Simon Marlow

Ah, the package name of course :)

Any chance you could turn the example into a test case?

The example is compelling, but I worry that it would turn accidental pattern match failure into deadlock, which would make debugging much harder. Happy to hear other opinions though.

I'd rather not make `async` depend on `lens`. Could these be put into a separate package `async-lens`, say in a module `Control.Concurrent.Async.Lens`?

It would be a breaking change, which is unfortunate, and all call sites would have to be updated. So while it would indeed have been a better choice originally, I'm...

My guess is it probably isn't worth it. A completed `ThreadId` doesn't take much space.

There's a bit of documentation here https://hackage.haskell.org/package/base-4.18.0.0/docs/Control-Concurrent.html#g:14 In particular **it's a debugging feature** so it doesn't have a well-defined semantics. To have a semantics would mean specifying the garbage collector,...

I don't have any examples unfortunately.

Yes, it would be nice to fix this properly. Pull requests accepted!

I'm fairly sure that you would also need to look at the parts of Happy that generate code, in particular https://github.com/simonmar/happy/blob/master/src/ProduceCode.lhs, and the part where we generate the imports, https://github.com/simonmar/happy/blob/master/src/Main.lhs#L490...