Summer of Code is ramping up. Every year the common complaint is that not enough Python core projects get proposed by students, and of course a big reason for that is often the only encouragement we offer prospective students is a link to the PEP index.
So let's make this year different.
Accepted students are paid a total of $4500 to work for roughly 30 hours a week, 12 weeks, on their proposed project.
The challenge is finding project ideas for them that could reasonably occupy them for the entire Summer and which the results of their work can be demonstrated. They're being paid for specific projects so "Spend the Summer fixing bugs on the tracker" is a no-go, and Google has outlined that Summer of Code is about code, not documentation.
I've seen and heard that a lot of work is still needed on http://svn.python.org/view/python/trunk both during the 3.1 release cycle, optimization possible all over the place. It'd be great if those of you working closely with this can shout out some ideas, brainstorm a bit.
PSF was announced as one of the mentoring orgs today, this week before student applications are open is for students to talk to their prospective mentors and iron out the wrinkles in their plans, so there's not much time to get core project ideas together.
How about porting PIL to 3.0? There were many such requests on python-list and image-sig (including mine :))
I have ported it to the stage where its tests passes (which are far from covering all the code) and some of my own tests, there is a git repo on the image-sig that points to it. I wasn't really careful with some of the things (and I would even consider redoing some of them), but only one or two people got a copy of it so apparently people don't want/need it on python 3.0 just yet (not it alone at least).
I did a "git clone git://gpolo.ath.cx/pil-py3k.git" but it failed:
gpolo.ath.cx[0: 220.127.116.11]: errno=Connection timed out fatal: unable to connect a socket (Connection timed out) fetch-pack from 'git://gpolo.ath.cx/pil-py3k.git' failed.
Thanks for noticing that, maybe more people had this same problem then, I will consider using github or some similar service (or maybe take the chance to bazaar, or mercurial, or svn, or..).
By the way the reason I think few people checked it out is that people mostly are waiting for an "official" PIL release that is known to be stable. Did you try making your port part of the "official" PIL distribution?
I have talked with Fredrik, he said he would be running it on another test suite to check how much of it really works. But, no, I didn't really try pushing it to be integrated into the next PIL release and it also wouldn't be possible without distributing a py3k version only -- I didn't do the port with the ability to work in python 3.x and python 2.x but this can be arranged.
Maybe I'm misunderstanding you but I didn't mean to say that this version should work on both python 2.x and python 3.x. Ideally, there would be a PIL distribution for 2.x only and another one for 3.x only. The only thing is that people (myself included) will only be comfortable with the PIL for 3.x version if it comes with the blessings of Fredrik, i.e. if I were you I'd try pushing this with Fredrik. After all if all tests pass including the ones Fredrik has for himself, there should be no problem and I suppose he would be happy to have a PIL for python 3.x.
Until then I'd be happy to check out your own port, whenever you have a working repository copy please let us know.