Jonathan, David, Aristotle:

With such smart and helpful people as you, all problems can be solved.

Looking at the test report, EU::MM was only 6.42, which is before
it could process BUILD_REQUIRES, which makes David correct in his
diagnosis.  So I'll put CONFIG_REQUIRES => {EU:MM => 6.56} into
Makefile.PL, and check that it propagates to META.yml.

CPAN on the smokebox was 1.9402, which can presumably act on the
CONFIG_REQUIRES in META.yml.  Otherwise there would seem to be a
semantic knot wherein the 6.42 EU:MM doesn't know how to process
the new CONFIG_REQUIRES so it can replace itself!

Also I'll replace LWP by LWP::UserAgent in BUILD_REQUIRES as
Aristotle suggests.

The tester is being uncharacteristically quiet on this matter,
for someone named David :-)

Thanks to all,
cmac


On Mar 25, 2010, at 3:01 PM, Aristotle Pagaltzis wrote:

* cr...@animalhead.com <cr...@animalhead.com> [2010-03-25 18:00]:
1. Change 'LWP' to 'LWP::UserAgent' in BUILD_REQUIRES

This won’t fix this problem, but you should do that anyway. Do
not mention distributions. The CPAN client will figure out which
distributions the user needs. Your job is to list the *modules*
you require – every single one of them, regardless of what the
distribution is called or whether several of them are part of the
same distribution.

Distributions can and do get split or combined.

If you list modules, you will be unaffected by that: the CPAN
client will simply figure out the right set of distributions to
install regardless of how they’ve evolved since the time when
your own code was released.

It seems we have 3 categories: config_requires, build_requires,
and requires, and it's not clear which of the last 2 applies to
something required by a test.

None of these, really. There should really be a `test_requires`.
However, in absence of that option, what remains to be said is
that both `configure_requires` and `requires` are more wrong than
`build_requires`, so that’s what you should pick for now.

Regards,
--
Aristotle Pagaltzis // <http://plasmasturm.org/>

Reply via email to