You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
As @devyn is exploring for us, truly useful parallelization seems extremely difficult with the current design. On top of that, there's issues like #4, that prove that the implicit ordering-of-serial-operations isn't really all that effective within the framework we've designed.
@devyn's suggestion is that we simply discard those constraints entirely: anything may happen in any order, subject only to the constraints of A) dispatches from a particular routine (i.e. the un-Tires-esque strictly-serial nature of our scripts), and B) the responsibility system's marshalling of separate scripts' executions. (To put it another way, implementation-wise: “get rid of the global operations queue entirely.”)
I'm still not remotely sure of the implications of doing so, though.
The text was updated successfully, but these errors were encountered:
As @devyn is exploring for us, truly useful parallelization seems extremely difficult with the current design. On top of that, there's issues like #4, that prove that the implicit ordering-of-serial-operations isn't really all that effective within the framework we've designed.
@devyn's suggestion is that we simply discard those constraints entirely: anything may happen in any order, subject only to the constraints of A) dispatches from a particular routine (i.e. the un-Tires-esque strictly-serial nature of our scripts), and B) the responsibility system's marshalling of separate scripts'
execution
s. (To put it another way, implementation-wise: “get rid of the global operations queue entirely.”)I'm still not remotely sure of the implications of doing so, though.
The text was updated successfully, but these errors were encountered: