Re: Mainline now closed to all changes which do not fix regressions

[Andrew Pinski <pinskia at physics dot uc dot edu>]

On Oct 12, 2005, at 9:49 AM, Adriaan van Os wrote:

> Mark Mitchell wrote:
>
> > As previously announced, here:
> >
> > http://gcc.gnu.org/ml/gcc/2005-10/msg00093.html
> >
> > the mainline is now subject to the usual release-branch rules: only
> > fixes for regressions.
>
> The development rules at <http://gcc.gnu.org/develop.html#stage3> use 
> the word "bug", not the word "regression". Should I conclude that the 
> rules have silently been changed or is this simply a sympton of a 
> complete lack of interest in bugs that are not "regressions" ?

No, we are running late for stage3 and the release manager wants the 
release
to go out.  This is not new to GCC development.  Every once in a while
regression only periods can happen to stabilize the tree.

There is __NO__ silently change in the rules for stage3.  The release 
manager
can decide that we need to stabilize the tree at any time and change 
what
kind of mode GCC mainline/branch is at.

And if you read the other part of the linked email, you would see that
patches which fix bugs could be approved but not applied until the RM
says we are opening it up for that.  And the mainline would open up
for a short time for those patches to be applied and then the mainline
would branch.  I don't see why you think this is changing the rules
at all.


> So, we have come to the point that not only old (and very old) bugs 
> are retargeted (if they were ever targeted at all), but also the new 
> "regression" bugs ? Sic transit gloria mundi ! Interestingly, one 
> could argue that a retargeted (and thus unfixed) regression bug is no 
> longer a regression bug, because it no longer "previously worked" ....

Regressions which are not bugs in the primary/secondary targets and 
primary languages
can never block a release, getting them fixed is nice but not a must.


-- Pinski