I'm pretty sure that most committers would want to apply the
rm_identity part in a separate commit first. Could you make it
two patches, one introducing rm_identity, and another with
xlogdump using it?
Carp the code:
const char *
clog_identify(uint8 info)
{
switch (info)
{
)
WITH (
OIDS=FALSE
);
There are redundant call when sorting?
On Tue, Jul 1, 2014 at 2:16 PM, Martijn van Oosterhout
klep...@svana.org wrote:
On Sun, Jun 29, 2014 at 10:05:50PM +0800, gotoschool6g wrote:
The simplified scene:
select slowfunction(s) from a order by b limit 1;
is slow
The simplified scene:
select slowfunction(s) from a order by b limit 1;
is slow than
select slowfunction(s) from (select s from a order by b limit 1) as z;
if there are many records in table 'a'.
The real scene. Function ST_Distance_Sphere is slow, the query:
SELECT ST_Distance_Sphere(s,
