David Rowley <dgrowleyml@gmail.com> writes:
> On Wed, 30 Sep 2020 at 11:22, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> Indeed. However, this complaint shows that the error message is not clear
>> enough. I propose changing it to
>> ERROR: unique constraint on partitioned table must be a superset of the partitioning columns
>> or perhaps
>> ERROR: unique constraint on partitioned table must include all partitioning columns
> I prefer the former. Although I'd rather see the constraint type
> mentioned explicitly rather than using the word "unique" regardless of
> what the constraint type is.
If the DETAIL has that, which it does, I'd rather have the primary text
lay out the general rule not a specific case. Anybody who is not clear
that a pkey is a kind of unique constraint will get that from reading the
DETAIL. Also, avoiding that insertion should allow more robust
translation of the primary message, which is something we should worry
about if we are concerned that users won't understand the message.
FWIW, after a few minutes' more thought I think I like the second wording;
it's just as precise and does not rely on two-dollar words. I'm not dead
set on it though.
regards, tom lane