On Wed, 2006-03-08 at 17:07 -0500, Bruce Momjian wrote:
> Committers go through the same approval process as non-committers
No, they don't: committers can commit changes directly to CVS, whereas
non-committers need to send them to -patches and have someone else
review and apply them. That significantly lowers the barriers to
modifying Postgres. Of course, there is still oversight by other
developers: someone else is liable to review your code once it's in the
tree, and it is considered bad practise for the non-core committers
(e.g. me) to commit major patches without sending them to -patches
first. But the fact remains that there is a significant difference in
the workflow between committers and non-committers (particularly when it
takes several weeks or months for a patch to be applied, as can
sometimes be the case).
> so it is only the physical commit action that separates committers from
> non-committers, so for us, commit privileges aren't a good indicator.
Sure they are: having the commit bit partly reflects the degree of trust
that the developer has earned based on their prior contributions. The
significance of having commit privileges depends on the project: in
Postgres it typically takes a *long* time for an individual to become a
committer, whereas other projects are more liberal about it. But that's
a matter of degree: in both cases having the commit bit infers something
about the project's trust in a contributor.
-Neil
