state of 3.2.1-pre: how far from release?

[Joe Buck <jbuck at synopsys dot com>]

We currently have 16 high-priority PRs against the 3.2 branch
(that is, excluding those PRs marked [mainline regression]).

There is no hope that the 3.2.1 release will be regression-free in the
way that we are currently defining regressions: that *some* preceding
release compiled the code correctly.  Such a criterion would require that
3.2.1 fix every issue introduced into GCC since 3.0.  Clearly we're not
going to do that, and will wind up needing 3.2.2 (unless we decide to
go straight to 3.3, which might not be a great idea considering the
15 additional regressions in the trunk).

I think it's reasonable to expect that 3.2.1 will not be worse than
3.2, and we have one regression against 3.2:

c++/8338	3.2 works, branch gives segfault in cc1plus
(the synopsis is no longer correct, it is an ICE not an infinite loop).

Still, this is for illegal code.

The following testcases have been broken in all releases from 3.0 onward:

c++/8021 c++/8372 c++/8453

I propose that we punt on those for 3.2.1.

The following testcases have been broken from 3.1 onward, and are also
broken on the branch, but are fixed (or at least don't occur) in the CVS
trunk:

c/5351 c++/8117 c++/8205

I think we should decide whether backporting is worthwhile, or if we
should just say "This issue will be fixed in 3.3".  Nothing here should
block 3.2.1.

The following testcases are broken for gcc 3.1 and later, and are still
broken in the branch and the trunk:

libstdc++/6746 c++/8036 c++/8116 c/8439 c++/8442

I don't think that any of these should block 3.2.1.

The following two PRs are for m68-elf/rtems:

target/8343
	This was narrowed down to a change Honza made, and the log says
	he's looking at it.  Any updates?
target/8314
	Here Joel was testing a patch and reassigned to Jeff Law.
	Any updates here?

Given that these are in progress, clearly fixes could go in, but again
nothing should block 3.2.1.

Finally, there are two that I am unclear about:

bootstrap/8362
	This is PowerPC specific and appears to be a blocker.

bootstrap/8146
	We still don't know what the issue is here (with building
	2.95.3).  I'd like to understand it.  But even if we do
	nothing, 3.2 has the same issue as the 3.2.1 branch, at
	least it's no worse.

My conclusion, IMHO:

3.2.1 has a vast number of bug fixes over 3.2.  If we can get 8362
fixed, I'd say ship the sucker, it is in the best interest of the
users to get the fixes out there (especially all the fixes to the
x86 backend).  The other bugs are not that important.