> Wouldn't a better alternative be to have a wait function which can
> deal with readable pipe connections and integer handles?
> 
> On Unix this would just delegate to select().
> 
> On Windows it could work as follows:
> * initiate an overlapped read on each connection
> * call WaitForMultipleObjects()
> * cancel each overlapped read
> * continue any read which succeeded but only gave a partial message
> * store read messages on associated connection objects

This sounds like a good direction to me.
(of course it must also preserve existing functionality;
test_concurrent_futures will check for that)