Attached are three patches: 1) Reverts the incompatible change to UNIVERSAL::VERSION such that once again it will return the stringified version object representation of $CLASS::VERSION. A long sidebar with David Golden, myself, and two different pumpkings came to the conclusion that the original code is less bad. There may be a future patch to make it return a version object (but only for 5.16.x and above), but that is still pending a ruling from the bench. 2) I discovered in perlguts that there is already an accepted method to determine if a given scalar is both a blessed object and derived from a given class name, so all such code version objects will use this instead of assuming SvROK(sv) is sufficient. 3) Clean up the version.t file with the forthcoming CPAN release, include reverting the test for the case in patch #1. These are supposed to be non-controversial (now that #1 has been decided) and really need to get in now (before 5.15.6)... JohnThread Next