Re: Regression: Problems with Timestamp arguments

  You're technically correct.

  Maybe this can be applied to NULL arguments only? Would that be
'more' acceptable?
  Lately I've been proven wrong with many of my (reasonable)
expectations, but I *hope* NULL timestamps with or without time zones
are the same!

2013/9/11 Kris Jurka <books@ejurka.com>:
>
>
> On Tue, 10 Sep 2013, Lachezar Dobrev wrote:
>
>>   I've made a blunt stab at working around the issue.
>>   The attached patch adds a new parameter to the JDBC URL: timestamp.
>> The parameter can be wither 'with_timezone' or 'without_timezone'. If
>> the parameter is missing, or has a different value then the current
>> behaviour (using Oid.UNSPECIFIED) is exhibited. With the parameters
>> specified correctly one of the Oid.TIMESTAMP or Oid.TIMESTAMPTZ is
>> used when sending Timestamp arguments to the server.
>
> This won't really work in general though.  If you have a database that has
> both timestamp and timestamptz fields in it, there is no valid setting of
> this configuration option.  Setting it either way will result in silent
> corruption of your data.
>
> Kris Jurka