This PR replaces the new use of Atomic<Int> with a Locked<Int> in the implementation of deliverExpectationCheckedEvents. Why? Because we're running into some environments where the Synchronization module isn't available (e.g. older host macOSes) and this is simpler. The performance profile is comparable: on my system, running the repeatedlyExpect() test takes 0.55s instead of 0.49s to call #expect() 1,000,000 times, so it's still a significant win over the implementation we had earlier.
Checklist:
[x] Code and documentation should follow the style of the Style Guide.
[x] If public symbols are renamed or modified, DocC references should be updated.
grynspan
commented
1 week ago
This PR replaces the new use of
Atomic<Int>with aLocked<Int>in the implementation ofdeliverExpectationCheckedEvents. Why? Because we're running into some environments where the Synchronization module isn't available (e.g. older host macOSes) and this is simpler. The performance profile is comparable: on my system, running therepeatedlyExpect()test takes 0.55s instead of 0.49s to call#expect()1,000,000 times, so it's still a significant win over the implementation we had earlier.Checklist: