Closed omus closed 2 years ago
Merging #387 (a7ff9f8) into master (bab4798) will not change coverage. The diff coverage is
n/a
.
@@ Coverage Diff @@
## master #387 +/- ##
=======================================
Coverage 95.12% 95.12%
=======================================
Files 35 35
Lines 1743 1743
=======================================
Hits 1658 1658
Misses 85 85
Continue to review full report at Codecov.
Legend - Click here to learn more
Δ = absolute <relative> (impact)
,ø = not affected
,? = missing data
Powered by Codecov. Last update bab4798...a7ff9f8. Read the comment docs.
I've subverted the thread-safe cache to validate that these thread-safety tests still fail with this change.
Tests still pass even without the thread-safe cache. I have managed to produce a thread-safety failure locally so I think the best way to proceed is to merge this PR (with the proper thread-safe cache) and hopefully I can get some additional information in regards to #386
Digging into #386. Seems like these failures are less random than they first appeared. Running the thread tests in a separate Julia process result in stack traces being lost and only allow for speculative debugging. By enabling threads on the CI and only running the tests when threads are enabled we should be able to get useful information to address the root of the problem.