Re: Rollback on close [Was: Fwd: [DB-SIG] conn.close() idempotence]

From: Daniele Varrazzo <daniele(dot)varrazzo(at)gmail(dot)com>
To: Marko Kreen <markokr(at)gmail(dot)com>
Cc: psycopg(at)postgresql(dot)org
Subject: Re: Rollback on close [Was: Fwd: [DB-SIG] conn.close() idempotence]
Date: 2011-10-19 18:34:11
Message-ID: CA+mi_8YA4Qmdgz99DaGDoypxJ+rzSvE-z020P3t80u1ti9Q3TQ@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: psycopg

On Wed, Oct 19, 2011 at 6:50 PM, Marko Kreen <markokr(at)gmail(dot)com> wrote:

> Let's clarify that further - DB-API *MUST* say that .close()
> [and also .__del__()] will rollback.  Eg - imagine
> driver accessing .dbf / .csv files directly.
>
> But there is no good reason to require communication
> with backend if the backend will rollback anyway
> on connection close.

Yes, this is our current position. Changing from it could only improve
the interaction between "sloppy code" and middleware, and as you have
pointed out this would make edge case more troublesome.

Thank you for the discussion, I wanted to hear the opinion of somebody
on the other side of the socket :)

-- Daniele

In response to

Browse psycopg by date

  From Date Subject
Next Message Frank Kauff 2011-10-19 19:35:45 Re: Error: no Python Codec for client encoding
Previous Message Marko Kreen 2011-10-19 17:50:15 Re: Rollback on close [Was: Fwd: [DB-SIG] conn.close() idempotence]