Hi, On 2019-06-04 14:55:16 -0700, Andres Freund wrote: > On 2019-06-04 17:37:35 -0400, Alvaro Herrera wrote: > I think it might be worthwhile to optimize things so that all typlen > 0 > && typlen <= sizeof(Datum) are allowed for byval datums.
Maybe even just deprecate specifying byval at CREATE TYPE time, and instead automatically infer it from the type length. We've had a number of blunders around this, and I can't really see any reason for specifying byval = false when we internally could treat it as a byval. Greetings, Andres Freund
