I'm fooling around with getting the parser to report an error cursor
location if input conversion fails for a constant in a SQL command.
For instance:
regression=# select 42 = 'foo';
ERROR: invalid input syntax for integer: "foo"
LINE 1: select 42 = 'foo'; ^
regression=# select '2000-jax-01'::date;
ERROR: invalid input syntax for type date: "2000-jax-01"
LINE 1: select '2000-jax-01'::date; ^
regression=#
This seems like it'd be a pretty useful thing to have in long queries,
but in short queries it looks a bit like overkill. And it affects
the expected output of a whole lot of the regression tests.
Does anyone think this might be "too chatty"?
regards, tom lane