>>> Gregory Stark <stark@enterprisedb.com> wrote:
> "Kevin Grittner" <Kevin.Grittner@wicourts.gov> writes:
>> In what way would an application want to treat deadlocks and update
>> conflicts differently? Both result from conflicts with concurrent
>> transactions and can be retried automatically. It seems like an
>> implementation detail with little chance of impact on applications
to
>> me. Can anyone provide a contrary example or argument?
>
> Well generally deadlocks are treated differently in that they are
treated by
> rewriting the application to not cause deadlocks.
I certainly don't propose changing the PostgreSQL error number or the
content of what is logged. Just the SQLSTATE. How would that make
what you suggest harder? It would certainly allow applications and
frameworks which are SQLSTATE-aware to automatically recover from
these until the rewrite is complete, which can hardly be considered a
bad thing.
-Kevin