>
>>>> The #defines would be less brittle in
>>>> the event, for example, that the postgres epoch were ever changed.
>>>
>>> I don't think it is real, and even in such case all constants are
>>> collected together in the file and will be found and changed at once.
>>
>> I agree that they would be found at once. I disagree that the example
>> is not real, as I have changed the postgres epoch myself in some builds,
>> to be able to use int32 timestamps on small devices.
>
> Wow! I'm sorry, I didn't know about it.
> But in such case (tighten to int32) there are more than two places
> which should be changed. Two more (four with disabled
> HAVE_INT64_TIMESTAMP) constants are not big deal with it.
Please, do not worry about that. I do not mean that your code needs
to be compatible with my fork.
Regards,
Mark Dilger