Re: gcc 3.0 branch report for 34 embedded targets

[Joe Buck <jbuck at synopsys dot COM>]

Benjamin Kosnik wrote:
> > >From what I understand, the arm problems have been fixed in current CVS.
> > 
> > I would find this more useful if this testing was against current CVS,
> > ie gcc-3.1, than against the gcc-3.0 branch. At this point, it's
> > easier to fix issues in current CVS than on the branch.
> > 
> > Do you have any plans to test cross compilers for the gcc-3.1 branch?
> > If so, can you detail them?

Joel Sherrill writes:
> What do you mean?  

He means that we already know that gcc-3.0.x sucks for embedded targets,
and we would like to know whether gcc-3.1 will suck less.  We have some
reasons to believe that at least on some of them you should see better
results.  If not, it would be good information.

> My goal is to be able to provide RTEMS prebuilt binaries for all *-rtems
> targets.  I periodically check out and test the more full set of 
> embedded targets because there seemed to be no one who reported on the
> state of the cross targets at the same time.  I have simulators set
> up for all the targets that have them in gdb and can run the GCC
> test suites on many of them.  

Great.  When you have time, could you re-run your tests on the current
CVS and see if we do better?

> I will repeat this on the current CVS if it will encourage someone
> to fix things. :)

Yes, please.