Hi Meskes-san Thank you for your comment. I write three points in this mail.
1. > This also seems to be conflicting with > bd7c95f0c1a38becffceb3ea7234d57167f6d4bf. If we want to keep this > commit in for the release, I think we need to get these things fixed. I understand it. My idea is that add an argument for statement-name to ECPGdo() or add a new function that is a wrapper of ECPGdo() and has the argument. The argument is used for lookup related connection. Is it good? 2. > I wrote: > But it doesn't allow to use host variable in parameter clause of EXECUTE > statement like the following. I found a way to support host variables in parameter list of EXECUTE statement. ecpg_build_params() replace each parameter to string-formatted data that can be created by ecpg_store_input(). I will try it. 3. I found a bug in my patch. Replacing $ to @ in AS clause is not good because @ is absolute value operator. Therefore, the replacing cannot accept valid statement like the following. exec sql prepare st(int) select $1 + @1; -- It equals to "select $1 + 1" I choose $$1 unavoidably. Other character seems to be used as any operator. P.S. - PREPARE with FROM is the standard for Embedded SQL. - PREPARE with AS is not defined in the standard. - PREPARE with AS clause is PostgreSQL style. - Oracle and MySQL support only the standard. Regards Ryo Matsumura
