Simon Peyton Jones
Friends
-dcore-lint is switched off in our compile-time tests testsuite/tests/perf/compiler
But I'm making some changes whose primary effect will be on Lint. How can I get a before-vs-after analysis of the tess in perf/compiler, with -dcore-lint switched on?
I'm afraid we don't have a good way of doing that without modifying the .T file, which explicitly disables linting. The following patch will drop the appropriate line: diff --git a/testsuite/tests/perf/compiler/all.T b/testsuite/tests/perf/compiler/all.T index 82390ff27e0..61092d2e1e1 100644 --- a/testsuite/tests/perf/compiler/all.T +++ b/testsuite/tests/perf/compiler/all.T @@ -1,7 +1,3 @@ -# Tests that call 'collect_compiler_stats' are skipped when debugging is on. -# See testsuite/driver/testlib.py. -setTestOpts(no_lint) - test('T1969', [# expect_broken(12437), collect_compiler_residency(20), With this patch, linting will be enabled for the perf/compiler tests (since the default testsuite compiler arguments include -dcore-lint; see Settings.Builders.RunTest.runTestGhcFlags). Concretely, I would: 1. run the testsuite as described on [1] 2. apply the above patch and commit 3. run the testsuite again as described on [1] 4. compare the results using the `perf_notes.py` tool as described on [1] Naturally, I'd be happy to answer any questions that arise. Cheers, - Ben [1] https://gitlab.haskell.org/ghc/ghc/-/wikis/building/running-tests/performanc...