Deprecate --fast option
#8970
Deprecate --fast option
#8970
Conversation
|
To explain the meaning behind my shipit above: Twitter will be ready for this. |
Eric-Arellano
changed the title
--no-fast and deprecate --fast option--fast option
Eric-Arellano
force-pushed
the
remove-fast
branch
from
cfe14c8
to
8a0c761
Compare
| @@ -848,7 +843,6 @@ def test_coverage_issue_5314_all_source_roots(self): | |||
| self.assertEqual([1, 2, 5, 6], all_statements) | |||
| self.assertEqual([2], not_run_statements) | |||
|
|
|||
| @ensure_cached(PytestRun, expected_num_artifacts=1) | |||
There was a problem hiding this comment.
I removed this because @ensure_cached is hardcoded to check that there is only one cache_subdir. With sharding, we have two subdirs. Rewriting TaskTestBase to allow for multiple cache_subdirs would be a lot of work for this legacy code.
| def test_coverage_auto_option_mixed_multiple_targets(self): | ||
| all_statements, not_run_statements = self.run_coverage_auto(targets=[self.green, self.red], | ||
| failed_targets=[self.red]) | ||
| self.assertEqual([1, 2, 5, 6], all_statements) | ||
| self.assertEqual([], not_run_statements) | ||
| self.assertEqual([2], not_run_statements) |
There was a problem hiding this comment.
I couldn't figure this one out.
This is now the same output as test_coverage_auto_option_red. I suspect what's happening is that self.green says every statement ran, but self.red says statement 2 did not, so the combined result says statement 2 did not run?
cc @jsirois, do you recall if V1 Pytest --no-fast will attempt to merge coverage results? I know that's a tricky task @TansyArron has spent much time on for V2.
Eric-Arellano
force-pushed
the
remove-fast
branch
from
8a0c761
to
c3f4cce
Compare
Eric-Arellano
force-pushed
the
remove-fast
branch
from
c3f4cce
to
f629b8a
Compare
Eric-Arellano
requested review from
pierrechevalier83,
illicitonion and
stuhood
|
Finally got this green! Ready for review. |
| @@ -184,10 +183,7 @@ def test_unit_tests_with_different_sets(self): | |||
| ]) | |||
| self.assert_success(run) | |||
|
|
|||
| @expectedFailure | |||
This change is meant to give a nudge to users in the direction of the V2 test runner. By having them switch to
--no-fastnow, we lower the barrier to entry to the V2 test runner, which must run with--no-fast.This is meant to avoid a Python 2 vs. Python 3 situation where the cost to migrate from the V1 test runner to the V2 implementation is too costly such that users stick with V1 and the Pants community now bifurcates.