[SciPy-dev] Scipy workflow (and not tools).

Mike Hansen mhansen@gmail....
Tue Feb 24 16:17:00 CST 2009


On Tue, Feb 24, 2009 at 2:02 PM, Charles R Harris
<charlesr.harris@gmail.com> wrote:
> Are we adding a lot of broken code? Does the gain offset the pain? I think
> we need more folks with commit privileges and interest. In the short term I
> would propose the following.

Many would argue that any untested code is broken code and will become
an maintenance burden in the future.

> 1) Additions get posted on the mailing list for comment before commit. I'll
> bet few knew of the additions to optimize.

Does this really provide motivation for people to look at code?

> 2) We look for folks with patches in trac and consider giving them commit
> privileges to fix things up.

Ideally, you don't want anything committed until it's "fixed up".
Using SVN makes this a bit more difficult to do.

> 3) We put together a list of needed tests. Then we will see how serious folks are about writing tests.

It's definitely important to know what you're actually testing and
what you're not.  Also, being able to do so in an automated way is
important.
http://ivory.idyll.org/blog/feb-09/people-who-dont-use-code-coverage-are-idiots.html

--Mike


More information about the Scipy-dev mailing list