On 3/8/07, Tels <[EMAIL PROTECTED]> wrote:
> CPAN::Reporter tries to find prerequisites in an environment similar
> to what test scripts would find (assuming they don't mess with @INC).
I think my testscripts "mess" with @INC, but I am not sure if this is a
problem (I generally test modules only manually, not with in some
test-framework)
What I mean is that CPAN::Reporter kicks off a subprocess to gather
information on prereqs. This way, it gets the exact same PERL5LIB that
CPAN.pm sets for test programs that get kicked off by "make test".
> If this is an old installation, yet CPAN::Reporter saw 1.80, then
> there may be a bug in CPAN::Reporter too. Please let me know how this
> turns out.
I have no idea how I could even replicate this except setting up complete
testing-framework, which I lack the time to do.
I should have been clearer -- please just copy me if you and the
tester correspond privately and figure out the problem.
One thing that might be special is that Math::BigInt is a core-module, and
so needs to overwrite a already existing core module upon install. If you
install it into the "site" dir instead of "perl", the old core would still
get loaded first.
If that's the case CPAN::Reporter should pick that up in the prereq
check -- unless CPAN.pm has prepended a successful test to the
PERL5LIB and from the original test report, PERL5LIB was empty.
And v1.80 of Math::BigInt now uses Module::Install, so this might have
something to do with that, but I am not sure.
Always possible. Module::Install is still a little quirky in some areas. ;-)
David
