Ron Mayer <rm_pg@cheapcomplexdevices.com> writes:
> It still accepts one odd input that 8.3 rejected:
> regression=# select interval '1 1' hour;
> Perhaps the additional patch below fixes that?
Hmm, not sure about that one. We decided a week or two back that we
don't want the thing discarding higher-order field values, and this
seems pretty close to that. As the code is set up (plus my patch)
I think it's the case that only the rightmost field specification
of the interval qualifier makes any difference for parsing the value;
the leftmost field doesn't really affect what we think the constant
means. That seems like a nice simple consistency property ...
regards, tom lane