[Numpy-discussion] let's use patch review

[David Cournapeau]

David Huard wrote:
>
> There is about 5 commits/day, I don't think it's a good idea to wait 
> for a vote on each one of them.

There is definitely a balance to find, and I am not convinced it would 
work well with subversion (it really makes sense to have those review 
with merge request, not per commit). For example, in scons, they have a 
fairly heavy review process which IMO prevents it from getting more 
contribution. In bzr, it works pretty well (they use a gateway system), 
but most main developers are paid for it.

Having a somewhat official review process would also help solving one of 
my problem with trac: when someone sends a patch on trac, we don't know 
it, we have to look for it, and some of them are lost/duplicated. 
Requesting unit tests for new contributors is too much, I think, I hate 
it for other projects where I am less involved than numpy/scipy. But say 
I have 20 minutes to spend on reviewing patches: with a system which a 
list of available patches, it would be easy to do so. Maybe it is 
possible with trac and I just missed it.

cheers,

David