My personal preference is to abandon broken CPAN modules whose authors are unresponsive, because there's no way to predict the future of the module. Maybe the author comes out of his coma and becomes responsive, maybe a new author takes it up and is driven by the sum of the bug reports to make design decisions that are counter to my needs, maybe the namespace is surrendered to someone who has an incompatible (though not necessarily incorrect) perception of what ought to occupy it.