On Mon, Jul 28, 2014 at 7:54 PM, James K. Lowden <jklow...@schemamania.org> wrote: > On Mon, 28 Jul 2014 16:32:42 -0500 > Nico Williams <n...@cryptonector.com> wrote: >> Raising errors is not that useful here: most users who ever run into >> such timestamps will not really know what to do with them. > > I think you're quite wrong here. Until presented with evidence to the > contrary, I would assert the opposite: that users who run into leap > seconds *do* know what to do with them because they need to, for > astronomical reasons or somesuch. And have already done so.
That is a fair point. Note that SQLite3 apparently does no corrections for leap seconds anyways in date arithmetic (which it can't do if you're using numeric arithmetic on Julian days anyways!), which it would have to do (since UTC is non-uniform). I'm not entirely sure what it means for anyone who really cares about TAI, or who they might be besides astronomers. If all you want to do is parse dates, then indeed, raising an error would be better than ignoring a leap second, but if you want to do any arithmetic on dates, then you're already in trouble, given which I think the fair thing to do is ignore leap seconds, but I'll freely admit that it's not ideal. Nico -- _______________________________________________ sqlite-users mailing list sqlite-users@sqlite.org http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
