Re: Beyond GCC 3.0

[Gerald Pfeifer <pfeifer at dbai dot tuwien dot ac dot at>]

On Wed, 27 Jun 2001, Mark Mitchell wrote:
>  - Patches that cause regressions, even on the mainline, must
>    be in the process of being fixed within 48 hours, or else can be
>    reverted by anyone with global write privileges, if they think
>    that's best.

I believe that ought to be "by anyone with write privilege to that
part of GCC" instead?

And, like others, I'd make the actual amount of time somwhat dependent on
the "severity" of the problem. That is, for an ia32 bootstrap failure,
revert the bug rather sooner than later, while for PDP11 it's okay to wait
three or four days.

>    Months 1,2: "Free for All"
>    --------------------------
>
>    Months 3,4: "Hack On"
>    ---------------------

I believe the borders between these two categories are not that strict
(and that's certainly fine so). Perhaps we could accept more risky stuff
during months 1-3 and become increasingly careful at the end of that
period.

>    Months 5,6: "Stabilize"
>    -----------------------

That's probably the single most important part of your scenario: The
branch should happen after much more of stabilization than we had for GCC
3.0 (where, for example, we even had to move new features, most notably EH
and libstdc++, on the release branch).

>    Note that this time will immediately follow a "Stabilize" period
>    on the mainline.  Therefore, it is reasonable to expect that we
>    are going to be in decent shape by the time we branch -- we will
>    have just spent two months fixing bugs.

Excellent.

One thing that also went wrong with our past procedure is that the HEAD
immensely detoriorated during the release process as many (like me) just
do not have time and resources to closely monitor and test two branches,
immediately notifying those -- involuntarily -- breaking something.

Gerald
-- 
Gerald "Jerry" pfeifer@dbai.tuwien.ac.at http://www.dbai.tuwien.ac.at/~pfeifer/