[Netkit.users] Clarification about the contributions process ?

Massimo Rimondini rimondin at dia.uniroma3.it
Thu Jun 6 11:00:11 CEST 2013

Dear Julien,

thank you very much for your reply and, once more, my apologies for 
being so slow: these mails require a bit of thinking before replying, 

Il 18/05/2013 23.08, Julien Iguchi-Cartigny ha scritto:
> Dear Massimo,
> [snip]
>> As you have observed, development has been a bit frozen for a while,
>> due to engagement of (human) resources in many other
>> research-related activities. However, a fair survey should also
>> reckon that labs are updated on a regular basis (also see the
>> "Recently updated pages" report at
>> http://wiki.netkit.org/index.php/News), support requests are always
>> answered, and the software is in a good enough shape to be fully
>> functional on a wide variety of platforms, with few exceptions, thus
>> decreasing the urge to apply critical fixes.
> First I need to apologies for my previous rough mail, I know it is
> difficult to find time / human ressources (especially in the academic
> sector). I also want to thank you for all your work you've done for Nekit.

Don't worry: your arguments are of course completely understandable.

>> At the same time, our
>> presence on Github is a precious opportunity to collect issues and
>> proposed fixes/enhancements (pull requests) that, although currently
>> stale, are available to us for being merged upstream in a more
>> development-friendly moment.I'm omitting that forks, which enable
>> independent development by others, are of course possible, as you
>> have experienced yourself.
>> That said, I don't think that "dead" is an appropriate adjective to
>> describe the project status. I understand that our current
>> "stuckness" in accepting pull requests may discourage further
>> submissions, but if asking for changes to go upstream is too slow a
>> path, another possibility in the meantime is to send to us a packaged
>> contributed release, which we would be happy to publish on the web
>> site. This would take much shorter and is independent of our current
>> "development power".
> I believe I have another usage / vision about Nekit. You're looking for
> a stable architecture and adding / upgrading labs. In my case the netkit
> software is more interesting with my own labs. But for my goal netkit is
> not usable for several reasons:
> - the "untar and run" idea of netkit which permit to use it without
> installing software or having the root rights is not totally
> supported at the present time: for 64 bits architectures it requires
> some libs (32 bits libc for instance). It could be problematic for some
> cases (for my case: i was a computer room in a foreign country for a
> network course and i couldn't use nekit because computers had only a
> fresh 64 bits install of ubuntu and a slow internet connection).
> - the integration of uml_dump is really needed, i've received and helped
> several people for integration of uml_dump in their netkit installation,
> and my archive of netkit 2.8 with uml_dump has been downloaded more than
> 100 times on 2 years.
> - building the netkit-uml-filesystem is definitively broken, as it is
> based on sid which is not maintened anymore and has a very annoying bug
> with debootstrap (it forgets some dependencies for dpkg, the only way is
> to collect a bunch of deb files and install them by hand before
> launching the debootstrap process)
> - furthermore, there is urgent need of an updated filesystem to have
> recent tools like a working version of bind when using dnssec
> - same story with the kernel, i've already fixed some bugs with the
> actual building process.
> I think to have a better communauty and expansion as a teaching tool,
> netkit needs:
> - a netkit-core with all dependencies (static linking) and with uml_dump
> support in two versions: 32 and 64 bits (already available in my repository)
> - a netkit-uml-filesystem and netkit-uml-kernel which can be build with
> recent version of debian (these are also available in my repositories
> but there is still some bugs in the filesystem). A modular build system
> which can build various version and offer simple way to add new
> softwares and modifications by netkit users is a plus.

So, as far as I can see and ascertain myself, the main problems can be 
classified in the following categories:
- dependency requirements on host-side software
- lack of some additional tools
- freshness of existing components
- flaws in the build systems

> We are really close to this goal (but it has also some drawbacks, for
> instance the lack of MPLS support). I'm proposing the following
> improvements to the netkit development process:
> - I will create a simple web page with all new versions builded and with
> aditionnal documentations. I will try to update this page with fork and
> fixes and update my repository. People can then try the new versions and
> report bug (preferably on your issue tab to avoid dissemination of
> issues accross forked repositories.

I think that a web page where you could gather your contributions would 
be very useful.
If you could give us a pointer, we would be happy to include it in the 
Netkit web site.

> - When you will consider it ready, you can integrate the build and data
> in netkit wiki and pull the commits from my repository to yours

As I said above, maybe a direct pointer to your page could be an 
effective way to not even subject access to your contributions to pulls 
to our repository.

> - if possible, could you give public or at least to the main
> contributors the right to close issues in your repositories.

I think we still need to understand if/how this can be done, but we will 
consider it (or attempt to close the issues ourselves).

> Of course, it is just a proposition and I wait your feedback.

It's a very welcome proposition, of course. Thank you!


> Regards.
> Julien
> _______________________________________________
> Netkit.users mailing list
> Netkit.users a list.dia.uniroma3.it
> http://list.dia.uniroma3.it/mailman/listinfo/netkit.users

More information about the Netkit.users mailing list