Explanation of pg_column_size

Hi all,

Can someone please explain the following to me:

mse=# with l as (select 0.1::numeric as numlit) select
pg_column_size(numlit) as sz_numlit, pg_column_size(0.1::numeric) as
sz_expr from l;
sz_numlit | sz_expr
-----------+---------
5 | 8
(1 row)

It looks as though the expression is being cast to double precision, but
maybe it is invalid to use pg_column_size in this way.

I'm using 9.5.3 on OS X.

Thanks,

Steve