On Wed, 2012-06-20 at 09:55 -0400, J Chapman Flack wrote:
> I've never hacked on pgAdmin yet, so I'm curious from those who have:
Sorry for not answering sooner. Way overloaded here.
> What sounds like the smartest way to make updates in the Edit Grid
> not fail when connected to a database with default_transaction_read_only
> set to on? :
> 1. Just send every update wrapped in START TRANSACTION READ WRITE
> ... COMMIT ?
> 2. Don't do that by default, but trap the "error: transaction is
> read only" and reissue the update with START TRANSACTION READ WRITE
> in that case?
> 3. Query the backend for SHOW DEFAULT_TRANSACTION_READ_ONLY and then
> issue updates wrapped in START TRANSACTION READ WRITE only if it
> was ON?
> 4. Just send a SET SESSION DEFAULT_TRANSACTION_READ_ONLY TO OFF and
> forget about it?
> I'm asking because we noticed in 1.14.2 that it doesn't do any of
> these things yet, but just surprises the innocent user with the
> "error: transaction is read only".
Well, if the DBA sets default_transaction_read_only to off, it's for a
good reason. So, we shouldn't override this by default. It would be
better that the edit tool finds that default_transaction_read_only is
on, and asks if the user wants to turn it off so that he could make
In response to
pgadmin-hackers by date
|Next:||From: Jiby James||Date: 2012-06-28 07:55:17|
|Subject: Compile Source Code|
|Previous:||From: pgAdmin Trac||Date: 2012-06-27 20:46:25|
|Subject: [pgAdmin III] #365: pg_terminate_backend() can be used on backends
with matching role on 9.2|