I think there's a bug with NULL handling
I was genuinely surprised that this was not caught by any test in the regression.
There are 32 cases, next patch will verify both the plan and the output
FOR r IN
SELECT idx_dir, idx_nf, qry_dir, qry_nf, sign
FROM unnest(ARRAY['+','-']) AS sign,
unnest(ARRAY['ASC','DESC']) AS idx_dir,
unnest(ARRAY['FIRST','LAST']) AS idx_nf,
unnest(ARRAY['ASC','DESC']) AS qry_dir,
unnest(ARRAY['FIRST','LAST']) AS qry_nf
{ oid => '2308', descr => 'nearest integer >= value',
- proname => 'ceil', prorettype => 'float8', proargtypes => 'float8',
- prosrc => 'dceil' },
+ proname => 'ceil', prosupport => 'arg0_asc_slope_support',
+ prorettype => 'float8', proargtypes => 'float8', prosrc => 'dceil' },
{ oid => '2320', descr => 'nearest integer >= value',
proname => 'ceiling', prorettype => 'float8', proargtypes => 'float8',
prosrc => 'dceil' },
Shouldn't 2320 also have the same change? Same for 1711/2167
{ oid => '183',
- proname => 'int42mi', prorettype => 'int4', proargtypes => 'int4 int2',
- prosrc => 'int42mi' },
+ proname => 'int42mi', prosupport => 'diff_slope_support',
+ prorettype => 'int4', proargtypes => 'int4 int2', prosrc => 'int42mi' },
Similarly shouldn't int24/int42 mul and div also be included?
Here I am genuinely NOT surprised :) I will take another look there before
submitting the next patch.
Regards,
Alexandre
