[Mesa-dev] Summer of Code ideas (maybe just an idea wishlist?)

[Bruno Jimenez]

Hi,

Last year, I participated in GSoC, (yeah, I still read the mailing list
every day :) ) so I will give my 2 cents to the topic.

My background isn't at all related to graphics or computer science, I am
a physicist, although my main interest is in simulation. So that took me
to OpenCL, which in turn took me to clover and finally landed in Mesa. I
started looking at the code to see how things worked in the backstage
(you know, we physicists usually like to know how things work :) ), and
I spotted a couple of things that I thought could be done better. At
first, I was mostly afraid of doing anything, mostly because I am not a
programmer (I still don't think of myself as one) and I didn't know if
what I was doing and how I was doing it was ok. But I ended sending some
patches to the mailing list, and one thing followed another and I ended
doing a GSoC.

And here I am still, mostly without free time to try to continue where I
left everything, but mostly willing to help again with whatever I can.

My points would be more or less two: First, not to just think about
graphics and cs people as possible help, it is true that they may be the
best for the job, but other backgrounds may also help. And second, be
welcoming to new people, for some of us, the step of sending something
to the list, specially the first time, may be very stressfull, and
receiving an answer with a plain 'do this' may be a bit discouraging.

As said, just my 2 cents.
Bruno

On Mon, 2015-03-16 at 09:32 -0700, Laura Ekstrand wrote:
> That was basically my background (mechanical engineering + lots of
> OpenGL) when I started six months ago, but I have found the lack of
> mentoring to be a large roadblock.  At that time, I wrote tests, but
> there were few people willing to review them and give timely feedback.
> I was advised to go ahead and push the tests after a month, but then
> others came back weeks later with lots of late reviews after the fact.
> They were highly critical and made me feel unwelcome in the community.
> I've had more success working directly on the Mesa driver.
> 
> 
> So I'm not sure we can attract and retain these types of students.
> 
> 
> On Mon, Mar 16, 2015 at 6:23 AM, Marek Olšák <maraeo at gmail.com> wrote:
>         On Fri, Mar 13, 2015 at 6:24 PM, Laura Ekstrand
>         <laura at jlekstrand.net> wrote:
>         > We should try to steer people away from just writing Piglit
>         tests for GSoC,
>         > unless they have a specific mentor in mind and have already
>         talked to him or
>         > her.  In my experience, Piglit tests are difficult to do
>         well because each
>         > one is drastically different from the others and involves
>         cultivating a
>         > fairly deep understanding of the the OpenGL function in
>         question.
>         >
>         > A project pairing a specific extension with relevant tests
>         (like Martin and
>         > I have done with ARB_dsa) would be better as long as it's a
>         fairly specific
>         > domain of the OpenGL spec.  That way, the student can study
>         the spec for one
>         > specific set of objects or entry points and cultivate the
>         necessary
>         > understanding they need to write the related tests.
>         >
>         > A lot of the emails we've gotten from students saying "I
>         want to write 4.x
>         > Piglit tests" have been too broad/generic and would be
>         difficult for a
>         > student to master in a summer without lots of
>         mentoring/direction from the
>         > community.
>         
>         We should also take into account that there are people having
>         a degree
>         in or studying computer science with specialization in
>         computer
>         graphics or having strong knowledge of OpenGL already. Such
>         people are
>         difficult to find, but they would be very effective with very
>         little
>         (if any) mentoring. Gamedev-related forums (gamedev.net,
>         opengl.org,
>         etc.) should have a lot of talented people suited for this
>         job, but
>         none of them are probably aware of the Mesa/Piglit GSoC.
>         
>         Marek
> 
> 
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/mesa-dev