You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
For an unknown reason, in some runs (not very rarely) the macOS action fails while executing the stress test.
It's unclear whether this also happens in the other operating system actions.
It's very weird that it doesn't always happen, as the stress test seed is fixed - all runs should perform the same exact actions.
One possible explanation could be the ECS having UB in some very specific scenario, like accessing uninitialized memory, and working only when that memory has 'well-behaved' values, like 0, or something.
Probably the first step to solve this would be to improve the error logging on this so that we get more information on what's going on.
The error we get is:
[18:31:55.199] [registry.cpp:46 at] [thread0x2048f4f40] critical: Assertion mTables.contains(id) failed
[18:31:55.200] [registry.cpp:46 at] [thread0x2048f4f40] critical: No such sparse relation table
Stack trace (most recent call first):
[doctest] doctest version is "2.4.11"
#0 0x0000000101443277 at /Users/runner/work/cubos/cubos/build/core/libcubos-core.dylib
[doctest] run with "--help" for options
===============================================================================
../../../core/tests/ecs/stress.cpp:32:
TEST CASE: ecs::* stress test
../../../core/tests/ecs/stress.cpp:32: FATAL ERROR: test case CRASHED: SIGABRT - Abort (abnormal termination) signal
===============================================================================
[doctest] test cases: 18 | 17 passed | 1 failed | 43 skipped
[doctest] assertions: 69703 | 69703 passed | 0 failed |
[doctest] Status: FAILURE!
#1 0x0000000101460d2b at /Users/runner/work/cubos/cubos/build/core/libcubos-core.dylib
#2 0x00000001008dfe17 at /Users/runner/work/cubos/cubos/build/cubos-core-tests
#3 0x0000000100943e13 at /Users/runner/work/cubos/cubos/build/cubos-core-tests
#4 0x00000001007820ef at /Users/runner/work/cubos/cubos/build/cubos-core-tests
#5 0x000000010078419f at /Users/runner/work/cubos/cubos/build/cubos-core-tests
#6 0x000000019c713153 at /usr/lib/dyld
/Users/runner/work/_temp/154d423e-9441-494d-99eb-fe19361ac68c.sh: line 1: 9170 Abort trap: 6 ./cubos-core-tests
Reproduce
Unable to reproduce deterministically as of now.
The text was updated successfully, but these errors were encountered:
Description
For an unknown reason, in some runs (not very rarely) the macOS action fails while executing the stress test.
It's unclear whether this also happens in the other operating system actions.
It's very weird that it doesn't always happen, as the stress test seed is fixed - all runs should perform the same exact actions.
One possible explanation could be the ECS having UB in some very specific scenario, like accessing uninitialized memory, and working only when that memory has 'well-behaved' values, like 0, or something.
Probably the first step to solve this would be to improve the error logging on this so that we get more information on what's going on.
The error we get is:
Reproduce
Unable to reproduce deterministically as of now.
The text was updated successfully, but these errors were encountered: