[SciPy-dev] Scipy workflow (and not tools).
Travis E. Oliphant
Wed Feb 25 16:52:55 CST 2009
> And I will give an example that involves myself. I added a patch that
> was partly tested and not properly benchmarked to the 0.7 matlab io
> and rendered it more or less unusable for large datasets.
> I speak only for myself, but I like having people have a look at my
> code, teaching me stuff I don't know, or just checking things that I
> didn't think of myself.
> So, how about this:
> A proposal
> We set up a patch review policy. The review involves checking for and
> suggesting tests and documentation. That's the default. If you don't
> want this to happen to your code, then you ask for an opt-out.
Who do you ask? Who decides whether or not you get one? I think it's
better to have a recommended policy of review but not a requirement.
Then, a tool like coverage that shows which code has been reviewed by
more than one pair of eyes. And then let that information along with
who committed the code and who are the people "curating" a particular
sub-package guide the release manager.
More information about the Scipy-dev