Migrate block 116.

Migrate block 116.

Interesting bits:

- Removed some behavior that is now unreachable in the presence of
  type errors.
- Made if_null_behavior_test not a multitest since it doesn't need to
  be any more.
- Likewise, if_null_evaluation_order_test never needed to be a
  multitest.

R=lrn@google.com

Committed: https://github.com/dart-lang/sdk/commit/a719910d78357557bca925822d938753d9de1a62

[Bob Nystrom, 2017-08-25 23:28:31]

rnystrom@google.com changed reviewers:
+ lrn@google.com

[Bob Nystrom, 2017-08-29 00:18:51]

Ping!

[Lasse Reichstein Nielsen, 2017-08-29 07:43:40]

lgtm

https://codereview.chromium.org/3003933002/diff/1/tests/language_2/hidden_imp...
File tests/language_2/hidden_import_test.dart (right):

https://codereview.chromium.org/3003933002/diff/1/tests/language_2/hidden_imp...
tests/language_2/hidden_import_test.dart:15: new Future(); //# 01: static type
warning
Why is there type warning?

I assumed that Dart 2 removed all warnings.

https://codereview.chromium.org/3003933002/diff/1/tests/language_2/identical_...
File tests/language_2/identical_closure2_test.dart (right):

https://codereview.chromium.org/3003933002/diff/1/tests/language_2/identical_...
tests/language_2/identical_closure2_test.dart:12: }
Point class isn't used.

https://codereview.chromium.org/3003933002/diff/1/tests/language_2/identical_...
tests/language_2/identical_closure2_test.dart:14: main() {
Add comment and/or TODO that this doesn't work when compiled to JS.

The numbers seem to be picked explicitly to not work in JS.

https://codereview.chromium.org/3003933002/diff/1/tests/language_2/if_null_pr...
File tests/language_2/if_null_precedence_test.dart (right):

https://codereview.chromium.org/3003933002/diff/1/tests/language_2/if_null_pr...
tests/language_2/if_null_precedence_test.dart:25: Expect.equals(2, true ?? 1 ? 2
: 3);
Should this be a type problem in Dart 2? 
Or is it an implicit downcast from Object to bool of (true ?? 1)?

https://codereview.chromium.org/3003933002/diff/1/tests/language_2/language_2...
File tests/language_2/language_2_kernel.status (right):

https://codereview.chromium.org/3003933002/diff/1/tests/language_2/language_2...
tests/language_2/language_2_kernel.status:67: if_null_evaluation_order_test:
Crash # VM does not support BottomType
Worrisome because there shouldn't be any "bottom" type in that test. The type of
the `null` literal is Null these days, not bottom (unlike the type of `throw 0`
which is still non-nullable bottom).

[Bob Nystrom, 2017-08-29 22:50:53]

Description was changed from

==========
Migrate block 116.

Interesting bits:

- Removed some behavior that is now unreachable in the presence of
  type errors.
- Made if_null_behavior_test not a multitest since it doesn't need to
  be any more.
- Likewise, if_null_evaluation_order_test never needed to be a
  multitest.
==========

to

==========
Migrate block 116.

Interesting bits:

- Removed some behavior that is now unreachable in the presence of
  type errors.
- Made if_null_behavior_test not a multitest since it doesn't need to
  be any more.
- Likewise, if_null_evaluation_order_test never needed to be a
  multitest.

R=lrn@google.com

Committed:
https://github.com/dart-lang/sdk/commit/a719910d78357557bca925822d938753d9de1a62
==========

[Bob Nystrom, 2017-08-29 22:50:55]

Committed patchset #1 (id:1) manually as
a719910d78357557bca925822d938753d9de1a62.

[Bob Nystrom, 2017-08-30 18:21:23]

https://codereview.chromium.org/3003933002/diff/1/tests/language_2/hidden_imp...
File tests/language_2/hidden_import_test.dart (right):

https://codereview.chromium.org/3003933002/diff/1/tests/language_2/hidden_imp...
tests/language_2/hidden_import_test.dart:15: new Future(); //# 01: static type
warning
On 2017/08/29 07:43:40, Lasse Reichstein Nielsen wrote:
> Why is there type warning?

<shrug> There's no change in this test. I think we've warned on shadowing a
dart: imported name for a long time.

> I assumed that Dart 2 removed all warnings.

It removes almost all, but it looks like there are a couple of things that
remain warnings. Another example is a getter and setter with the same name whose
property don't match. It's not an error because they don't need to match for
soundness, but the warning is still there.

https://codereview.chromium.org/3003933002/diff/1/tests/language_2/identical_...
File tests/language_2/identical_closure2_test.dart (right):

https://codereview.chromium.org/3003933002/diff/1/tests/language_2/identical_...
tests/language_2/identical_closure2_test.dart:12: }
On 2017/08/29 07:43:40, Lasse Reichstein Nielsen wrote:
> Point class isn't used.

Done.

https://codereview.chromium.org/3003933002/diff/1/tests/language_2/identical_...
tests/language_2/identical_closure2_test.dart:14: main() {
On 2017/08/29 07:43:40, Lasse Reichstein Nielsen wrote:
> Add comment and/or TODO that this doesn't work when compiled to JS.

Done. Also, there are status entries disabling this on dart2js and dartdevc, at
least one of which mentions the bigints bug.

> The numbers seem to be picked explicitly to not work in JS.

Yeah. My guess is they wanted to test that identical is correct in bigints on
the VM?

https://codereview.chromium.org/3003933002/diff/1/tests/language_2/if_null_pr...
File tests/language_2/if_null_precedence_test.dart (right):

https://codereview.chromium.org/3003933002/diff/1/tests/language_2/if_null_pr...
tests/language_2/if_null_precedence_test.dart:25: Expect.equals(2, true ?? 1 ? 2
: 3);
On 2017/08/29 07:43:40, Lasse Reichstein Nielsen wrote:
> Should this be a type problem in Dart 2? 
> Or is it an implicit downcast from Object to bool of (true ?? 1)?

Yes, I believe that's right. You get Object as the LUB of true and 1, then
downcast in the Boolean conversion for ?:.

https://codereview.chromium.org/3003933002/diff/1/tests/language_2/language_2...
File tests/language_2/language_2_kernel.status (right):

https://codereview.chromium.org/3003933002/diff/1/tests/language_2/language_2...
tests/language_2/language_2_kernel.status:67: if_null_evaluation_order_test:
Crash # VM does not support BottomType
On 2017/08/29 07:43:40, Lasse Reichstein Nielsen wrote:
> Worrisome because there shouldn't be any "bottom" type in that test. The type
of
> the `null` literal is Null these days, not bottom (unlike the type of `throw
0`
> which is still non-nullable bottom).

Oops, that was a mistake on my part. This is passing now. I just didn't update
the status correctly. (See the next line.)