Re: Gabriel Dos Reis will be RM for the 3.2.2 release

[Gabriel Dos Reis <gdr at integrable-solutions dot net>]

Joe Buck <jbuck@synopsys.com> writes:

| I wrote:
| > | There isn't a schedule yet; it is up to Gaby to decide.
| 
| On Thu, Jan 23, 2003 at 05:47:56AM +0100, Gabriel Dos Reis wrote:
| > Well, I'd propose the following:
| > 
| >    o I would like to make a pre-release tarball on 2003-01-29.
| > 
| >    o We will have one week testing, i.e. from 2003-01-29 til 2003-02-04.
| > 
| >    o If no critical regression shows up, then we can do a release on
| >       2003-02-05.
| > 
| > Comments?
| 
| One question is what the policy should be on adding anything else to
| the 3.2 branch in the meantime.

I see no compeling reason to change the policy that has been in effect
since GCC-3.1 dot releases, which more or less is what you outlined in
your message.

|  My suggestion (and it's only a
| suggestion) is to do something like this:
| 
| - allow fixes to problems in the 3.2 branch that don't appear in 3.2.1
|   (idea: no new failures in 3.2.2 that aren't in 3.2.1, but no effort
|   to fix problems that have been around for a while that haven't already
|   been fixed)

Yes, this corresponds to the regression fixes with respect to previous
(here specifically 3.2.x) releases policy.  

| - otherwise, a fix would have to have a very strong justification, it
|   would fix some other regression, be important to multiple users or
|   fix a bug in some significant free software package, and be low risk

Right.  However, I would suggest such change not to be made 2 days
before the planed release date, to allow testing.

| I only know of one fix that might fall into the second category: Toon has
| been asking for the following fix (approved by Richard) to be applied to
| the 3.2 branch:
| 
| http://gcc.gnu.org/ml/gcc/2003-01/msg00689.html

Mark --
Since this patch has been approved by RTH, I would recommand it for
3.2 branch.

Thanks,

-- Gaby