fixes #231, mixin application (aka ClassTypeAlias) implicit ctors

fixes #231, mixin application (aka ClassTypeAlias) implicit ctors

R=vsm@google.com

Committed: https://github.com/dart-lang/dev_compiler/commit/96e7827f97698bdd4a44aa341ebc80d4adc4d9b1

[Jennifer Messerly, 2015-06-19 23:00:21]

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

[vsm, 2015-06-19 23:20:40]

lgtm

https://codereview.chromium.org/1193083002/diff/20001/lib/src/codegen/js_code...
File lib/src/codegen/js_codegen.dart (right):

https://codereview.chromium.org/1193083002/diff/20001/lib/src/codegen/js_code...
lib/src/codegen/js_codegen.dart:332: // If we've already emitted this class,
skip it.
Is this comment still valid?

https://codereview.chromium.org/1193083002/diff/20001/lib/src/codegen/js_code...
lib/src/codegen/js_codegen.dart:342: var fun = js.call('function() {
super.#(...#); }', [
It'd be nice to have the args here to show the actual API.

[Jennifer Messerly, 2015-06-22 15:55:44]

Thanks!

https://codereview.chromium.org/1193083002/diff/20001/lib/src/codegen/js_code...
File lib/src/codegen/js_codegen.dart (right):

https://codereview.chromium.org/1193083002/diff/20001/lib/src/codegen/js_code...
lib/src/codegen/js_codegen.dart:332: // If we've already emitted this class,
skip it.
On 2015/06/19 23:20:40, vsm wrote:
> Is this comment still valid?

Good catch, removed (from 3 places!)

https://codereview.chromium.org/1193083002/diff/20001/lib/src/codegen/js_code...
lib/src/codegen/js_codegen.dart:342: var fun = js.call('function() {
super.#(...#); }', [
On 2015/06/19 23:20:40, vsm wrote:
> It'd be nice to have the args here to show the actual API.

Agreed. However, there are a few reasons for not doing that:
* we don't have a FormalParameterList here, just the element model, so can't
reuse existing visit* code.
* we don't have an ArgumentList, so we'd have to emit parameters in argument
position. Happens to be valid at the moment, probably won't be when we have
destructuring.
* it's less error prone to do forward all arguments this way. It avoids obscure
bugs that could happen, like "default value is incorrect in mixin application
constructor super calls" or "named arguments don't work in mixin application" :)

In the very long term I'm in favor of this kind of readability tweak, but I
think we have a lot of higher priority things to do first.

[Jennifer Messerly, 2015-06-22 15:56:01]

Committed patchset #3 (id:40001) manually as
96e7827f97698bdd4a44aa341ebc80d4adc4d9b1 (presubmit successful).