Re: patch: streaming of bytea parameter values

Kris Jurka wrote:

>>Being user-friendly when dealing with application errors is nice
>>behaviour to have, but it *is* a separate issue..
>
>
> Separate from what, correctness?

Yes.

> We both agree that a short stream or
> IOException is an error, but I'm not advocating the JDBC equivalent of
> System.exit() as a viable solution. What makes it OK for this
> particular error to destroy a Connection while no other application level
> error can?

The implementation of SQLState (or connectionErrorOccurred) across
different drivers is quite patchy. When writing portable applications
(or even driver-specific applications, to a lesser extent) you have to
assume that any SQLException you don't specifically recognize means your
connection is toast.

I don't see that tearing down the connection for a certain class of
hard-to-handle errors makes this situation any worse.

-O