#3 make failure for GNAT

[George Chapman (LMCO)]

Makefile in the GNAT directory has an incorrect value
for the CPUID variable.
Attached is a diff between the original Makefile and
the corrected one.

[George Chapman (LMCO)]

diff of original Makefile with corrected one.

[George Chapman (LMCO)]

Logged In: YES
user_id=1286538

Version 20051222.

[Simon Wright]

Logged In: YES
user_id=917025

Is this a 'stopper' for the user?

Is there a way we can automate this on OS2, Windows? (as we already do for
Linux, Darwin -- NB Darwin doesn't understand the GNU long options, eg
uname --machine).

Would it be 'good enough' to just say for example x86?

As you can see, Darwin wants to call the CPUID 'powerpc' (don't know what
happens on Intel Macs), would it be good to accept the native view here? (I
suppose there are reasons downstream in the Makefiles etc).

[George Chapman (LMCO)]

Logged In: YES
user_id=1286538

Not a stopper for the user.
Can be automated for Windows but would take some effort on
my part to figure it out.
It would be good enough for Windows to pass CPUID from an
environment variable (i.o.w. have user set CPUID); if
CPUID is not set, then default to lowest common
denominator (i386 ?).
native view - no opinion on PowerPC because I won't try a
PowerPC VxWorks cross-build for another few weeks; don't
have a Mac PowerPC or Intel (work & house are
x86/Windows+VxWorks & Linux).