Closed
Description
There's something fishy in the family definition for SEND -- the family name is incorrectly for_iter. But it's not so simple to fix because apparently SEND_GEN has an inconsistent cache effect (which wasn't detected previously because of the duplicate family name -- it's a bug in the cases generator that it doesn't complain about that).
We should both
- Fix the stack effect of SEND_GEN
- Fix the generator to complain about duplicate family names