]> granicus.if.org Git - postgresql/commit
I think I've finally identified the cause of the off-by-one-second
authorTom Lane <tgl@sss.pgh.pa.us>
Mon, 31 May 2004 18:31:51 +0000 (18:31 +0000)
committerTom Lane <tgl@sss.pgh.pa.us>
Mon, 31 May 2004 18:31:51 +0000 (18:31 +0000)
commit87de80e95a7c5999dfaf4b769502c97cbe56248b
tree3bd50b67729b185646616a4b03a6ad6111faaf1c
parentd534b9ee9e8fec0b4d295e7450f6654b08490690
I think I've finally identified the cause of the off-by-one-second
issue in timestamp conversion that we hacked around for so long by
ignoring the seconds field from localtime().  It's simple: you have
to watch out for platform-specific roundoff error when reducing a
possibly-fractional timestamp to integral time_t form.  In particular
we should subtract off the already-determined fractional fsec field.
This should be enough to get an exact answer with int64 timestamps;
with float timestamps, throw in a rint() call just to be sure.
src/backend/utils/adt/timestamp.c