On Tue, Apr 11, 2017 at 4:02 AM, Robert Haas <robertmhaas@gmail.com> wrote:
> On Mon, Apr 10, 2017 at 2:09 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> ilmari@ilmari.org (Dagfinn Ilmari Mannsåker) writes:
>>> Why bother with the 'rte' variable at all if it's only used for the
>>> Assert()ing the rtekind?
>>
>> That was proposed a few messages back. I don't like it because it makes
>> these functions look different from the other scan-cost-estimation
>> functions, and we'd just have to undo the "optimization" if they ever
>> grow a need to reference the rte for another purpose.
>
> I think that's sort of silly, though. It's a trivial difference,
> neither likely to confuse anyone nor difficult to undo.
+1. I would just do that and call it a day. There is no point to do a
mandatory list lookup as that's just for an assertion, and fixing this
warning does not seem worth the addition of fancier facilities. If the
function declarations were doubly-nested in the code, I would
personally consider the use of a variable, but not here.
--
Michael
