Re: [PATCH] GROUP BY ALL

Peter Eisentraut <peter(at)eisentraut(dot)org> writes:
>> The code in transformGroupClause() currently says:
>>      * Likewise, TLEs containing window functions are not okay to add
>>      * to GROUP BY.  At this writing, the SQL standard is silent on
>>      * what to do with them, but by analogy to aggregates we'll just
>>      * skip them.

> At the most recent SQL standard meeting, I had this issue addressed, and
> the handling of window functions in GROUP BY ALL is now specified to
> work like PostgreSQL already works. I propose the attached patch that
> updates the code comment about that, and also makes the documentation
> more specific that we are only talking about aggregate functions and
> window functions referring to the same query level, which is another
> thing the original specification was silent about but which has since
> been fixed.

Cool, thanks for seeing to that. The proposed patch seems okay,
except that the docs wording makes it sound like outer-level window
functions are a thing, which they are not AFAIK. You could instead
write

- contain either an aggregate function or a window function.
+ contain either an aggregate function referring to the
+ same query level or a window function.

However, that seems to read a trifle more awkwardly, and the
distinction is quite pedantic. If you prefer your own wording
I won't complain.

regards, tom lane