C. Titus Brown wrote:
On Wed, Mar 18, 2009 at 09:24:25PM +0000, Antoine Pitrou wrote: -> Rather than performance, I think some more interesting areas would be related to -> some of the standard library modules. For instance, the unittest module could -> welcome some new features (test discovery, support for skipped tests, probably -> others that I'm forgetting about). Since this is pure Python stuff not needing -> any deep experience with the interpreter's internals, it would be appropriate -> for an outsider.
interesting idea -- but I've seen too many arguments about what the *right* functionality to add to unittest would be to want to give it to a student. I think a student would probably not be willing or able to fight the battles necessary to get his/her changes into the core...
Yeah, that was my thought on that specific example as well - since there isn't anyone to say "we're doing it this way" for the unittest module the way we have with the language itself (i.e. Guido) or with other modules (e.g. Raymond for itertools, Vinay for logging), unittest improvement discussions tend to get stuck on the details :P
If an existing module has a strong owner that can say yes/no to large changes then work in that area can be a good student project (especially if the owner can serve as the mentor). But modules with only collective python-dev responsibility can be tricky to work on more from the social angle rather than the technical one.