On 9/27/07, Graham TerMarsch <[EMAIL PROTECTED]> wrote: > I can actually verify that CPANPLUS doesn't send automated "failure" reports > when no Makefile is generated; I tested that locally myself this morning > after reading your previous message.
That's good news. > That's what I'm finding too; the vast majority of CPAN Testers reports that I > see are from CPANPLUS. Most of the smoke testers are CPANPLUS based. I believe that Andreas Koenig and David Cantrell have each hacked up something using CPAN/CPAN::Reporter. > Although CPAN doesn't support this in a release yet, knowing that doing "exit > 0" will at least help keep CPANPLUS from filing false failures is a step in > the right direction. :) Absolutely. Now we need to publicize it. I'll go post something on the CPAN Testers wiki shortly. > Cool. For my own edification, how does CPAN behave when it encounters a > missing "configure_requires"? Does it simply install the dependency and then > continue, or does it install it and then re-run "perl Makefile.PL"? Or, > alternatively, does it just choke outright and abort the build? >From a quick skim of the code, it looks like it processes configure_requires right after it downloads and unpacks the distribution tarball. If configure_requires can't be satisfied, it doesn't continue at all. As Eric said, configure_requires does have to be in the META.yml, though. David
