Reproducing the exact sequence of events would be very difficult.
Essentially, the same JDBC connection was used simultaneously (in
multiple threads) for various selects and updates, which is a bad
enough thing to do, I presume, as the connection instance is not
thread-safe. I don't think that much should be invested in helping
detect programming errors like this. Let the onus of dealing with such
situations rest on the application programmers -- it is their fault
anyway. (Oracle makes it somewhat [not significantly much] more easier
to do this, but I find it normal that customers get at least some
extra features for their bucks.)
Thanks
Peter
2009/4/13 Kevin Grittner <Kevin.Grittner@wicourts.gov>:
> Péter Kovács <maxottovonstirlitz@gmail.com> wrote:
>> I found the source of the problem: the client
>> application made SQL calls in invalid sequences.
>
>> I ran the same test case against Oracle as well
>
>> it gave a more informative error message ("protocol violation");
>> and, also, the error message was emitted much closer to the place in
>> the execution path where the actual programming error occurred.
>
> Could you share information about what you did, what you would like or
> expect as an error message, and what you got instead? It might help
> us improve PostgreSQL.
>
> -Kevin
>
