[SciPy-dev] Scipy workflow (and not tools).
Travis E. Oliphant
Wed Feb 25 16:44:25 CST 2009
> for dead code. Two examples for stats related functions: The recent
> removal of var and mean from scipy stats broke several functions
> that didn't have test coverage and so didn't show up in the tests.
> The second case is the recent addition of curvefit where the
> documentation didn't correspond to what was actually calculated.
Yes, and this is a perfect example of "people who care" catch the
problem even after commit and it gets fixed (extremely quickly) ---
especially with this kind of documentation problem. I don't see how a
formal review process would have helped in this case given that I
probably would not have spent time on it had I had to jump through that
And, the parameter estimation was working fine, but the error estimate
was not (at least for a day).
> In both cases the review and corrections happened after the commit,
> since I keep an eye on any stats related commits. Without the
> review we might get misleading (or incorrect) numbers and broken
> code. And I've seen a lot of both in stats.
This is exactly what we need. More people like Josef looking at
check-ins and offering suggestions / assistance that corresponds to
their area of interest / expertise. I was very grateful for the
assistance (and the unit-tests based on NIST data).
I don't think we need any formal rules except the natural ones. We
definitely need better tools to "track which code has been reviewed",
check builds on multiple platforms, merge development branches, etc.,
etc. We also need more cycles to spend on SciPy.
More information about the Scipy-dev