On Mon, Feb 7, 2011 at 5:06 PM, Nicolas Grilly <nicolas@gardentechno.com> wrote:
> On Tue, Feb 1, 2011 at 14:18, Marko Kreen <markokr@gmail.com> wrote:
>>
>> No, the error message should arrive immediately. But it may be deficiency
>> of
>> libpq that you cannot aquire it before ending the copy.
>
> I have analyzed the PostgreSQL protocol using Wireshark (an open source
> packet analyzer), and I confirm that the PostgreSQL backend, while doing a
> COPY ... FROM STDIN, reports errors as soon as possible (especially errors
> related to invalid data).
> Therefore, the "late" reporting of errors while doing a COPY ... FROM STDIN
> is not a limitation of the underlying protocol; it is a limitation (or a
> design choice) of the libpq library.
> It looks like this is a well known issue because it is listed on the todo
> list:
> http://wiki.postgresql.org/wiki/Todo#COPY
> And was discussed before:
> http://archives.postgresql.org/pgsql-hackers/2008-04/msg01169.php
>
>>
>> I'm not sure about that actually. It should be possible to call
>> select() & PQconsumeInput
>> between copy calls, thus also PQgetResult to get the error.
>
> I have tried that, and many other things, and I have carefully read libpq
> source code, and my understanding is that it is impossible with the current
> implementation of libpq.
> It is impossible because the function pqParseInput3 (file fe-protocol3.c)
> does not parse error responses while the connection is in PGASYNC_COPY_IN
> state. We have to call PQputCopyEnd to switch the connection to PGASYNC_BUSY
> and start error messages parsing.
> Can Marko or someone else confirm this analysis? Any idea to improve that?
> Regards,
> Nicolas Grilly
> PS : Considering that both psycopg2 and psql are built on top of libpq, this
> a quite logical they behave the same regarding late reporting of copy
> errors, contrary to pg8000 that could easily read, parse and report the
> errors early.
Yeah, seems your analysis is right and it's libpq (design?) bug,
so no way to work around it.
--
marko
