GCC 3.3 ICE policy

[Mark Mitchell]

On Tue, 2003-06-24 at 11:08, jbuck@synopsys.com wrote:
> 
> 
> 
> On Tue, Jun 24, 2003 at 09:18:54AM -0700, Mark Mitchell wrote:
> > > > If a PR meets the following criteria:
> > > > - It is an ICE that occurs after a valid error message
> > > > - It occurs on the 3.3 branch
> > > > - It is already fixed on the mainline
> > > > I am going to mark the PR as WONTFIX.
> 
> On Tue, 2003-06-24 at 09:40, Joe Buck wrote:
> > > Is really the correct way to use Bugzilla?  After all, we DO plan to
> > > fix the bug, in fact we already have.  3.4 will contain the bug fix.
> 
> On Tue, Jun 24, 2003 at 10:23:22AM -0700, Mark Mitchell wrote:
> > Well, sort of.  We often haven't really so much fixed that particular
> > bug, as we have (for example) written a new parser which doesn't have
> > the old problem.  It's not like there's a patch to point people at to
> > apply to 3.3 that will fix the problem.
> 
> The folks who filed the bugs don't care if we fix them by means of a
> patch, or by rewriting the affected code entirely.

No -- but people who want to pull patches out of CVS to backport do
care.

> If I were the bug-filer, I would interpret "WONTFIX" as "you will have to
> live with this bug forever".  That's not the message we want to send.

You'd have gotten an email that said "this is a 3.3-only ICE already
fixed on the mainline."

I see your point, but I still feel like WONTFIX is more accurate.

However, I'm not going to worry about this issue until some kind of
consensus arises.  I just don't see this as a critical issue.  When
there's at least one more opinion on this topic, I'll go with the
majority opinion.

-- 
Mark Mitchell
CodeSourcery, LLC
mark@codesourcery.com