I'd like to propose a radical new notion for releases. Prior to tag, we use perl-framework to actually check that the current server is fully operational [as defined by our wonderful test-dev group!!!] If that passes, tag. Let other platforms run it through the test suite. If they pass, we roll. If noone speaks up for three days after the tag [on a given platform] ohy veh - try again later.
The SSL randomness is a showstopper, right now, for tagging .30. We shouldn't be releasing any new segfaults, least of all with something we can test [vis a vie the test suite.] No sense in wasting a perfectly good tag on a perfectly invalid point in code development. Bill