[commit: ghc] master: Always run explicitly requested ways (extra_ways) for fast runs. (b89c491)

git at git.haskell.org git at git.haskell.org
Sun Sep 20 10:41:59 UTC 2015


Repository : ssh://git@git.haskell.org/ghc

On branch  : master
Link       : http://ghc.haskell.org/trac/ghc/changeset/b89c49138fcd62a1175d67fac62d59ac2497bfdd/ghc

>---------------------------------------------------------------

commit b89c49138fcd62a1175d67fac62d59ac2497bfdd
Author: Edward Z. Yang <ezyang at cs.stanford.edu>
Date:   Sun Sep 20 12:16:50 2015 +0200

    Always run explicitly requested ways (extra_ways) for fast runs.
    
    To keep validates fast, we only one run one way.  But I think that
    it's important for some tests to run them a few ways, just to
    make sure functionality, e.g. the profiler, is working.  This commit
    changes the logic so that any way specified in extra_ways is always
    run for fast.  The big changes is now profiling tests are run on
    validate.
    
    I also made it so the G1 garbage collector tests only run on slow.
    
    Signed-off-by: Edward Z. Yang <ezyang at cs.stanford.edu>
    
    Test Plan: validate
    
    Reviewers: austin, thomie, bgamari
    
    Reviewed By: austin, thomie, bgamari
    
    Subscribers: thomie
    
    Differential Revision: https://phabricator.haskell.org/D1251


>---------------------------------------------------------------

b89c49138fcd62a1175d67fac62d59ac2497bfdd
 Makefile                                 | 7 +++++--
 testsuite/driver/testlib.py              | 9 ++++++++-
 testsuite/tests/codeGen/should_run/all.T | 2 +-
 3 files changed, 14 insertions(+), 4 deletions(-)

diff --git a/Makefile b/Makefile
index 38c7eb9..348375f 100644
--- a/Makefile
+++ b/Makefile
@@ -191,12 +191,15 @@ endif
 #           cd tests  config.  many   many  by
 # validate  && make   speed=   tests  ways  whom
 # =============================================================================
-# --fast    fast      2        some   1     Travis (to stay within time limit)
-# --normal  test      1        all    1     Phabricator (slow takes too long?)
+# --fast    fast      2        some   1+exs Travis (to stay within time limit)
+# --normal  test      1        all    1+exs Phabricator (slow takes too long?)
 # --slow    slow      0        all    all   Nightly (slow is ok)
 #
 #           accept    1        all    1
 #
+# `--fast` and `--normal` run one default way, as well as any other ways which
+# are explicitly requested by the test using extra_ways().
+#
 # `make accept` should run all tests exactly once. There is no point in
 # accepting a test for multiple ways, since it should produce the same output
 # for all ways.
diff --git a/testsuite/driver/testlib.py b/testsuite/driver/testlib.py
index 1700392..34a3fb8 100644
--- a/testsuite/driver/testlib.py
+++ b/testsuite/driver/testlib.py
@@ -676,8 +676,15 @@ def test_common_work (name, opts, func, args):
 
         # Only run all ways in slow mode.
         # See Note [validate and testsuite speed] in toplevel Makefile.
-        if config.speed > 0:
+        if config.accept:
+            # Only ever run one way
             do_ways = do_ways[:1]
+        elif config.speed > 0:
+            # However, if we EXPLICITLY asked for a way (with extra_ways)
+            # please test it!
+            explicit_ways = filter(lambda way: way in opts.extra_ways, do_ways)
+            other_ways = filter(lambda way: way not in opts.extra_ways, do_ways)
+            do_ways = other_ways[:1] + explicit_ways
 
         if not config.clean_only:
             # Run the required tests...
diff --git a/testsuite/tests/codeGen/should_run/all.T b/testsuite/tests/codeGen/should_run/all.T
index f9e0e0e..ab2ce60 100644
--- a/testsuite/tests/codeGen/should_run/all.T
+++ b/testsuite/tests/codeGen/should_run/all.T
@@ -1,5 +1,5 @@
 # Test +RTS -G1 here (it isn't tested anywhere else)
-setTestOpts(extra_ways(['g1']))
+setTestOpts(unless(fast(), extra_ways(['g1'])))
 
 test('cgrun001', normal, compile_and_run, [''])
 test('cgrun002', normal, compile_and_run, [''])



More information about the ghc-commits mailing list