That was the conclusion we came to: either do them all or none of them.
The problem is with the amount of time we spent working on, maintaining, and fielding support requests for rubygems, the last thing we wanted to do was multiply that work supporting additional systems.
This decision wasn't made lightly, but the service and support will be better off with it gone.
Then another option would have been to do the same thing for non-Ruby projects. Integrate with CPAN; build eggs (or whatever it is that Python uses); create Cabal packages.