Closed finagolfin closed 1 year ago
Yeah I think depending on the specific numbers of the capacity is probably wrong. I'd happily take a patch that measures the capacity before removeAll
and then confirms it's unchanged, rather than relying on hardcoded numbers. Are you open to making that patch?
Submitted your suggestion as #395.
393 broke my Android CI, which runs that test in an Android x86_64 emulator, with these 4 test failures:
All four are related to checking the capacity of the header array and not getting the expected size.
I can reproduce on my Android AArch64 phone, so I stuck these
print()
s in the initializer and found the offending line when the test is run:That prints the following for this test:
I have no idea if this
map
should be allocating a larger capacity or if this even matters: shouldn't the array capacity just be an implementation detail?Opening this to see what you all think of this strange failure, let me know.