On Fri, May 10, 2002 at 01:57:44PM -0500, Randy Kobes wrote:
> On Fri, 10 May 2002, Automated Perl Test Account wrote:
> > This is an error report generated automatically by CPANPLUS.
> > Below is the error stack during 'make test':
> > PERL_DL_NONLAZY=1 /usr/bin/perl -Iblib/arch -Iblib/lib
> > -I/usr/lib/perl5/5.6.1/i386-linux -I/usr/lib/perl5/5.6.1
> > test.pl 1..1 Can't locate Apache/AuthCookie.pm in @INC
> [ ... ]
> Although the existence of required modules can be checked for
> within Makefile.PL with the use of PREREQ_PM, not everyone uses
> this, or some do it in different ways, so it seems extreme to
> give this a failure based on Apache::AuthCookie not being
> previously installed.

Yes, I agree a FAIL seems a little bit extreme. However, as PREREQ_PM
is the de facto way to specify dependencies, it would also be desirable
to notice the author, perhaps with a more encouraging message, e.g.
"hey, it looks like a PREREQ_PM would be nice. why not do that?".

IIRC, that was also one of Michael Schwern's original ideas in CPANTS.

> Perhaps CPANPLUS should be configured to automatically send
> only PASS reports, with the other grades needing human
> inspection before being sent?

CPANPLUS currently only send FAIL reports that failed during the
'make test' stage -- i.e. when the Makefile.PL thinks it's okay
to install this module, and after 'make' successfully executed without
problems. I think it's safe to say that, if the module has entered
that stage, it's unlikely that a 'make test' error would be a
false-negative.

I'll try to find volunteers to hack the CPANPLUS::Internals::Report
to add the aforementioned "Can't locate" error detection, and methinks
it will be included in 0.04. But whether if it warrants a FAIL is open
to debate. Any thoughts?

Thanks,
/Autrijus/

Attachment: msg33288/pgp00000.pgp
Description: PGP signature

Reply via email to