johnf wrote:
> On Thursday 28 February 2008 09:29:29 am Paul McNett wrote:
>> johnf wrote:
>>> On Sunday 24 February 2008 08:08:10 pm Adrian Klaver wrote:
>>>> On Sunday 24 February 2008 7:50 pm, johnf wrote:
>>>>> On Sunday 24 February 2008 07:37:50 pm Adrian Klaver wrote:
>>>>>> Where do you see the error appear
>>>>> Where does it appear. Anytime I close my forms I see
>>>>> "Dabo Info Log: Sun Feb 24 19:38:22 2008: Application finished"
>>>>> And that's all.
>>>> It is appearing in the Postgres log.
>>> Adrian,
>>> You have been following the discuss on the psycopg list. Have you any
>>> suggestion on how to handle the problem in Dabo?
>>>
>>> My testing has not provided a simple solution. Your suggested change
>>> only changes the error message from a dropped client to an abort. To be
>>> honest the error messages are not preventing Dabo from working but do
>>> imply something could go wrong in the future.
>>>
>>> BTW just adding a connection.execute("commit") does not prevent the error
>>> messages from Dabo.
>> If one follows example of pure-python dbapi access to postgres, using
>> psycopg, does the error occur or no? If no, what specifically is Dabo
>> not doing that it needs to be doing? If yes, we don't need to blame this
>> on Dabo but on either psycopg or postgres.
>>
>> Do we need a hook on app exit to explicitly close connections or something?
>>
>> Paul
> I am not suggesting this has anything to do with Dabo other than I would like
> a work around. To answer your question - my testing (without Dabo or with
> Dabo) shows a bug in the closing of the connection if some sort of
> transaction has been created. The error may depend on several factors:
> 1. the isolation setting in postgres
> 2. the type of transaction executed
> 3. the way the connection is closed/disconnected.
>
> What I'd like to see is some sort work around in Dabo that determined the
> isolation setting and took the right steps to close a Dabo app. That said -
> I have no idea what those steps are. I'm hoping open discussion (centered on
> a Dabo fix) will help determine the steps. Again I don't see this as Dabo
> bug but something Dabo has to deal with. I also do not see the psycopg guys
> taking this to seriously - after all it does not appear to effect any of the
> data transactions.
My point is, if we can reproduce the problem and the fix in pure python,
then it should be clear how to make the workaround in Dabo. Don't put
the cart before the horse!
Paul
_______________________________________________
Post Messages to: Dabo-users@leafe.com
Subscription Maintenance: http://leafe.com/mailman/listinfo/dabo-users
Searchable Archives: http://leafe.com/archives/search/dabo-users
This message: http://leafe.com/archives/byMID/dabo-users/[EMAIL PROTECTED]
