TracedStepRangeLen

[jumerckx]

I imagine some code should maybe be moved around a bit?

[giordano]

Uhm, the history of this PR contains unrelated commits

[jumerckx]

Uhm, the history of this PR contains unrelated commits

Sorry, should be fixed

[jumerckx]

failing CI is all:

...
  15864.0 ms  ✓ Reactant → ReactantCUDAExt
   4266.6 ms  ✓ LuxLib → LuxLibReactantExt
   5220.7 ms  ✓ Lux → LuxReactantExt

[2255] signal (2): Interrupt: 2
in expression starting at /Users/runner/work/_temp/8434ebd3-cb2c-4917-ad80-59675f07439b:3
kevent at /usr/lib/system/libsystem_kernel.dylib (unknown line)
unknown function (ip: 0x0)
Allocations: 6396017 (Pool: 6381939; Big: 14078); GC: 10
Error: The action 'Run Tests' has timed out after 60 minutes.

Is this due to flaky Github runners or should I attempt debugging this? (@giordano)

[giordano]

yes, we've had those hangs in the macOS-latest runners after precompilation in a while, reason why we did #843.

[giordano]

Note: instead of pushing a new commit which restart some 50 new jobs (11 of which on macOS, for which we have 5 runners across the entire organisation), you can simply go to the summary page of CI run and restart only failed jobs:

Of course now it's too late to that here since the 50 new jobs have restarted already, but good to keep in mind for the future 🙂

[jumerckx]

Note: instead of pushing a new commit which restart some 50 new jobs (11 of which on macOS, for which we have 5 runners across the entire organisation), you can simply go to the summary page of CI run and restart only failed jobs:

I'll keep this in mind in the future! In this case it was a merge commit from main, though, shouldn't CI fully rerun in that case?

[wsmoses]

looks good, maybe add a quick test?

[giordano]

In this case it was a merge commit from main, though, shouldn't CI fully rerun in that case?

Ok, if you did want to test on top of main then that was fine.

[jumerckx]

Added a test—and fixed a bug in getindex that was unearthed by it.

[jumerckx]

I think this is ready to be merged