Обсуждение: Minor quibble with description of WINDOW clause

The way the WINDOW clause docs are currently worded, it's clear from the
grammar tree that such clauses come toward the end of the query. But this
phrase made me wonder, incorrectly, whether they were supposed to come before
calls to window functions, somehow:

"...where window_name is a name that can be referenced from subsequent window
definitions or OVER clauses"

See http://www.postgresql.org/docs/9.3/static/sql-select.html#SQL-WINDOW for
the full context.

I'd have preferred to see "...from OVER clauses or subsequent window
definitions", which the attached simple patch does. And if I'm the only one
that likes the idea of such a change, I'll gladly go away quietly. :)

--
Joshua Tolley / eggyknap
End Point Corporation
http://www.endpoint.com

eggyknap <eggyknap@gmail.com> writes:
> The way the WINDOW clause docs are currently worded, it's clear from the
> grammar tree that such clauses come toward the end of the query. But this
> phrase made me wonder, incorrectly, whether they were supposed to come before
> calls to window functions, somehow:

> "...where window_name is a name that can be referenced from subsequent window
> definitions or OVER clauses"

> See http://www.postgresql.org/docs/9.3/static/sql-select.html#SQL-WINDOW for
> the full context.

> I'd have preferred to see "...from OVER clauses or subsequent window
> definitions", which the attached simple patch does. And if I'm the only one
> that likes the idea of such a change, I'll gladly go away quietly. :)

I think you're right.  Applied and back-patched.

            regards, tom lane