Re: Value for os.arch on Intel

[Mark Wielaard]

Hi,

On Fri, 2005-03-04 at 11:52 +0000, Robert Lougher wrote:
> I notice that gij also uses uname, and gives i686 on an Athlon (at
> least gij 3.4.2 does).  I think it's a "no-brainer"  -- if
> applications depend on it, we need JDK behaviour and I'm changing
> JamVM.  Any opinions?

I think native code depending on this is broken since these values are
not clearly defined for all platform/architecture type. And we should
encourage developers to fix such code/build systems. (What if someone
wants to compile such code on ia64, arm/xscale, mingw, sh-3/4, etc.)

That said, this is what libgcj recently did:
http://gcc.gnu.org/ml/java-patches/2005-q1/msg00382.html

Note that for the various windows platforms there doesn't seem to be any
consensus at the moment:
http://gcc.gnu.org/ml/java-patches/2003-q1/msg00445.html

Cheers,

Mark

signature.asc
Description: This is a digitally signed message part