On 2 July 2016 at 21:37, Andreas Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de> wrote: > I'd vote for reverting the change and putting it in a branch until > Variable::Magic has got the right medicine. +1: Variable::Magic not being broken very much should be a blocking condition. It is used by B::Hooks::EndOfScope, ranked 129th in the river as of last February[1]. But the February data is wrong, because it does not consider namespace-clean as a consumer ( which is itself, ranked 96th ) 96: 5011 namespace-clean 129: 3432 B-Hooks-EndOfScope That Variable-Magic is not itself represented as a dependency of B-Hooks-EndOfScope should be seen as a defect in the river analysis, but its understandable, because its a dynamic requirement[2] So this breakage is very much "break the CPAN entirely" grade bad as far as I can tell. 1: February River https://gist.github.com/xdg/af7a32c5b21d45a6b255 2: Dynamic Requirement code in B-Hooks-EndOfScope: https://metacpan.org/source/ETHER/B-Hooks-EndOfScope-0.21/Makefile.PL#L52-55 -- Kent KENTNL - https://metacpan.org/author/KENTNLThread Previous | Thread Next