[Numpy-discussion] GSOC 2010
Hi All, I don't feel that numpy/scipy did as well in GSOC 2009 as it could have. I think this was mostly due to lack of preparation on our part, we weren't ready when the students started showing up on the lists. So I would like to put together a selection of suitable projects and corresponding mentors that we could put on the wiki somewhere and advertise. Just to start things off, here are two things that come to mind. - Python 3k transition. I think it is time to start looking at this seriously. - Best of breed special functions in cython. These could be part of a separate numpy extras package where code is restricted to C, Cython, and Python. Thoughts? Chuck ___ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion
Re: [Numpy-discussion] GSOC 2010
On Wed, Oct 21, 2009 at 3:02 PM, Charles R Harris charlesr.har...@gmail.com wrote: Hi All, I don't feel that numpy/scipy did as well in GSOC 2009 as it could have. I think this was mostly due to lack of preparation on our part, we weren't ready when the students started showing up on the lists. So I would like to put together a selection of suitable projects and corresponding mentors that we could put on the wiki somewhere and advertise. Just to start things off, here are two things that come to mind. Python 3k transition. I think it is time to start looking at this seriously. Best of breed special functions in cython. These could be part of a separate numpy extras package where code is restricted to C, Cython, and Python. Thoughts? for scipy: more stats, gsoc2009 went very well. Josef Chuck ___ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion ___ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion
Re: [Numpy-discussion] GSOC 2010
On Wed, Oct 21, 2009 at 1:11 PM, josef.p...@gmail.com wrote: On Wed, Oct 21, 2009 at 3:02 PM, Charles R Harris charlesr.har...@gmail.com wrote: Hi All, I don't feel that numpy/scipy did as well in GSOC 2009 as it could have. I think this was mostly due to lack of preparation on our part, we weren't ready when the students started showing up on the lists. So I would like to put together a selection of suitable projects and corresponding mentors that we could put on the wiki somewhere and advertise. Just to start things off, here are two things that come to mind. Python 3k transition. I think it is time to start looking at this seriously. Best of breed special functions in cython. These could be part of a separate numpy extras package where code is restricted to C, Cython, and Python. Thoughts? for scipy: more stats, gsoc2009 went very well. Yes, it seems so. I had the impression that planning for that project was undertaken pretty early on with the involvement of Skipper. What exactly *was* the history of that project and what can we learn from it? Chuck ___ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion
Re: [Numpy-discussion] GSOC 2010
On 21-Oct-09, at 3:02 PM, Charles R Harris wrote: • Best of breed special functions in cython. These could be part of a separate numpy extras package where code is restricted to C, Cython, and Python. I think a lot of SciPy could be usefully brought over to Cython, as well (not all the C code, but some of it). Having Cython do the wrapping should reduce the burden in the eventual Py3k transition. David ___ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion
Re: [Numpy-discussion] GSOC 2010
On Wed, Oct 21, 2009 at 3:23 PM, Charles R Harris charlesr.har...@gmail.com wrote: On Wed, Oct 21, 2009 at 1:11 PM, josef.p...@gmail.com wrote: On Wed, Oct 21, 2009 at 3:02 PM, Charles R Harris charlesr.har...@gmail.com wrote: Hi All, I don't feel that numpy/scipy did as well in GSOC 2009 as it could have. I think this was mostly due to lack of preparation on our part, we weren't ready when the students started showing up on the lists. So I would like to put together a selection of suitable projects and corresponding mentors that we could put on the wiki somewhere and advertise. Just to start things off, here are two things that come to mind. Python 3k transition. I think it is time to start looking at this seriously. Best of breed special functions in cython. These could be part of a separate numpy extras package where code is restricted to C, Cython, and Python. Thoughts? for scipy: more stats, gsoc2009 went very well. Yes, it seems so. I had the impression that planning for that project was undertaken pretty early on with the involvement of Skipper. What exactly *was* the history of that project and what can we learn from it? Short(-ish) version of some general thoughts from my end: GSoC was brought to my attention as a fruitful endeavor (and it definitely was!). There was a list of potential topics posted on SciPy SoC mentoring page, and I just kind of went through all of them to see where the most value-add would be (both ways from me to the SciPy project and from the SciPy project to my studies/work). So that list of topics was the main driving force, and I'm glad we're starting to push for ideas now (I have a few ideas of my own motivated mostly by needs of stats/statistical modeling, but I need some more time to think). However, we obviously should be open to new ideas from students coming to the project. Another thing is the importance of the application process. The thing that pushed me was reading about other successful applicants for SoC in general (there is a lot of really good advice and write-ups out there). It is a very competitive program, so your proposal needs to be very, very well thought out. That includes drafts of proposals with feedback from the community and mentors well before the official application process even starts, so the earlier that's taken care of, the better. Beyond that, students should know what's expected of them coming into the program (what development tools they need to be familiar with, numpy/scipy standards, familiarization with the code base), and what's expected of the end product (high quality code, test driven development, etc.). I also can't stress enough how helpful it was to have Alan and Josef as mentors, as well as the availability to use the MLs for more general questions. Obviously, the level of engagement of the mentor is going to depend on the project and the student, but I for one couldn't have learned as much as I did nor gotten as far as we did without their help. If these comments are seen as helpful, I can try to work on some more detailed ones/links to detailed ones, as I think this would be beneficial to establish as something to look forward to. The availability of this program (Thank you, Google) allows significant strides in development to be made each summer and that should not be overlooked (I don't think it is). Cheers, Skipper ___ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion
Re: [Numpy-discussion] GSOC 2010
On Wed, Oct 21, 2009 at 3:23 PM, Charles R Harris charlesr.har...@gmail.com wrote: On Wed, Oct 21, 2009 at 1:11 PM, josef.p...@gmail.com wrote: On Wed, Oct 21, 2009 at 3:02 PM, Charles R Harris charlesr.har...@gmail.com wrote: Hi All, I don't feel that numpy/scipy did as well in GSOC 2009 as it could have. I think this was mostly due to lack of preparation on our part, we weren't ready when the students started showing up on the lists. So I would like to put together a selection of suitable projects and corresponding mentors that we could put on the wiki somewhere and advertise. Just to start things off, here are two things that come to mind. Python 3k transition. I think it is time to start looking at this seriously. Best of breed special functions in cython. These could be part of a separate numpy extras package where code is restricted to C, Cython, and Python. Thoughts? for scipy: more stats, gsoc2009 went very well. Yes, it seems so. I had the impression that planning for that project was undertaken pretty early on with the involvement of Skipper. What exactly *was* the history of that project and what can we learn from it? Skipper started early in the preparation, and with the help of Allan and me had a pretty concrete proposal. Because of final exams, the actual work on statsmodels started a bit late. From my perspective a few issues that helped: Skipper, Alan and I have the same background (in econometrics), so I knew roughly what knowledge I could expect. Skipper was willing and able to work his way through several textbooks for the models that he, and I, didn't know much (or anything) about. Cleaning up stats.models was a relatively well defined project, with relatively easy to define goals. I kept reminding him about writing tests, and to verify results with other packages, so that we knew when we had a model correctly cleaned up. Skipper spend a lot of time on this. For most parts, I worked on the code in parallel with him, checking on his progress, looking at the problems we had with matching the results of the other statistical packages, finding bugs and writing some draft code. During July, August we had almost daily long email threads. I think, this helped a lot, so that Skipper didn't get stuck or sidetracked, and that I was able to keep up with the changes (and learn some of the statistical background). Josef Chuck ___ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion ___ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion
Re: [Numpy-discussion] GSOC 2010
On Wed, Oct 21, 2009 at 12:02 PM, Charles R Harris charlesr.har...@gmail.com wrote: I don't feel that numpy/scipy did as well in GSOC 2009 as it could have. I'd be curious to hear why you felt that numpy/scipy didn't do as well this year. We had more projects than any other year and I think that most of the code ended being used. It could be that the work done wasn't publicized enough or that the most of the contributions end up contributed to related projects like in a scikit or (hopefully soon to be merged work) in cython. At any rate, I'd be curious to hear more about your concerns so that they we don't repeat them next year (assuming the program is run again next year). I think this was mostly due to lack of preparation on our part, we weren't ready when the students started showing up on the lists. So I would like to put together a selection of suitable projects and corresponding mentors that we could put on the wiki somewhere and advertise. Just to start things off, here are two things that come to mind. Regardless, better preparation would be a huge help. Having detailed lists of summer projects will be useful even if the SoC program doesn't get approved for next year. Python 3k transition. I think it is time to start looking at this seriously. Best of breed special functions in cython. These could be part of a separate numpy extras package where code is restricted to C, Cython, and Python. Both of these ideas sounds very interesting. Personally, I would like to see ideas like these make there way into fully fleshed out NEPs: http://projects.scipy.org/numpy/browser/trunk/doc/neps -- Jarrod Millman Helen Wills Neuroscience Institute 10 Giannini Hall, UC Berkeley http://cirl.berkeley.edu/ ___ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion
Re: [Numpy-discussion] GSOC 2010
On 10/21/2009 3:23 PM, Charles R Harris wrote: What exactly *was* the history of that project and what can we learn from it? Imo, what really drove this project forward, is that Skipper was able to interact regularly with someone else who was actively using and developing on the code base (i.e., Josef). While I am confident Skipper would have made a worthwhile contribution without this, I think he would agree that he both learned more and was more productive because he was able to interact with Josef. One other thing that was important was focus: Skipper (and Josef) focused in on making sure an important but doable (summer is very short!) piece of the stats code was refactored, extended, documented, and tested. Alan Isaac PSI do not mean to diminish the importance of the feedback kindly provided by others. ___ NumPy-Discussion mailing list NumPy-Discussion@scipy.org http://mail.scipy.org/mailman/listinfo/numpy-discussion
