On 14/12/11 17:15, Daniele Varrazzo wrote:
> The only widespread use of a numeric value I've seen (including many
> official psycopg2 examples) is for autocommit=0, so when I've changed
> the values I've cared to keep that value stable only. For this reason,
> my preference between the solutions goes to 2 too (in the unlikely
> case READ UNC started meaning anything for Postgres we can
> differentiate it from 1).
>
> Are you making the patch or shall I do it myself? If you change the
> values, take care of this check too:
> <https://github.com/dvarrazzo/psycopg/blob/devel/psycopg/connection_int.c#L1042>.
> In particular, before releasing, let me run the test against all the
> supported PG versions: these branch are exhaustively checked.
I'll do that later today. Thanks for the note.
federico