Oliver Elphick <[EMAIL PROTECTED]> writes:
> 1. There are regression failures on timestamptz and horology which seem
> to have come about either on input or output of timestamps with
> fractional seconds.
Hmm ... I'm seeing slightly different misbehaviors in the
integer-datetimes and float-datetimes cases. I think we are about to
have to dig down to the root of the off-by-one-second cases that Tom
Lockhart noticed but never really solved. I suspect that the issue has
to do with platform-specific rounding of negative values to integer, but
I haven't got the full story quite yet.
> I tried debugging this but got a segmentation fault and apparent stack
> corruption in gdb, with the reported break point not anywhere I had set
> one. I don't know what to do about that.
Recompiling with -O0 might help debuggability.
> 2. If the postmaster is not running, there is garbage in psql's error
> message:
Works OK for me --- could you trace that one more fully?
> 3. There is a compilation warning that a constant will not fit into a
> long in adt.c. There are two more files where INT64CONST() is required
> but not supplied. Patch attached.
Not sure about these ... does anyone else get these warnings?
regards, tom lane
---------------------------(end of broadcast)---------------------------
TIP 9: the planner will ignore your desire to choose an index scan if your
joining column's datatypes do not match
