Tue Jun 1 13:59:48 CDT 2010
On Tue, Jun 1, 2010 at 6:50 AM, Charles R Harris
> Numpy/Scipy has changed from the days when there were just a few folks
> involved and the urgent need was to get some code, any code, out there. I'm
> sure many projects start that way because in beginning the idea is the
> important thing, the perfection of the implementation not so much. But as
> things progress and more people use the code, correctness becomes important.
> The numpy/numeric C code itself shows this process, with the early code
> quality being what I would classify as "undergraduate" C. That doesn't mean
> Numeric wasn't useful, obviously many people found it so or we wouldn't be
> here, but it does mean that the code wasn't easy to maintain or understand.
> Now the basic ideas have been worked out and the originators have moved on
> while at the same time the code has become more widely used, so the need
> becomes maintenance, correctness, distribution, and attracting the people to
> do those things. That requires a different sort of process.
And reliability, i.e., it is not enough to claim that the code is correct,
people (scientists whose reputations are at stake) need to be able to *rely*
on it being correct.
Perhaps it hasn't been long enough, but I note two things at this point:
0) No one has disputed that we have (and have had for some amount of time,
i.e., it didn't go into affect yesterday) a standing policy that new code
submissions are supposed to have passing tests and a Standard-compliant
docstring *before* being checked-in, and
1) No one has indicated a specific place where this (or any other standing
policy) may be found for reference.
So, I propose that we establish such a place - even if we don't presently
populate it with *anything* - so that, if we wish to discuss, e.g., whether
or not rules may be subject to individuals' "style," we can at least all
know exactly what rules we're discussing.
> SciPy-Dev mailing list
Mathematician: noun, someone who disavows certainty when their uncertainty
set is non-empty, even if that set has measure zero.
Hope: noun, that delusive spirit which escaped Pandora's jar and, with her
lies, prevents mankind from committing a general suicide. (As interpreted
by Robert Graves)
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the SciPy-Dev