>>Or was the code incorrectly used?
Hard for me to say, but I think its about caller context. The way I am using it
might be different ... hey the function was static ... copy & paster be warned!
The code appears to be doing the same thing as the backend (with the exclusion
of backend stuff like HasCTZSet and forced conversions). I plan to do an
extensive test sometime today. So far, I am getting the correct timestamp
conversions across the board.
Andrew
Michael Meskes wrote:
> On Sun, Dec 09, 2007 at 11:54:25AM -0500, Andrew Chernow wrote:
>> That explains why my libpq code was getting 3AM for "without time zone"
>> values. I am using code from src/interfaces/ecpg/pgtypeslib/timestamp.c
>> timestamp2tm(). That uses localtime() after converting the timestamp to an
>> epoch value. I changed this code so that it calls gmtime() for
>> TIMESTAMPOID and localtime() for TIMESTAMPTZOID. Now it works perfectly :)
>
> Does this mean pgtypeslib is buggy? This code has been taken from the
> backend ages ago, so some changes might have occured that I'm not aware
> of. Or was the code incorrectly used?
>
> Michael
