Patch: embedded target fixes

[Tom Tromey]

>>>>> "Anthony" == Anthony Green <green@redhat.com> writes:

Anthony> 	* configure.host: Add support for xscale-elf embedded
Anthony> target.

I think this is fine.

Anthony> 	* include/config.h.in: Rebuilt.
Anthony> 	* acconfig.h: Add DISABLE_MAIN_ARGS.
Anthony> 	* prims.cc (_Jv_RunMain): Use DISABLE_MAIN_ARGS.
Anthony> 	* configure: Rebuilt.
Anthony> 	* configure.in: Add --disable-main-args option.

What are the symptoms of this problem?
If argv==NULL, couldn't we just test for that at runtime and avoid
a configure option:

    _Jv_ThisExecutable (argv == NULL ? "[Embedded App]" : argv[0]);

?

Anthony> 	* java/lang/natSystem.cc (init_properties): Don't use
Anthony> 	_Jv_Environment_Properties if DISABLE_GETENV_PROPERTIES is
Anthony> 	defined.

Definitely ok.

Anthony> 	* java/net/natPlainDatagramSocketImpl.cc (_Jv_bind): Don't build
Anthony> 	when DISABLE_JAVA_NET.
Anthony> 	* java/net/natPlainSocketImpl.cc (_Jv_bind, _Jv_connect,
Anthony> 	_Jv_accept): Don't build when DISABLE_JAVA_NET.

If a net-less build works with this, it is fine.

Anthony> 	* include/config.h.in: Rebuilt.
Anthony> 	* acconfig.h (NO_TZNAME): New define.
Anthony> 	* configure: Rebuilt.
Anthony> 	* configure.in (NO_TZNAME): Define for newlib target.

I think a ChangeLog entry is missing for the natSystem.cc change.
Will this patch really work?  It looks to me like we use tzinfo[1]
unconditionally as an argument to strcmp.  I think this will fail
since in the patch this is NULL.

Tom