[Openicc] 2008 Google SoC preparations

[Hal V. Engel]

On Monday 03 March 2008 00:34:25 Cyrille Berger wrote:
> On Sunday 02 March 2008, Kai-Uwe Behrmann wrote:
> > > I also think we need to have a section to provide some guidance to
> > > students concerning our expectations on the applications and the
> > > application process. Perhaps we can reduce the number of copy and paste
> > > type applications and get more applications that can be seriously
> > > considered.   The same thing also applies to expectations for the
> > > period between students being selected and the beginning of the actual
> > > project work.
> >
> > Yes, I'd like to pick Boudewijns suggestion and communicate almost on a
> > emaillist. Beside transparency, we can require a *nix environment as
> > fallback. Eigther a student shows she/he can build open source libraries
> > on Xcode/VC or we make a call to use Linux/BSD. The later are regulary
> > to meet among all sort of developers.
>
> I think *nix environment should be a requirement. Last year, my student and
> me lost a considerable amount of time on making run the various available
> tonemapping algorithms on Windows.

This was also an issue for my project and from reading the various GSoC forums 
is apparently a fairly common issue.  We lost perhaps 2 full weeks getting 
things working on Windows and this definitely placed the project in a bad 
position.   I think that part of this is that setting up a build environment 
on a Windows system is significantly more difficult than on a *nix system and 
students think they have a handle on it because they have things working for 
their relatively simple school projects.  They don't realize how much work 
they will need to do to get this working for the much more complex GSoC 
projects.  

Fortunately I was able to work with the student to make up most of the lost 
ground but I would have preferred to not have had to deal with the issue.   
On the flip side there were also some benefits to the student working on 
Windows.  For example, he did fix a number of preexisting Windows specific 
bugs.

I am inclined to be a little flexible since I don't have an issue with the 
student working on Windows as long as the project is intended to be cross 
platform AND the student has everything working on Windows well before the 
beginning of actual project work.  The rub with the last one is how do you 
ensure that the student has in fact gotten things working on Windows before 
project start up?  

This is a heads up to potential mentors.  If your student wants to work on 
Windows then you must insist that the student has a fully functional build 
environment well before the start of the GSoC project work.  Next time around 
I will insist that the student sends me a working Windows binary of current 
CVS at least a month before project start up.

Hal