Closed WyriHaximus closed 1 year ago
@WyriHaximus I agree that it might be useful to remove the child process adapter for now given the major refactoring (#97 and friends). I still think this might be useful again in the future, but I'm happy to remove any excess code for the time being :+1:
(Not sure about the PR labels, I'll add feature and BC break, but feel free to update as you see fit.)
@clue When the time comes, and we can get it to run it's test in under a second we should consider adding it. But a 100 fold in performance difference is just to big.
While fully non-blocking the child process adapter is a factor 100 slower than all the other adapters combined, including the blocking fallback adapter. As such removing it wil increase performance for everyone not having ext-uv or ext-eio (coming in a follow-up PR) installed. Even at the cost of blocking for a short period of time.