[Ksummit-2008-discuss] Fixing the Kernel Janitors project

James Bottomley James.Bottomley at HansenPartnership.com
Wed Jun 4 10:17:14 PDT 2008


On Wed, 2008-06-04 at 09:12 -0700, Grant Grundler wrote:
> On Wed, Jun 4, 2008 at 4:35 AM, Jiri Kosina <jkosina at suse.cz> wrote:
> >...
> > On the other side, I know that universities (in their computer-science
> > oriented study programmes, of course) have often problems with not having
> > enough interesting projects for "Operating systems" kind of classes; even
> > coming up with topics for master theses could sometimes be painful for
> > them, if there are more students than creative ideas for work topics.
> 
> I was told the Master's thesis needs to involve some theoretical work
> and not just a coding exercise. But undergrad students certainly can
> use the "work experience" on more basic projects.

Depends on University.  For my university in the UK, the requirement for
a master's thesis is "a useful contribution to learning" and the
requirement for a PhD is "a significant contribution to learning".

The US system is different because you don't get either a master's or
doctorate via pure research like you do in the UK, so the US coursework
counts as "theoretical work".  The main problem is getting the academic
committees to recognise open source contributions as "contributions to
learning" for a thesis.  I have actually spent some time talking to
academics about this, particularly in the field of systems research,
which is where most of the OS related stuff ends up nowadays.  Right at
the moment, the non-systems computer science people seem to veto the
idea of code as "contribution to learning" and instead focus on output
"as in what the code does", which is why we get a lot of academic papers
using older versions of linux with lots of results but the code tends to
get lost.

> >
> > So maybe
> >
> > - keeping the list of projects / TODOs interesting enough (*)
> > *and*
> > - putting some more effort into making sure that universities are aware of
> >  the fact that we have such list of things that could save them effort
> >  because they wouldn't have to pull work assignments out from the air
> >  completely (**)
> 
> Subsystem maintainers can register as "mentors" for Google Summer Of Code
> and equivalent programs at other companies. AFAIK, universities are targeted
> for promoting those projects. Unfortunately, 2008 SoC is already off and running
> but should keep it in mind for next year. Perhaps Linux Foundation could
> be the focal point to help organize this for linux kernel in 2009.

The google summer of code is really too short for university projects
(other than things which go as small components to theses).  However, we
can use it for other things ... like enhance your entry credentials to
masters or doctoral programmes.  It's also a bit short for getting into
kernel internals because the learning curve becomes too steep.

However, the Linux Foundation could look into finding a few projects, I
suppose ... if it works OK for the fBSD people, it might work for us.

> > we could then receive non-trivial amount of newcomers providing useful
> > code and getting closely acquainted with kernel coding for really low
> > price.
> >
> > (*) I guess every subsystem maintainer has a lots of things he would like
> > to have implemented in his area; at least I have :)
> > (**) This would mean really some slight PR aimed towards universities;
> > maybe Linux foundation could help here a little bit?
> 
> Linux Foundation was historically more focused on "Industry Partners". I think
> those partners have better connections to the Universities since they are
> the ones attempting to recruit.

Actually, that's not really true anymore: OSDL, it's predecessor
organisation was focused on Industry; the Linux Foundation is focused on
collaboration between all players in the ecosystem, of which
universities are a part (there's even a special member class for them).

James




More information about the Ksummit-2008-discuss mailing list