Re: 5.9.1 release now integrated into ponie

On Wed, Apr 07, 2004 at 04:10:25PM +0100, Nicholas Clark wrote:
> ../lib/Math/BigInt/t/bigints.t  255 65280    51  102 200.00%  1-51

This turned out to be me not understanding CVS - I turns out that a vendor
branch update does not cause old vendor files to be deleted (even when never
edited)
So the old copy of a .pm was being found first. Now gone.

> ../lib/warnings.t                           503    1   0.20%  224

This is this test:

# doio.c [win32_execvp]
BEGIN {
    if ($^O eq 'MacOS') {
	print <<EOM;
SKIPPED
# no exec on Mac OS
EOM
	exit;
    }
}
use warnings 'exec' ;
exec $^X, "-e0" ;


and it is failing with ENOTSUP. Really. I checked with ktrace:

 15981 perl     CALL  sigaction(0x8,0xbffff430,0xbffff500)
 15981 perl     RET   sigaction 0
 15981 perl     CALL  execve(0x1b0f3c0,0x1b0b7f0,0x1b0a340)
 15981 perl     RET   execve -1 errno 45 Operation not supported
 15981 perl     CALL  stat(0x1b0f3c0,0xbffff410)
 15981 perl     NAMI  "./perl"
 15981 perl     RET   stat 0
 15981 perl     CALL  sigaction(0x8,0xbffff440,0)
 15981 perl     RET   sigaction 0
 15981 perl     CALL  write(0x2,0x2021600,0x41)
 15981 perl     GIO   fd 2 wrote 65 bytes
       "Can't exec "./perl": Operation not supported at tmp0002 line 12.
       "
 15981 perl     RET   write 65/0x41


(see doio.c, around line 1589)

Of course, this can't be happening, as the OS X man page doesn't list ENOTSUP
as a possible return from execve. Hence I don't know where to look next.
(Darwin source code....?)

Nicholas Clark