The following test fails on Racket CS on Windows (only): (test 1969 date-year (seconds->date -86400)) Here, seconds->date uses time-utc->date at the rumble layer. On Chez, time-utc->date calls localtime_r and the Windows version of localtime_r doesn’t support negative time.

Since Racket BC implements the core of seconds->date in rktio, does it make sense for Racket CS to switch seconds->date to using rktio? Or should this test be considered as platform-dependent?

Probably CS should switch to rktio for seconds->date.

Okay. Sounds good. I’ll give it a try