GeezerVison

Not sure what is going on here, but the test failures have survived a re-run.

I can see my Emscripten tip of tree job started failing on September 30th, so something committed that day or the day before probably. That narrows things down a bit but I was hoping to get an email or something when the job failed.

Job failed on emscripten-core/emscripten@59f4772
Succeeded on emscripten-core/emscripten@5bd3cf2

So there are these commits:

• 59f4772b8 Fix for MAIN_MODULE+1 + pthreads + dynamic linking (Fix for MAIN_MODULE=1 + pthreads + dynamic linking emscripten-core/emscripten#27799)
• f99d5010b Fix test_futimens_nodefs under Windows (Fix test_futimens_nodefs under Windows emscripten-core/emscripten#27801)
• f2875ce53 Always return the event loop in emscripten_semaphore_async_acquire (Always return the event loop in emscripten_semaphore_async_acquire emscripten-core/emscripten#27798)
• f6916853b Rebaseline tests after accidental checkin of invalid results
• 89631a5b2 Add support for new TARGET_ARCH dylink sub-section (Add support for new TARGET_ARCH dylink sub-section emscripten-core/emscripten#27797)
• 81cdc6d08 [WebGL] Remove surplus argument in glDrawElements ([WebGL] Remove surplus argument in glDrawElements. NFC emscripten-core/emscripten#27793)
• 436755800 [FS] Reduce rounding errors when setting file timestamps ([FS] Reduce rounding errors when setting file timestamps emscripten-core/emscripten#27794)

Presumably it must be emscripten-core/emscripten#27794.

Looks like the fix for this was published in 6.0.13?

Yeah. There was another problem in 6.0.13 but we fixed that on our side in #160264 so I am pretty sure it will be green now.

Opened #160854.

I also updated my tip-of-tree CI so that it will send me an email if it goes from green to red, so hopefully next time I can catch any regression before it gets to a published Emscripten version. We should also explore adding a test to Emscripten's CI. I opened an issue on Emscripten about it:
emscripten-core/emscripten#28097