[libvirt] Google Summer of Code 2013 ideas wiki open

[Daniel P. Berrange]

On Mon, Apr 15, 2013 at 09:43:55PM +0800, harryxiyou wrote:
> On Mon, Apr 15, 2013 at 5:13 PM, Daniel P. Berrange <berrange at redhat.com> wrote:
> [...]
> > I'm not hugely comfortable with the idea of "capability support" being
> > done by a student. IMHO to do a good job on that design-wise requires
> > someone with a very good understanding of libvirt architecture & application
> > needs.
> >
> 
> I understand. However, i think our Libvirt is developing so we should give more
> choices to learners who are very interested in some field of Libvirt.(Like me, i
> love the storage system of Libvirt very much). Maybe this is the essence of
> GSOC, isn't it? Actually, some student is not only interested in
> Libvirt but also
>  wanna to join this community and contribute to this community forever. (Like
> me, i love the community because i can learn more knowledge from it.)
> I believe that interest is the best teacher. No matter how the problem is
> difficulty i will try my best to achieve it if  i am very interested
> in it. Another
> key point is that GSOC just let students join the community and finish easy
> jobs firstly. GSOC wanna train more core developers for our community. If i
> can finish a job a bit difficulty, i can also accomplish it after GSOC
> continuously.
> All in all, i think you should not worry about this matter ;-).

Enthusiasm of the students to learn libvirt isn't the only consideration.
The project mentors have to put non-trivial effort into GSoC and want to
have some level of confidence that the project will result in a positive
outcome for both the student & the project.

I think a project looking at adding 'rename' API support for all objects
in libvirt has a high liklihood of success, since it is a clearly defined
problem with easily measurable success criteria and a fairly unambiguous
design to follow that will not require much debate.

A project looking at "capabilities" has a much lower liklihood of success
because it very focused on architectural design. Getting such design right
requires significant knowledge of libvirt, and will require significant
debate & discussion by many parties on the list. Design discussions of this
kind pay no attention to any external deadlines, that programs like GSoC
have. I'm not saying a student couldn't write something, but I'm not
confident that the result would be something we'd be prepared to merge,
and by that benchmark could be considered a failure.

I think GSoC projects should focus on something where the design is clearly
defined upfront, and the task is mostly about learning the libvirt codebase
& getting a thorough implementation done.

Regards,
Daniel
-- 
|: http://berrange.com      -o-    http://www.flickr.com/photos/dberrange/ :|
|: http://libvirt.org              -o-             http://virt-manager.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: http://entangle-photo.org       -o-       http://live.gnome.org/gtk-vnc :|