSAGE-----
I would like to discuss some issues with the em28xx-new project and the
re-merging process with the upstream code and the linuxtv project.
First I would like to say that I am not an active developer on this
project, but a very heavy user that followed the project very close and
made some documentation, packages, helped getting some issues discussed
and some devices added. This project is important for me because I need
it to be merged successfully and with width support so it will become
maintainable on large basis.
This is the situation:
Markus Rechberger has worked for more than 3 years on a clone/fork of
the official code, creating the em28xx-new  project. He has worked
very very hard on this project and produced a lot of code and provides a
lot of support to the users of his project. It takes lot of energy and
resources to pull of this kind of work! I do really respect this work
and hope others also do this!
In a good free software development process somebody would clone some
work to his own personal repository, then add or fix one feature. Create
a diff compared with the upstream code and then. Commit the patch with
good description to the upstream project, so this feature and its code
changes can be reviewed, tested and supported. After this one feature
fix and patch, a developer moves to his next feature and repeats to
process so step by step and patch by patch the project grows. This
understanding of the code bye a larger group of people will lead to a
platform of developers that can maintain and contribute to the project.
This would be the proper way of doing things it most cases, but a
developer need to know and understand or mentored with this from the start.
However sometimes a developer works very hard on the code creating lots
of new features, adding support for lots of new devices and fixes lots
of issues without committing patches during the process. After 3 years
or more its understandable that it can lead to a code base almost
unrecognizable from the initial cloned code base. So what to do now! To
ask from the developer to completely separate all his work in individual
patches that can be committed to upstream is almost unrealistic. A good
plan needs to be created and discussed by the people that matter.
In this case I have seen several attempts of Markus to make large
patches and some smaller to try getting his code merged upstream, but
they were after short discussion rejected, because they did not fit the
one patch for one fix, feature, etcetera standard/idea.
I am going to ask for understanding of both the side of Markus that
worked very hard on his code, and that of the upstream developers. There
are both valid reasons on how they did there things.
But we need a solution to get all the code back into the upstream
project so it can go into the kernel project and eventually be delivered
at the Linux distributions and all there users, so no custom compiling,
custom package install are required and non transparent bug reports can
Is it possible for an upstream developer to step forward and take on the
task of merging the code of Markus back into mainstream, all questions
on the code can be discuses on several mailing-list  of choice.
Current the situation is kind of a hold-of, the issues are not being
discussed, the problem is not addressed, so no process is made and
during this time users are suffering from non working nor good supported
devices for there hybrid dvb-t/analog broadcast experiences under Linux.
I hope this lead to a productive discussion that will get the code to
the end users through there main distribution systems.
Jelle de Jong
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org
-----END PGP SIGNATURE-----
More information about the linux-dvb