This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [PATCH] Enforce GMP/MPFR version requires and fix fortran/28276,27021


Geoffrey Keating <geoffk@apple.com> writes:

> > I agree that we have a patch review problem.  I don't think it is as
> > serious now as it has been in the past, but I think it is serious.  I
> > have proposed, on IRC, some ideas which I think would help.  Either
> > way, I would strongly resist any sort of "implicitly approved" rule.
> 
> Could you post those on the mailing list?  Some of us were not
> on IRC at the time...

I believe that it would help to create a new class of maintainers:
people who are permitted to approve patches in certain areas, but are
not permitted to commit patches in those areas without approval from
some other maintainer.

Right now maintainers are always able to check in their own patches,
however ill-considered.  I believe that there are a number of people
who can and should be trusted to recognize good code in areas of the
compiler for which they are not necessarily trusted to make their own
changes.  It's easier to see the flaws in somebody else's code than it
is to see the flaws in your own code.

One argument against this is that it then becomes possible for two
maintainers to approve each others patches.  While true, I don't see
that as significantly more risky than the current situation.  It is
still important to pick trustworthy people.

Another argument against this is that it creates second class
maintainers.  But we already have that, so I don't see this as an
issue.  There are a small number of people with blanket write
privileges, most of whom rarely review patches these days.  Everybody
is second class to them.  I have heard that there is some reluctance
to approve new blanket write maintainers--indeed the most recent of
the current blanket maintainers to be approved was Bernd Schmidt in
November, 2000.

Ian


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]