Re: no "+" operator for smallint and bigint

Why not do it like C does, promote to the RESULTANT data type.

If he had rewritten his statement like so, then it would work
as per my statement above:

operator# select (8::smallint + 8::bigint)::biginit;

2/24/2003 6:59:44 AM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:

>Martijn van Oosterhout <kleptog(at)svana(dot)org> writes:
>> On Mon, Feb 24, 2003 at 08:19:20PM +0900, Jean-Christian Imbeault wrote:
>>> PG=3D#
>>> ERROR: Unable to identify an operator '+' for types 'smallint' and 'bigi=
>> nt'
>
>> Since there is no explicit operator it doesn't know if the result should be
>> a smallint or a bigint so it's asking you to tell it.
>
>More specifically, the parser finds two equally plausible choices: cast the
>smallint to int and apply "int4 + int8", or cast the smallint to bigint
>and apply "int8 + int8". Not knowing which to prefer, it has to punt.
>
>> Well, it not quite a straight forward problem and there has been much
>> discussion about how to solve it.
>
>That's an understatement :-(.
>
>I currently like the idea of getting rid of as many cross-datatype
>operators as possible --- offering a native "int4 + int8" operator
>doesn't seem to have any obvious benefit over letting the parser
>insert a promotion. But there are some pitfalls in that idea too.
>
> regards, tom lane
>
>---------------------------(end of broadcast)---------------------------
>TIP 4: Don't 'kill -9' the postmaster
>