Re: [RFA:] Fix PR target/6838 CRIS, request to revise 3.1 check-inpolicy

[Mark Mitchell <mark at codesourcery dot com>]

--On Wednesday, May 29, 2002 05:33:33 PM +0200 Hans-Peter Nilsson 
<hans-peter.nilsson@axis.com> wrote:

> Please revise check-in policy for 3.1,
> <URL:http://gcc.gnu.org/ml/gcc/2002-05/msg01212.html>.  I
> believe that port-local bug-fixes for new ports should be
> fixable by a person with approval rights without release manager
> approval.

After the first release of a port/language we do not want to introduce
regressions there.  It's true that CRIS cannot have regressed relative
to a previous release of GCC -- but it's true that it *could* regress
in GCC 3.1.1 if we blow it.

For new ports, I've been very flexible *before* the first release
because things are often in considerable flux, and if the new port
is somewhat broken, everyone will still be pleased that is there at
all.  But *after* the release, we must start to assume that there are
people depending on it.

Does that make sense?

I do think it makes sense to be a little more lenient with new ports;
there are probably fewer people depending on them and more "showstopper"
bugs; it's worth taking more risks.

The patch is OK for the branch.

Thanks,

-- 
Mark Mitchell                mark@codesourcery.com
CodeSourcery, LLC            http://www.codesourcery.com