Make pub strong-mode clean.

Make pub strong-mode clean.

R=rnystrom@google.com

Committed: https://github.com/dart-lang/pub/commit/391b445dc6b28958794658b18caedb69d8cf8719

[Bob Nystrom, 2016-07-29 00:31:22]

The Future<...> type arguments to schedule() look wrong to me. Since schedule()
returns a Future<T>, it means you'll get Future<Future<T>>, which is probably
not going to do nice things.

Otherwise, LGTM.

https://codereview.chromium.org/2184303002/diff/1/lib/src/ascii_tree.dart
File lib/src/ascii_tree.dart (right):

https://codereview.chromium.org/2184303002/diff/1/lib/src/ascii_tree.dart#new...
lib/src/ascii_tree.dart:131: bool isLast: false}) {
Did dartfmt do this? If so, can you file a bug. It's icky.

https://codereview.chromium.org/2184303002/diff/1/lib/src/lock_file.dart
File lib/src/lock_file.dart (right):

https://codereview.chromium.org/2184303002/diff/1/lib/src/lock_file.dart#newc...
lib/src/lock_file.dart:186: var packages = new Map<String,
PackageId>.from(this.packages);
I would expect this to be inferred from this.packages. Is it not? Is there a bug
to track this?

https://codereview.chromium.org/2184303002/diff/1/test/descriptor/git.dart
File test/descriptor/git.dart (right):

https://codereview.chromium.org/2184303002/diff/1/test/descriptor/git.dart#ne...
test/descriptor/git.dart:46: return schedule/*<Future<String>>*/(() async {
Shouldn't this just be <String>?

https://codereview.chromium.org/2184303002/diff/1/test/descriptor/tar.dart
File test/descriptor/tar.dart (right):

https://codereview.chromium.org/2184303002/diff/1/test/descriptor/tar.dart#ne...
test/descriptor/tar.dart:22: return schedule/*<Future<String>>*/(() async {
Ditto.

[nweiz, 2016-08-01 19:59:15]

> The Future<...> type arguments to schedule() look wrong to 
> me. Since schedule() returns a Future<T>, it means you'll
> get Future<Future<T>>, which is probably not going to do
> nice things.

This is actually necessary to get maybe-future methods working right with strong
mode. I don't know the details, but Leaf told me to annotate stuff as
Future<Future<T>> and that seems to make everything happy.

https://codereview.chromium.org/2184303002/diff/1/lib/src/ascii_tree.dart
File lib/src/ascii_tree.dart (right):

https://codereview.chromium.org/2184303002/diff/1/lib/src/ascii_tree.dart#new...
lib/src/ascii_tree.dart:131: bool isLast: false}) {
On 2016/07/29 00:31:22, Bob Nystrom wrote:
> Did dartfmt do this? If so, can you file a bug. It's icky.

No, I did. Changed to formatter style.

https://codereview.chromium.org/2184303002/diff/1/lib/src/lock_file.dart
File lib/src/lock_file.dart (right):

https://codereview.chromium.org/2184303002/diff/1/lib/src/lock_file.dart#newc...
lib/src/lock_file.dart:186: var packages = new Map<String,
PackageId>.from(this.packages);
On 2016/07/29 00:31:22, Bob Nystrom wrote:
> I would expect this to be inferred from this.packages. Is it not? Is there a
bug
> to track this?

A lot of .from() factory constructors don't use generic annotations for some
reason (see
https://github.com/dart-lang/dev_compiler/blob/master/tool/input_sdk/lib/core...).
Last I talked to Lasse about it, it sounded like it was by design, although I
can't say I understand the benefit.

[nweiz, 2016-08-01 19:59:33]

Code review changes

[nweiz, 2016-08-01 19:59:59]

Description was changed from

==========
Make pub strong-mode clean.

R=rnystrom@google.com
==========

to

==========
Make pub strong-mode clean.

R=rnystrom@google.com

Committed:
https://github.com/dart-lang/pub/commit/391b445dc6b28958794658b18caedb69d8cf8719
==========

[nweiz, 2016-08-01 20:00:01]

Committed patchset #2 (id:20001) manually as
391b445dc6b28958794658b18caedb69d8cf8719 (presubmit successful).