[Python-Dev] Summer of Code: call for ideas and mentors

[Nick Coghlan]

On Fri, Mar 4, 2011 at 6:35 PM, "Martin v. Löwis" <martin at v.loewis.de> wrote:
> Google Summer of Code is coming up again, and we will again
> be participating. Arc Riley will setup infrastructure later
> today, and we need to start thinking about possible projects.
>
> Traditionally, people (students and other projects) have been willing
> to give the Python Core the highest attention in SoC, as improvements
> to Python will help the whole community. Also traditionally, there has
> been a shortage of project ideas and mentors for Python Core. My guess
> is that the project shortage stems from the assumption that many of the
> projects are so tricky that you can't give them to a newcomer. I found
> this view both confirmed and rebutted in the past - it all depends on
> which students we can attract.
>
> The shortage of mentors is probably more inherent, and reflects the
> shortage of volunteers for other tasks. As a long-time mentor, I'd
> like to encourage all of you to consider mentoring this year. GSoC/PSF
> will have a strict rule "one student per mentor", co-mentorship
> is encouraged.
>
> Feel free to discuss Python-Core-GSoC here or on the GSoC mailing
> list(s) (soon to appear).

Are proof-of-concept projects acceptable as GSoC projects? I've long
had the belief that Python's import semantics could be bundled up into
an "import engine" object to get away from the process globals in the
sys module (i.e. modules, path, meta_path, path_hooks,
path_importer_cache) as well as the global import lock. The "default
engine" could then be implemented as a subclass that replaced the
essential attributes with properties that read from and wrote through
to the standard locations.

Experimenting with this idea became significantly more feasible since
Brett wrote importlib, but would still require a strong understanding
of Python's import system. I suspect even a proof of concept that was
tested against just filesystem imports and zipimport would prove quite
tricky.

Once the concept has been proven... I'm sure we could figure out
*something* useful to do with the idea. It would depend on the details
of what actually turns out to be feasible.

Cheers,
Nick.

-- 
Nick Coghlan   |   ncoghlan at gmail.com   |   Brisbane, Australia