Guillaume Lessard
Guillaume Lessard
> storage.swapAt(largestDescendantIdx, index) I wonder if `storage.withUnsafeMutableBufferPointer { $0.swapAt(largestDescendantIdx, index) }` would also exhibit the excessive allocations problem?
> it does perform significantly worse though at 7.7s instead of ~6s for the manually inlined version of swapAt Interesting. I would think that the manually-inlined would do worse than...
The proposed new test trips TSAN on Linux with: `ERROR: ThreadSanitizer: SEGV on unknown address 0x000000000000` This is not useful information, as we are expecting it to crash.
There should not be a data race because `cancel()` is required to be thread-safe. This is in the contract for `Subscription`. What got me down this road was trying to...
@mdznr Such an optimization would only work for a `RandomAccessCollection`, and these are not specialized for those indexing semantics (yet.)
@swift-ci please test
Why add `@discardableResult` on `stablePartition` but not `partition`?
@haaakon I was interested specifically in `AssertParse` and `AssertErrorMessage`, and those (currently) just work when copied to a test target.
I don't have complete tests yet, but thought I should check in for feedback anyway.
Interesting. I had relied on cppreference to be correct, and assumed I did not need to read the standard.