favonia

@Zoullx Hi, I haven't heard from you for a while. What I gathered is that you have encountered some problem (perhaps some bug) in the past, and you believe a...

I think this is moving in a brighter direction. Question: it seems impossible for `Task` to work with more than two _anonymous_ `Workset.t` at the same time, and that for...

Got it. (By "anonymous" I meant "worksets created internally by `Task.run`" and by "call `run` to switch" I meant exactly "nesting `run_with_workset`". Just fixed my previous comment to replace "`run`"...

I like the new API. I have another idea---can we drop the `parallel_` prefix? * `iter`/`for_`: the current `parallel_for` * `reduce`: the current `parallel_for_reduce` * `scan`: the current `parallel_scan` *...

As the "author" of this issue, I support rethinking the API. My original concern is really about the obvious inconsistencies (which are now fixed). Maybe `get_num_domains` indeed should not add...

> This is unexpected (to me), but closer to what the previous documentation would say Well, I think "previous documentation" here might be misleading. It was not clear (to me)...

> If the `promise` is not resolved at the time of the call, rather than sitting idle, the current domain also services tasks on the `pool`. What I understand is...

@panglesd This is a great idea! Personally I prefer `{code ...}` over `{c ...}` for several reasons: (1) for me it is easier to remember and (2) it matches the...

We thought they were useless in the presence of discrete types, but now without equality types (in this proof theory) discrete types seem useless. So, maybe there is room for...

It is unclear whether we want regular types, but we need to have a kind system first.