It is tempting to use very large `Duration` value to get a practically
infinite timeout.
Before this commit Tokio panics on checked Instant + Duration
overflow.
This commit implements very simple fix: if Instant + Duration
overflows, we use duration = 30 years. Better fix should avoid
firing a timer on duration overflow. It requires deeper understanding
how timers work, but also it is not clear, for example, what
`Sleep::deadline` function should return.
Similar fix is done for `sleep`.
The trait-level documentation accurately describes this method, but the
function-level documentation imples that an implementer should overwrite
the entire buf, rather than just the unfilled portion.
Before this commit `close` after successful `try_recv` panics.
My use case is this: on drop, I call `close` to prevent pushing a
message to the queue, and then fetch the message if any and process
it.
But if message is already processed, `close` panics. And there is
no API to know if message was already fetched or not (except for
writing a wrapped which would track that info, which would be an
overkill).
But generally `close` operation should be safe to be called any
time.
Since ed5a94eb2db95b7cc142045fbbd5d68c6276e04e, there is no longer a
limit of 32,768 threads on the worker threads and blocking thread pool,
but the documentation has not been updated. This commit removes that
part of the documentation.
The time driver stores an Instant internally used as a "base" for future
time calculations. Since this is generated as the Runtime is being
constructed, it previously always happened before the user had a chance
to pause time. The fractional-millisecond variations in the timing
around the runtime construction and time pause cause tests running
entirely in paused time to be very slightly deterministic, with the time
driver advancing time by 1 millisecond more or less depending on how the
sub-millisecond components of the `Instant`s involved compared.
To avoid this, there is now a new option on `runtime::Builder` which
will create a `Runtime` with time "instantly" paused. This, along with a
small change to have the time driver use the provided clock as the
source for its start time allow totally deterministic tests with paused
time.
