#define CvROOT(sv)      ((XPVCV*)SvANY(sv))->xcv_root_u.xcv_root
#define CvXSUB(sv)      ((XPVCV*)SvANY(sv))->xcv_root_u.xcv_xsub
 
#define CvOUTSIDE(sv)   ((XPVCV*)SvANY(sv))->xcv_outside
[download]

Looks like they both operate on the same thing (the CV*) to me

Yes - when you descend to that level they become very similar, indeed :-)

So the question is why does Perl and your function use different tests to determine the same thing?

Good question. My approach won't work on perl-5.6.2, so perhaps history and portability might be part of the explanation. But there could, of course, be other reasons. (I couldn't even get correct results on perl-5.6.2 with CvROOT and CvXSUB.)

Cheers,
Rob

Yes - when you descend to that level they become very similar, indeed :-)

At a high level:

	validate $x	defined &$x	ref $x && defined &$x
code ref	returns result	returns result	returns result
sub name	dies	returns result	returns false
other refs	dies	dies	dies
undefined	?	stringified	returns false

I couldn't even get correct results on perl-5.6.2 with CvROOT and CvXSUB

That's odd, because defined and sub calls use CvROOT(cv) || CvXSUB(cv) in 5.6.2 as well.