finding strict aliasing problems

[Mike Stump]

> Date: Tue, 1 May 2001 10:38:32 +0200
> From: Jakub Jelinek <jakub@redhat.com>
> To: Joe Buck <jbuck@racerx.synopsys.com>

> Right, on the other side IMHO making no-strict-aliasing be the
> default for 3.0 again would be a bad decision - people would be
> never forced to fix broken sources so we'd stay with
> no-strict-aliasing forever.

Let me say that I think that 3.0 should ship with alias analysis
turned on.  If we do a .0 release every decade, and if we want the .0
release to have the major changes that might throw people, then,
clearly, 3.0 should turn it on, otherwise, we have to wait a decade.

And if not now, when?  WRS is another vendor that has shipped
compilers with it on by default.