Hello.
This year we had a record of two and a half applications (one was on a cross section of PyPy and numpy) accepted for the Google SoC program. Since it ended a couple of weeks ago, we wanted to present the results that were achieved. All three projects were completed successfully, although the rate of success varied quite a bit.
The Numpy proposal progress significantly on making numpy compatible with PyPy's CPython's extension module support, but failed to bring PyPy's numpy implementation into a usable shape (which is a somewhat ambitious goal, one might argue). The experiments done during the projects are living on the micronumpy branch.
The Fast ctypes proposal did some useful experiments on how to JIT external calls from PyPy to C, however, the actual code as of now is not very interesting and it's quite far from providing a full ctypes replacement (or equivalent).
Definitely the most successful proposal was a 64bit (x86_64) backend for PyPy's JIT. It not only includes working 64bit JIT (merged into PyPy trunk), but also a working asmgcc for x86_64 linux platform, that makes it possible to run the JIT on this architecture with our advanced garbage collectors. One can say that x64_64 is now no longer a second-class citizen for PyPy, although it definitely didn't receive as much testing as the x86 platform. Expect this to be a major selling point for the next PyPy release :-)
Cheers, fijal & the PyPy team
Awesome news.
ReplyDelete