Add a leg_only test suite, and make sure to run it in our presubmit script.

Add a leg_only test suite, and make sure to run it in our presubmit script.
Committed: https://code.google.com/p/dart/source/detail?r=1303

[kasperl, 2011-11-08 11:34:13]

LGTM.

http://codereview.chromium.org/8496009/diff/1/frog/presubmit.py
File frog/presubmit.py (right):

http://codereview.chromium.org/8496009/diff/1/frog/presubmit.py#newcode86
frog/presubmit.py:86: test_cmd.extend(['language', 'corelib', 'leg', 'isolate',
'peg', 'leg_only'])
Long line.

[kasperl, 2011-11-08 11:41:09]

http://codereview.chromium.org/8496009/diff/1/frog/tests/leg_only/leg_only.st...
File frog/tests/leg_only/leg_only.status (right):

http://codereview.chromium.org/8496009/diff/1/frog/tests/leg_only/leg_only.st...
frog/tests/leg_only/leg_only.status:5: prefix leg
Should this be leg_only? I guess so.

http://codereview.chromium.org/8496009/diff/1/frog/tests/leg_only/leg_only.st...
frog/tests/leg_only/leg_only.status:8: *: Fail # Too many type errors in frog.
Maybe you could remove the commented out code in the tests in your next CL and
mark them as failing?

[ngeoffray, 2011-11-08 11:46:11]

Thanks!

http://codereview.chromium.org/8496009/diff/1/frog/presubmit.py
File frog/presubmit.py (right):

http://codereview.chromium.org/8496009/diff/1/frog/presubmit.py#newcode86
frog/presubmit.py:86: test_cmd.extend(['language', 'corelib', 'leg', 'isolate',
'peg', 'leg_only'])
On 2011/11/08 11:34:13, kasperl wrote:
> Long line.

Done.

http://codereview.chromium.org/8496009/diff/1/frog/tests/leg_only/leg_only.st...
File frog/tests/leg_only/leg_only.status (right):

http://codereview.chromium.org/8496009/diff/1/frog/tests/leg_only/leg_only.st...
frog/tests/leg_only/leg_only.status:5: prefix leg
On 2011/11/08 11:41:09, kasperl wrote:
> Should this be leg_only? I guess so.

Good catch.

http://codereview.chromium.org/8496009/diff/1/frog/tests/leg_only/leg_only.st...
frog/tests/leg_only/leg_only.status:8: *: Fail # Too many type errors in frog.
On 2011/11/08 11:41:09, kasperl wrote:
> Maybe you could remove the commented out code in the tests in your next CL and
> mark them as failing?

I forgot to update this file after svn copying the one in tests/leg. So the
comment is not right. Removed.

[ngeoffray, 2011-11-08 12:12:22]

PTAL, I had to make test.py to be more flexible with flags. The previous
solution was affecting all components, not just the configuration :)

Also removed comments in tests, and update status file.

[kasperl, 2011-11-08 12:21:34]

LGTM.

http://codereview.chromium.org/8496009/diff/1028/frog/tests/leg_only/leg_only...
File frog/tests/leg_only/leg_only.status (right):

http://codereview.chromium.org/8496009/diff/1028/frog/tests/leg_only/leg_only...
frog/tests/leg_only/leg_only.status:7: [ $component == leg || $component == frog
|| $component == frogsh ]
Are these really supposed to fail when running with frog/frogsh? Somehow that
seems strange to me.

[ngeoffray, 2011-11-08 12:28:11]

http://codereview.chromium.org/8496009/diff/1028/frog/tests/leg_only/leg_only...
File frog/tests/leg_only/leg_only.status (right):

http://codereview.chromium.org/8496009/diff/1028/frog/tests/leg_only/leg_only...
frog/tests/leg_only/leg_only.status:7: [ $component == leg || $component == frog
|| $component == frogsh ]
On 2011/11/08 12:21:34, kasperl wrote:
> Are these really supposed to fail when running with frog/frogsh? Somehow that
> seems strange to me.

That's because the test suite adds leg_only, so frog and frogs will also fail.

I don't want to make leg leg_only, as it will not help us find errors when we
should bail out.

I don't want to make frog and frogsh enable_leg by default, because this would
affect frog developers.

So having the status file this way was the best option for me.