On Wed, 3 Aug 2005 17:12:04 -0400
Michael <macallan18@earthlink.net> wrote:

> Hello,
> 
> > > May be it's 64-bit issue, I'm gonna get 32-bit userland up and
> > > running on the second hard disk and see if the problem persists.
> > 
> > It's certainly no endianness problem - my PowerPC boxes don't have it.
> > Probably just another case of someone assuming that sizeof(long)==4.
> 
> I've found the problem - it's a lot further down than I expected. One
> single source file in libgobject treats enum values as int while all the
> rest treats them as long. What happens is that all enums passing the
> marshaller are 'converted' to 0 or garbage.
> It will hopefully be fixed in the next glib2 release and I've sent a
> patch to the package maintainer. If you're impatient some sort of patch
> can be found in
> ftp://ftp.netbsd.org/pub/NetBSD/misc/macallan/sparc64/gvaluetransform.patch
> if it doesn't apply right away it should at least tell you what to do.
> 
> have fun
> Michael
> 

Cheers for that dude. There is so much software that breaks on 64-bit systems.
I still play around with sparc64, but my main multimedia desktop machine is
i386. Whatever they say, but 64-bit revolution won't happen just yet, people
need to be educated on how to write 64-bit clean code :-)