System.Text.Json.SourceGeneration intermittent failures #72101
Labels
area-Codegen-meta-mono
disabled-test
The test is disabled in source code against the issue
os-android
runtime-mono
specific to the Mono runtime
test-failure
Milestone
Comments
greenEkatherine
added
area-System.Text.Json
blocking-clean-ci
|
Tagging subscribers to this area: @dotnet/area-system-text-json, @gregsdennis Issue DetailsAssertions failed in several builds for tests:
Example failure: cc @jtschuster
|
|
The failing assertion according to the build logs is this one: Seems related to #69800. The output does not match the assertion inputs so wondering if this might be a runtime issue. |
|
Also hitting similar assert failures in System.Text.Serialization.Test.ContinuationTests |
|
Always and only Android and WASM. Result is NaN. Seems likely a Mono issue. |
|
@marek-safar @steveisok thoughts about the right area path for this? It seems that it's only Android and WASM. It would be unlikely that there is JSON bug specific to those, especially relating to numerical operations. |
|
Tagging subscribers to 'arch-android': @steveisok, @akoeplinger Issue DetailsAssertions failed in several builds for tests:
Example failure: cc @jtschuster Runfo Tracking Issue: System.Text.Json failures on Mono_Release
Displaying 100 of 941 results Build Result Summary
|
|
@danmoseley For android, there is a known floating point issue that would be fixed by #65723. This only impacts x86 and is not very high on the priority list. I think where we hit this on Android x86 we should skip. |
|
/cc @lewing as he may have a better idea for wasm. |
|
This looks like codegen bug in MonoVM but it's strange that it's NaN instead of 1.1. @lewing @SamMonoRT could you assign this? |
|
It'd be great if someone could add a skip for x86 Android as you suggest. |
|
The wasm failures look different from the android ones, i.e. the test runner seems to time out. |
|
most if not all of the wasm failures are in the threaded runtimes so those failures are not something to worry about at the moment |
|
Yeah, the WASM ones are clearly totally different - they are hangs. @lewing are you saying we should disable these tests for wasm too? that sounds good to me. |
|
@danmoseley we do want to test json on wasm but the threaded runtime lanes are just being brought up so we expect breakage there. @simonrozsival can you work with @radical to disable the tests for threaded lanes? |
|
@simonrozsival can you extract the wasm part into a separate issue? A tracking issue for all the threaded lane failures will be fine too. |
|
@lewing can someone on your team disable the tests ASAP please? In avergae it is 36 failures per day across PRs and Official runs -- lots of noise in Test monitoring. |
|
Great, thanks @lewing! |
|
@radical I'll look into it on Monday |
|
The latest runtime-extra-platforms rolling build doesn't have any System.Text.Json failures, so I will remove the blocking-clean-ci tag |
tlakollo
removed
the
blocking-clean-ci
simonrozsival
added
the
disabled-test
lewing
removed
arch-wasm
|
The failing tests have been disabled and there are tracking issues for the actual failures, closing this |
This was referenced Jul 26, 2022
ghost
locked as resolved and limited conversation to collaborators
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
Frequency:
Assertions failed in several builds for tests:
Example failure:
Assert.Equal() Failure Expected: 1.10000002 Actual: NaN log results
cc @jtschuster
Runfo Tracking Issue: System.Text.Json failures on Mono_Release
Displaying 100 of 1291 results
Build Result Summary
The text was updated successfully, but these errors were encountered: