Re: Small update to reversed_comparison_code

[Daniel Berlin <dberlin at redhat dot com>]

Daniel Berlin <dberlin@redhat.com> writes:

> Mark Mitchell <mark@codesourcery.com> writes:
> 
> > >>>>> "Joseph" == Joseph S Myers <jsm28@cam.ac.uk> writes:
> > 
> >   Robert Dewer volunteered for ACT
> >   <URL:http://gcc.gnu.org/ml/gcc/2000-11/msg00492.html> to provide a machine
> >   for a test-before-checkin system.  Perhaps they could provide such a
> > 
> > Yup.  Robert and I are in cahoots, here, actually.  We've talked about
> > this amongst ourselves for a while.  I know how to get a few machines,
> > including some from major UNIX vendors -- but we need more than a few,
> > sadly...
> > 
> 
> Is it just me, or is this starting to sound like a distributed CPU
> time project.
> 
> GCCTesting@Home.
> 
Oh, and I realize this is not a parallelizable task, but it doesn't
have to be, unless bootstraps take absurdly long times on some of the
targets we want to test.

I'm suggesting we just automate a way of doing the bootstrap,
reporting the results, etc, so that with a bit of marketing to users
likely to have the machine types we want, we might actually convince
some to let us use their spare machine cycles to do bootstrapping.

You are, IMHO, far more likely to get people to let you do this than to try
to find one of every machine type to put into an automated tester of
some sort that needs attention, or whatever.

You could at least get most gcc developers to let you use their idle
cycles for doing this, I would imagine.

Just a far out thought.

--Dan