Stefan Schindler

Hi Matt I think this feature would be great. What behaviour would you expect? That after the `stop_queue()`/`abort_pool()` the queue is drained in the sense that all jobs up to...

With the design of 2.0 you will be able to shutdown generations of tasks. So without knowing anything about the function running in the pool as soon as it is...

Thank you for the PR. I will look at it and see how it fits into the the new architecture.

The reason the test fail is because the oldest rustc we support is 1.13 currently. What is the oldest possible version crossbeam supports?

I guess we will switch the crate to `edition = 2018` with 2.0, but keep the 1.x release with the support of rust 1.13.x since this would be a breaking...

One of the problems is, that we don't have control over the OS and the other tasks on the machine. Depending on the load, sleeping for 100ms may not be...

Good idea. If you are using a SGE managed cluster you can use my latest crate [son of grid engine](https://crates.io/crates/son_of_grid_engine). You can use [get_pinned_threadpool()](https://docs.rs/son_of_grid_engine/0.1.0/son_of_grid_engine/struct.SystemInfo.html#method.get_pinned_threadpool) for a correctly pinned pool there....

Thanks! I will take a closer look at it later

I think a more exact/explicit wording is better for the builder.

The problem with putting the `ThreadPool` inside an `Arc` is, the `Sender` is not `Send` and has checks against moving it from thread to thread. ([See this example](https://gist.github.com/dns2utf8/828d338bf3cb4214a8127786d89fe085)) Once the...