Test container:hashtablez_force_sampling_test fails on OSX 10.14.2

[kkiningh]

Test command:

$ bazel test //absl/container:hashtablez_force_sampling_test
INFO: Invocation ID: e51fd711-88c7-422c-ae24-4f4e90d588f3
INFO: Analysed target //absl/container:hashtablez_force_sampling_test (0 packages loaded, 0 targets configured).
INFO: Found 1 test target...
FAIL: //absl/container:hashtablez_force_sampling_test (see /private/var/tmp/_bazel_kkiningh/ccf5c17c3d69615a2c80e87264add296/execroot/com_google_absl/bazel-out/darwin-fastbuild/testlogs/absl/container/hashtablez_force_sampling_test/test.log)
Target //absl/container:hashtablez_force_sampling_test up-to-date:
  bazel-bin/absl/container/hashtablez_force_sampling_test
INFO: Elapsed time: 0.267s, Critical Path: 0.10s
INFO: 1 process: 1 darwin-sandbox.
INFO: Build completed, 1 test FAILED, 2 total actions
//absl/container:hashtablez_force_sampling_test                          FAILED in 0.1s
  /private/var/tmp/_bazel_kkiningh/ccf5c17c3d69615a2c80e87264add296/execroot/com_google_absl/bazel-out/darwin-fastbuild/testlogs/absl/container/hashtablez_force_sampling_test/test.log

INFO: Build completed, 1 test FAILED, 2 total actions

Log:

$ cat /private/var/tmp/_bazel_kkiningh/ccf5c17c3d69615a2c80e87264add296/execroot/com_google_absl/bazel-out/darwin-fastbuild/testlogs/absl/container/hashtablez_force_sampling_test/test.log
exec ${PAGER:-/usr/bin/less} "$0" || exit 1
Executing tests from //absl/container:hashtablez_force_sampling_test
-----------------------------------------------------------------------------
Running main() from gmock_main.cc
[==========] Running 1 test from 1 test suite.
[----------] Global test environment set-up.
[----------] 1 test from kAbslContainerInternalSampleEverything
[ RUN      ] kAbslContainerInternalSampleEverything.Works
absl/container/internal/hashtablez_force_sampling_test.cc:52: Failure
Value of: samples
Expected: only contains elements that is equal to true
  Actual: { false, false, false } (of type bool [3]), whose element #0 doesn't match
absl/container/internal/hashtablez_force_sampling_test.cc:55: Failure
Value of: HashtablezInfoHandlePeer::IsSampled(Sample())
  Actual: false
Expected: true
[  FAILED  ] kAbslContainerInternalSampleEverything.Works (0 ms)
[----------] 1 test from kAbslContainerInternalSampleEverything (0 ms total)

[----------] Global test environment tear-down
[==========] 1 test from 1 test suite ran. (0 ms total)
[  PASSED  ] 0 tests.
[  FAILED  ] 1 test, listed below:
[  FAILED  ] kAbslContainerInternalSampleEverything.Works

Mac version:

$ sw_vers
ProductName:    Mac OS X
ProductVersion: 10.14.2
BuildVersion:   18C54

abseil commit:

$ git rev-parse HEAD
b312c3cb53a0aad75a85ac2bf57c4a614fbd48d4

[kkiningh]

I see this test is tagged no_test_darwin_x86_64, but it still runs when I run all tests as recommended in the quickstart guide (bazel test //absel/...).

[derekmauro]

Thanks for the report. This test should probably never have been released and the failures should be safe to ignore. I'll look into having it removed.

[JonathanDCohen]

I see this test is tagged no_test_darwin_x86_64, but it still runs when I run all tests as recommended in the quickstart guide (bazel test //absel/...).

That's expected behavior. Those tags are only functional for our own CI scripts, which are unfortunately internal only. You do something like bazel test --build_tag_filters=-no_test_darwin_x86_64 blahblahblah(relevant documentation) in order to skip all the tests which aren't supported on macos.

[derekmauro]

I see this test is tagged no_test_darwin_x86_64, but it still runs when I run all tests as recommended in the quickstart guide (bazel test //absel/...).

That's expected behavior. Those tags are only functional for our own CI scripts, which are unfortunately internal only. You do something like bazel test --build_tag_filters=-no_test_darwin_x86_64 blahblahblah(relevant documentation) in order to skip all the tests which aren't supported on macos.

Actually it's not the expected behavior. Everything we release is supposed to be portable. The library in question is an internal hack that wasn't supposed to be released. I'm removing all the no_test_* filters from our CI scripts. We only use those tags for internal tests.