Re: pgsql/src/backend/nodes (copyfuncs.c outfuncs.c print.c)

Hiroshi Inoue <Inoue@tpf.co.jp> writes:
> Tom Lane wrote:
>> Now that I look at it, the optimizer *already* prefers fast-start plans
>> for cursors.  Is LIMIT ALL really necessary as an additional hint,
>> and if so how should it interact with the bias for cursors?

> If LIMIT doesn't restrict the total count of rows which cursors
> could return,there's no problem. Otherwise LIMIT ALL would be
> needed.

But is there a reason to treat LIMIT ALL differently from no LIMIT
clause at all?

            regards, tom lane