While fixing the collection tests in Xcode 16b5 beta I realised that this is probably related to changes in swift-foundation. Updating the base image to the nightly we used in run #10 makes the collection tests behave the same again on macOS and Linux.
There were two additional tests that started failing, on Linux only. The reason is small rounding/representation errors when comparing dates. We have to temporarily make those with ~microsecond accuracy instead.
To do:
[x] Once tests pass, tag spi-base and use the tagged version
While fixing the collection tests in Xcode 16b5 beta I realised that this is probably related to changes in swift-foundation. Updating the base image to the nightly we used in run
#10makes the collection tests behave the same again on macOS and Linux.There were two additional tests that started failing, on Linux only. The reason is small rounding/representation errors when comparing dates. We have to temporarily make those with ~microsecond accuracy instead.
To do: