On Fri, Jun 15, 2007 at 03:00:21PM -0400, John Scoles wrote:
> testing with
> ftp://ftp.funet.fi/pub/languages/perl/CPAN/authors/Tim_Bunce/DBI-1.56_92.tar.gz
> got some differnent results this time
> Makefile.PL and nmake when whitout a hitch
> I have attaced the nmake test results
> seems it is the 40profile.t that is dieing now.
> t/40profile...............NOK 14
> # Failed test 'should return ref to leaf node'
> # in t/40profile.t at line 106.
> t/40profile...............NOK 15# Failed test 'should return ref to leaf
> node'
> # in t/40profile.t at line 107.
> # got: ''
> # expected: 'ARRAY'
> t/40profile...............NOK 16# Failed test 'should return ref to
> correct leaf node
> # in t/40profile.t at line 111.
> # got: undef
> # expected: 'ARRAY(0x1bfa680)'
> t/40profile...............ok 44/58# Looks like you failed 3 tests of 58.
> t/40profile...............dubious
> Test returned status 3 (wstat 768, 0x300)
That's odd. There's nothing windows specific about that code.
What version of perl are you using here?
(t/40profile.t is set up to skip all the tests if $] <= 5.008001.)
Try it with the attached patch and let me know what it says.
Thanks John.
Anyone else seeing this behaviour?
Tim.
--- DBI.xs (revision 9655)
+++ DBI.xs (working copy)
@@ -4172,6 +4172,7 @@
SvROK(method) ? SvRV(method) : method,
t1, t2
);
+ warn("dbi_profile %d, gimme=%d", SvOK(leaf), GIMME_V);
(void)cv;
if (GIMME_V == G_VOID)
ST(0) = &sv_undef; /* skip sv_mortalcopy if not needed */
