Re: Transaction aborts on syntax error.

Greg Stark <gsstark(at)mit(dot)edu> writes:
> So you picture the backend automatically introducing a mini-nested-transaction
> for every request and automatically rolling that back on any error. So the
> application or user wouldn't have to do anything to continue processing
> ignoring the error?

You're assuming a bunch of facts not in evidence about how we choose to
present this functionality to clients, including a rather dubious
assumption that we'd choose to break backward compatibility.

My guess is that there will be some way to get the above behavior
(possibly implemented by client-library code rather than the backend),
but that it won't be the default.

regards, tom lane