| |
|
Sebastiaan Koppe
| On Tuesday, 16 January 2024 at 21:19:29 UTC, Richard (Rikki) Andrew Cattermole wrote:
> On 17/01/2024 10:15 AM, Sebastiaan Koppe wrote:
>> It may be possible that the language coroutine support may be able
>> to drive it. Which could be worth considering.
>>
>> Its the other way around really.
>
> That wasn't what I meant.
> ...
> So library coroutine representation and sender/receiver library representation would both take as argument the language coroutine feature.
I don't know enough of the specific language coroutine feature you have in mind, but I don't think its the right base building block for async computations. For one they allocate too often, and I am not sure how they can handle cancellation without explicitly passing stoptokens around.
However, I absolutely do agree with you that whatever language support D gets, it needs to look like regular sequential code.
|