On Thursday 07 September 2006 01:48, Richard Fish wrote:
> I'm assuming of course that had the gcc team and
> others known how much things would break by stabilizing 4.1 that a
> bigger push would have been made to clean things up before-hand.

I believe the reason why gcc-4.1.1 was stabilized knowing that at least 75 
packages still needed to be stabilized in order to compile with gcc-4.1.1 was 
that release engineering didn't want to delay the release even further 
because of maintainers who didn't respond to stabilization requests in a 
timely manner.

They had to stabilize it before releasing 2006.1. Otherwise everyone who 
installed from that release would end up with a lot of broken keywords after 
the first sync.

The positive thing about this is that everyone who installs from release 
2006.1 won't have to go through the whole gcc upgrade procedure because it 
ships with 4.1.1. The cost was that some users of stable who chose to upgrade 
as soon as gcc-4.1.1 was stabilized had some extra hazzle with recompiling 
the system and those who find themselves unable to search bugzilla filed a 
lot of dupes.

The one thing that I think they could have done to make it easier was to refer 
everyone to the gcc-4.1.x stabilization tracker [1] in the announcements on 
gwn and warn that there were still a bunch of packages that needed to be 
stabilized in order to make a clean upgrade possible.

That way users would be able to make an informed decision about when to 
upgrade to gcc-4.1.1 and would have known where to look for known 
stabilization bugs. Yet the new users who install from the 2006.1 release 
wouldn't have issues with broken keywords after the first sync.

[1] http://bugs.gentoo.org/show_bug.cgi?id=140707

-- 
Bo Andresen