Another one from last night: fix for dlls/kernel/tests/time.c on Windows 98

[Alexandre Julliard]

Jakob Eriksson <jakov at vmlinux.org> writes:

> I  may be totally long but this is what I figured:
> 
> the low parts unit is 10 nanoseconds.  So one full dwLowDateTime is
> 2^32 10 nanoseconds, which is 120 hours.

I think you are off by a factor of 1000 (and the unit is 100
nanoseconds BTW).

> Or maybe we shouldn't test this particular corner case at all. As you said
> before, if it differs so wildly between different Windows versions, probably
> no real application depends on the behaviour.

Yes, I think the test is useless. Its whole point is to make sure the
function detects bogus input, but if it doesn't under Windows then
there is nothing to test.

-- 
Alexandre Julliard
julliard at winehq.org