Refactored EventLoop::timeout_ms to use Duration instead of u64

[zslayton]

Addresses #32. Removed _ms from the method name since the Duration class can represent a range of units.

[sschepens]

I was thinking of dropping a pull request for this too, we have to be careful when casting i64 to u64, if it's negative it will give a huve positive u64, maybe we can error o reject negative i64s? or maybe set timeout to 0 when it's negative

[zslayton]

I considered this -- in honesty, I'm not sure why Duration uses signed values. I'm not sure what it would mean to have a negative duration of time. "Two days ago" is still a duration of two days. Perhaps I should open a ticket upstream to discuss it.

In the meantime, @carllerche, how would you prefer that this case be handled?

[sschepens]

Yeah, i also thought about it, it's kind of misleading to have negative durations but, but golang seems to have them too and there are some use cases where it's actually handy to have negative durations.

[carllerche]

For now, if the duration is negative, I would set a 0ms timeout. I would also open an issue upstream. I am going to discuss this issue with @aturon. Besides that, the PR looks good. Thanks

[zslayton]

I've started a thread on discuss.rust-lang.org. I'll tweak the PR to set a 0ms timeout this evening.

[carllerche]

THanks for the PR :)