Ashutosh Bapat <ashutosh.bapat@enterprisedb.com> writes:
> On Tue, May 16, 2017 at 6:50 PM, Peter Eisentraut
> <peter.eisentraut@2ndquadrant.com> wrote:
>> On 5/15/17 23:45, Ashutosh Bapat wrote:
>>> +1. We should throw an error and add a line in documentation that
>>> collation should not be specified for hash partitioned table.
>> Why is it even allowed in the parser then?
> That grammar is common to all the partitioning strategies. It looks
> like it's easy to handle collation for hash partitions in
> transformation than in grammar. But, if we could handle it in grammar,
> I don't have any objection to it.
If you disallow something in the grammar, the error message is unlikely to
be better than "syntax error". That's not very desirable.
regards, tom lane