Update expectations files to get more strong mode inference tests to pass.

Update expectations files to get more strong mode inference tests to pass.

The logic to make these tests pass was already in place, but their
expectations files either didn't exist or needed tweaking.

R=scheglov@google.com, sigmund@google.com

Committed: https://github.com/dart-lang/sdk/commit/fe991b20dd8795950239e81a9ed966502a6f3ebd

[Paul Berry, 2017-05-12 18:05:37]

paulberry@google.com changed reviewers:
+ scheglov@google.com, sigmund@google.com

[scheglov, 2017-05-12 18:08:12]

LGTM

[Siggi Cherem (dart-lang), 2017-05-12 18:16:15]

lgtm - I'm not sure why didn't we see these failures before locally or on the
bots?

[Paul Berry, 2017-05-12 18:20:04]

On 2017/05/12 18:16:15, Siggi Cherem (dart-lang) wrote:
> lgtm - I'm not sure why didn't we see these failures before locally or on the
> bots?

It's because the expectation files were wrong, and the status files said "Fail",
so the tests failed as expected.  Unfortunately this is one of the drawbacks of
the "golden file" approach--when we fix a test but the expectation file has a
problem, we don't get any feedback from the bot.

(I discovered these tests by re-running the tests with -DupdateExpectations=true
-DskipVm=true, which caused the tests to pass.)

Incidentally, I'm planning to have a conversation with Peter sometime next week
to see if we can add more types of expectations to the status file so that when
a test starts failing for a different reason, we notice.

[Paul Berry, 2017-05-12 18:20:24]

Description was changed from

==========
Update expectations files to get more strong mode inference tests to pass.

The logic to make these tests pass was already in place, but their
expectations files either didn't exist or needed tweaking.
==========

to

==========
Update expectations files to get more strong mode inference tests to pass.

The logic to make these tests pass was already in place, but their
expectations files either didn't exist or needed tweaking.

R=scheglov@google.com, sigmund@google.com

Committed:
https://github.com/dart-lang/sdk/commit/fe991b20dd8795950239e81a9ed966502a6f3ebd
==========

[Paul Berry, 2017-05-12 18:20:25]

Committed patchset #1 (id:1) manually as
fe991b20dd8795950239e81a9ed966502a6f3ebd (presubmit successful).

[Siggi Cherem (dart-lang), 2017-05-12 18:35:26]

On 2017/05/12 18:20:04, Paul Berry wrote:
> On 2017/05/12 18:16:15, Siggi Cherem (dart-lang) wrote:
> > lgtm - I'm not sure why didn't we see these failures before locally or on
the
> > bots?
> 
> It's because the expectation files were wrong, and the status files said
"Fail",
> so the tests failed as expected.  Unfortunately this is one of the drawbacks
of
> the "golden file" approach--when we fix a test but the expectation file has a
> problem, we don't get any feedback from the bot.
> 
> (I discovered these tests by re-running the tests with
-DupdateExpectations=true
> -DskipVm=true, which caused the tests to pass.)
> 
> Incidentally, I'm planning to have a conversation with Peter sometime next
week
> to see if we can add more types of expectations to the status file so that
when
> a test starts failing for a different reason, we notice.

that makes sense - +1 on adding more types. My understanding is that
VerificationError is a recent addition, and it should be fairly simple to add
more.