At 2005-05-19 11:47:16 +0530, ams(at)oryx(dot)com wrote:
>
> + Datum
> + md5_bytea(PG_FUNCTION_ARGS)
> + {
> + /* It would be nice if we could avoid de-toasting the whole bytea,
> + * and feed it to md5_hash in small chunks instead. */
> + struct varlena *in = PG_DETOAST_DATUM(PG_GETARG_DATUM(0));
Oops, I guess that should be "bytea *in = PG_GETARG_BYTEA_P(0);" now.
-- ams