=# SELECT '1=>x'::hstore, 'x=>1.02'::hstore, 'x=>true'::hstore, 'x=>false'::hstore; hstore | hstore | hstore | hstore ----------+-----------+--------+-------- "1"=>"x" | "x"=>1.02 | "x"=>t | "x"=>f (1 row)
This is very simple select, but it demonstrates types support, thanks to a new binary storage, which now supports scalars, boolean and numeric values.
=# select 'scalar'::hstore, '100.1'::hstore, 'true'::hstore; hstore | hstore | hstore ----------+--------+-------- "scalar" | 100.1 | t (1 row)
That means, json could use our binary storage and finally be effectively usable !
As always, there are something we need to develop, but basic functionality is already ready.
We need to touch indexing stuff, which works, but we didn't checked yet, and binary recv/send.
Also, we are waiting for accepting gin patch (I already wrote here ), which greatly accelerates
hstore.