[committed] Change default for enable_hash_synchronization_default on PA-RISC

Michael Koch konqueror@gmx.de
Fri Jun 23 04:30:00 GMT 2006


On Thu, Jun 22, 2006 at 08:20:21PM -0400, John David Anglin wrote:
> > On Wed, Jun 14, 2006 at 09:01:33AM -0400, John David Anglin wrote:
> > > > >>>>> "Michael" == Michael Koch <konqueror@gmx.de> writes:
> > > > 
> > > > Michael> I would really welcome a backport tp 4.1 branch. This would help me a
> > > > Michael> lot to make GCJ reliable on Debian hppa.
> > > > 
> > > > Note that this change affects the ABI.  That means if you do a
> > > > backport then users have to recompile their programs -- at least,
> > > > programs using the "C++" ABI.
> > > 
> > > There's another ABI change in 4.1 (fix for passing of complex floats),
> > > so this additional change may not present much of a problem.  As a result,
> > > I don't think Debian is using 4.1 to any great extent yet.  I don't have
> > > a good assessment of how many others might be using GCJ on hppa and how
> > > much these users would benefit.
> > 
> > Well, we as Debian Java maintainers decided to use GCJ 4.1 as the
> > default runtime for Debian (for building and running
> > applications/libraries). I would call this a great extent. We build and
> > run e.g. Eclipse with it by default.
> 
> I've installed the change below.  The 'hppa-*' change fixes a typo.
> I also plan to install a backport of PA libffi changes from the trunk.
> With these two changes, I've had three clean builds and checks of
> libjava on the 4.1 branch with hppa-unknown-linux-gnu.

Thanks,
Michael
-- 
http://www.worldforge.org/



More information about the Gcc-patches mailing list