Strong mode inference tests.

Strong mode inference tests.

Add some tests for local inference.  Some of these are marked failing
since they test upcoming changes to inference that haven't landed yet.

BUG=
R=jmesserly@google.com

Committed: https://github.com/dart-lang/sdk/commit/e58515bfa162405e48f698b8805fbb0c51ef0e85

[Leaf, 2017-01-24 18:12:26]

leafp@google.com changed reviewers:
+ jmesserly@google.com

[Leaf, 2017-01-24 18:12:26]

Jen - These are tests that I wrote to help me clarify in my head the changes in
your CL.  Most of them pass with your CL, but I think one or two of them fail
and should either be fixed, or explained to me why my expectations are wrong...
:)

I'd suggest we land these, and then your CL should switch the failing ones to
non-failing.

WDYT?

[Jennifer Messerly, 2017-01-24 18:22:10]

On 2017/01/24 18:12:26, Leaf wrote:
> Jen - These are tests that I wrote to help me clarify in my head the changes
in
> your CL.  Most of them pass with your CL, but I think one or two of them fail
> and should either be fixed, or explained to me why my expectations are
wrong...
> :)
> 
> I'd suggest we land these, and then your CL should switch the failing ones to
> non-failing.
> 
> WDYT?

looks great! Do you mind if I merge this into my inference CL and get to work on
those failures?

[Leaf, 2017-01-24 18:28:35]

On 2017/01/24 18:22:10, Jennifer Messerly wrote:
> On 2017/01/24 18:12:26, Leaf wrote:
> > Jen - These are tests that I wrote to help me clarify in my head the changes
> in
> > your CL.  Most of them pass with your CL, but I think one or two of them
fail
> > and should either be fixed, or explained to me why my expectations are
> wrong...
> > :)
> > 
> > I'd suggest we land these, and then your CL should switch the failing ones
to
> > non-failing.
> > 
> > WDYT?
> 
> looks great! Do you mind if I merge this into my inference CL and get to work
on
> those failures?

That's fine.  What do you think about me just landing this CL, and then you will
just pick them up?  I think that may be a little simpler then juggling a bunch
of unlanded CLs, particularly if I want to add another round of these .

[Jennifer Messerly, 2017-01-24 19:03:35]

lgtm

https://codereview.chromium.org/2653833003/diff/1/pkg/analyzer/test/generated...
File pkg/analyzer/test/generated/strong_mode_test.dart (right):

https://codereview.chromium.org/2653833003/diff/1/pkg/analyzer/test/generated...
pkg/analyzer/test/generated/strong_mode_test.dart:97: source,
[StaticTypeWarningCode.TYPE_ARGUMENT_NOT_MATCHING_BOUNDS]);
might be worth a comment that the error is what we're missing, and that
inference is not expected to handle bounds such as on "f" above

https://codereview.chromium.org/2653833003/diff/1/pkg/analyzer/test/generated...
pkg/analyzer/test/generated/strong_mode_test.dart:156:
_isInstantiationOf(_hasElement(elementB))([_isDynamic])(type);
FYI: we don't pick "dynamic" in case of inference error, because that typically
causes "down cast composite" warnings that confuse users. We intentionally pick
a type that is more likely to place blame an argument, so the user gets a
narrower error scope they can more easily take action on.

whether that heuristic is right, i'm not sure, but we've had some user
experience with it

[Leaf, 2017-01-24 19:23:31]

https://codereview.chromium.org/2653833003/diff/1/pkg/analyzer/test/generated...
File pkg/analyzer/test/generated/strong_mode_test.dart (right):

https://codereview.chromium.org/2653833003/diff/1/pkg/analyzer/test/generated...
pkg/analyzer/test/generated/strong_mode_test.dart:97: source,
[StaticTypeWarningCode.TYPE_ARGUMENT_NOT_MATCHING_BOUNDS]);
On 2017/01/24 19:03:35, Jennifer Messerly wrote:
> might be worth a comment that the error is what we're missing, and that
> inference is not expected to handle bounds such as on "f" above

I think that's already in the comment, and in the expectation below.  The
comment says that we're testing that propogation does not happen, and the
expectation is that you get dynamic.

https://codereview.chromium.org/2653833003/diff/1/pkg/analyzer/test/generated...
pkg/analyzer/test/generated/strong_mode_test.dart:156:
_isInstantiationOf(_hasElement(elementB))([_isDynamic])(type);
On 2017/01/24 19:03:35, Jennifer Messerly wrote:
> FYI: we don't pick "dynamic" in case of inference error, because that
typically
> causes "down cast composite" warnings that confuse users. We intentionally
pick
> a type that is more likely to place blame an argument, so the user gets a
> narrower error scope they can more easily take action on.
> 
> whether that heuristic is right, i'm not sure, but we've had some user
> experience with it

Are you saying the expectation of B<dynamic> is wrong here?  I'm pretty sure
that's what showed up.  But in any case, if not, then feel free to change it
when you enable it in your CL.

[Leaf, 2017-01-24 19:24:05]

Description was changed from

==========
Strong mode inference tests.

Add some tests for local inference.  Some of these are marked failing
since they test upcoming changes to inference that haven't landed yet.

BUG=
==========

to

==========
Strong mode inference tests.

Add some tests for local inference.  Some of these are marked failing
since they test upcoming changes to inference that haven't landed yet.

BUG=
R=jmesserly@google.com

Committed:
https://github.com/dart-lang/sdk/commit/e58515bfa162405e48f698b8805fbb0c51ef0e85
==========

[Leaf, 2017-01-24 19:24:06]

Committed patchset #1 (id:1) manually as
e58515bfa162405e48f698b8805fbb0c51ef0e85 (presubmit successful).