At 12:59 AM 12/8/2001 +0300, Arioch wrote:

>as far as i got it, error code holds bits, that are unreliable as they were
>designed for vax vms, and currently has no sense. No good code will pay
>attention to them. There's no need to remap all the codes.

The error codes that concern me are 32 bit values which are
compared against other 32 bit values, using an integer to
integer comparison. Nobody, to my knowledge, checks just
the low order bits. These are the isc_ and gds_ error codes.
Applications that use the API get those values from the status
vector which often includes secondary information. We really
don't need to use high order bits of code value to pass quite
a lot of information.

The SQL codes are a different matter. Aside from being negative
when an error occurs, zero on success, and positive on warning,
they're not defined by the standard or by convention. We could
add more.

Perhaps I've completely missed the point?

Regards,

Ann
www.ibphoenix.com
We have answers.