Re: [Madeleine-devel] "non-monotonous time" error on Windows

[Anders B. <ndr...@ya...>]

> >   1) Error:
> > test_time_actor(TimeTest):
> > RuntimeError: non-monotonous time
> >     ./test_clocked.rb:71:in `forward_to'

> It seems that the sleep(0.1) call gets rounded down to sleep(0), even
> on Linux, so that test must work only by chance.
> It's only the test itself that is broken, not the code it's testing, so
> it's not a big problem. But we'll have it fixed before the next
> release.

Ok, this is what I've found: The fact that the sleep-method is a
disappointment when it comes to sub-second timing doesn't really affect
this test that much. Neither is the assumption that certain threads have
been scheduled in the test really wrong.
The assertion that breaked is only an extra check that time really has
progressed since the last call. When combining a fast machine with a
low-resolution system clock you sometimes get the same time twice.
I simply changed a "<=" to a "<", to only check that time doesn't go
backwards. A reasonable assumption, I think...

Could anyone who has seen this problem confirm that it's gone?

/Anders