@Aristotle: Your two one-syllable answers were very useful. The bug you spotted in your longer answer was from a previous release.
But armed with the knowledge that the voice of experience indicated that the problem almost certainly lay in my packaging, I went back and added the Marpa module to *both* the 'require' and 'build_requires' for Marpa::HTML. Now all seems to work perfectly. An obstacle for me was that Module::Build's actual behavior seems to contradict the documentation in Module::Build::API, which says of 'build_requires' that "Modules listed in this section are necessary to build and install the given module, but are not necessary for regular usage of it." Read literally, that says that any module necessary for regular usage should never be listed under 'build_requires'. As far as I can tell the actual behavior of Module::Build would be more in accord with the following: "Modules listed in this section are necessary to build and install the given module. If a listed module is necessary for regular usage as well as the build and install, the listed module also needs to be specified using the 'requires' argument." Thanks, jeffrey On Mar 18, 5:23 am, pagalt...@gmx.de (Aristotle Pagaltzis) wrote: > * Jeffrey <jeffreykeg...@jeffreykegler.com> [2010-03-18 13:00]: > > > 1.) Is this to be expected when a new module (first > > non-developer release) is a prerequisite? > > No. > > > 2.) Is my set-up still at fault? META.yml, which is created by > > Module::build has a "Marpa: 0.100" line, which I hoped would > > suffice. > > Neither your Makefile.PL nor your Build.PL list Marpa as > a requirement at all. It’s kinda funny that you go to extra > effect to get Module::Build to hack it into the META.yml, by way > of `meta_add`, when just listing the prereqs properly would make > it do that automatically for you. What made you do that? > > In any case the result reported by the Testers is entirely to be > expected, and is just what users will experience if they try to > install your module with any CPAN installer: they’ll be forced to > resolve the dependency manually before it will install. > > > 3.) Is this a problem with some of the cpantesters setups. > > No. > > Regards, > -- > Aristotle Pagaltzis // <http://plasmasturm.org/>
