Gigasecond: Fix failing tests

softmoth · September 21, 2023, 4:43pm

There are two causes of failures. First, if the two dates are joined at different milliseconds, then they will inherit different millisecond values during d:join, unless the full 10 items are specified in the array.

Second, there is a bug with d:+msec in 8th version 23.06 (I believe since in 23.02 https:/8th-dev.com/bin/bugs/tktview/b82069cc12), causing the time to change by the value of the local time zone. If your time zone GMT you won’t notice. It is fixed for the next release (23.07?).

glennj · September 21, 2023, 5:15pm

The 8th-test-runner repo runs in a Ubuntu 18.04 image, so the time zone for test runs will be UTC. And I see from the commit history that the 8th.zip is 23.01.01, so that’s a bullet dodged with that bug.

softmoth · September 21, 2023, 5:17pm

It’s important to fix it so users won’t be confused why their tests are failing locally, right?

glennj · September 21, 2023, 5:28pm

Indeed. Merged.

In other tracks I’m involved with, we added a warning to the instructions to use UTC, in the language-specific way.