On 27.07.2016 05:00, Joshua D. Drake wrote:
> On 07/26/2016 06:25 PM, Peter Eisentraut wrote:
>> On 7/5/16 4:24 AM, Albe Laurenz wrote:
>>> But notwithstanding your feeling that you would like your application
>>> to break if it makes use of this behaviour, it is a change that might
>>> make some people pretty unhappy - nobody can tell how many.
>>
>> What is the use of the existing behavior? You get back an arbitrary
>> implementation dependent value. We don't even guarantee what the value
>> will be. If we changed it to return a different implementation
>> dependent value, would users get upset?
>
> No they would not get upset because they wouldn't know.
>
> Can we just do the right thing?
Attached is a patch to "do the right thing". The verification is in
"to_date()" now, the extra function is removed. Regression tests are
updated - two or three of them returned a wrong date before, and still
passed. They fail now. Documentation is also updated.
Regards,
--
Andreas 'ads' Scherbaum
German PostgreSQL User Group
European PostgreSQL User Group - Board of Directors
Volunteer Regional Contact, Germany - PostgreSQL Project