Neil Conway <neilc@samurai.com> writes:
> - Is there a reason why you've made the type of SQLCODE `text', rather
> than integer?
The value isn't an integer ... which gets back to my point that this is
not compatible with Oracle's idea of SQLCODE and therefore we should *not*
use that name for it.
BTW: the patch has some memory-leak problems, I believe, because it is
studiously not following the var->freeval protocol. Now that I look,
it appears to be copied-and-pasted from some existing code that also
gets this wrong :-(
regards, tom lane