Chapman Flack <chap@anastigmatix.net> writes:
> On 2/5/20 10:20 AM, Tom Lane wrote:
>> * Third-party context types would have to force the compiler to take
>> context-type values that weren't among the known enum values ---
> Doesn't that seem like a long run for a short slide?
Well, one thing we could do is assign an "other" or "custom" code,
and people who were just doing one-off things could use that.
If they were going to publish their code, we could encourage them
to ask for a publicly-known enum entry. We have precedent for
that approach, eg in pg_statistic stakind codes.
regards, tom lane