On Tue, Sep 29, 2009 at 7:38 AM, Richard Fateman
<fateman at cs.berkeley.edu> wrote:
> This comment is not specifically about Andrej's change, but a comment on the
> notion that a change is OK "if no one objects" and it doesn't break the test
> suite.
>
> What would be better in the case of changes (especially where the original
> use is unclear) is a positive action -- a confirmation of correctness --.
>
> That is, someone else has looked at the proposed change, and positively
> asserts that he (or she) thinks this is a correct change.
I've thought about the possibility of requiring 2nd-party review
before committing changes. I don't think it would help.
At present we effectively have review after the fact.
A priori we have a lot of discussion of various arcane points, which
may or may not lead to actual code, but it certainly helps build a
certain groupthink which, apparently, prevents developers from
doing anything too crazy (i.e. too different from what the others
developers want).
FWIW
Robert Dodier