Support static/global field initializations.

Support static/global field initializations.

Works for literals (except String) and references to other fields.

Committed: https://code.google.com/p/dart/source/detail?r=3392

[kasperl, 2012-01-16 10:44:04]

DBC:

http://codereview.chromium.org/9211005/diff/2001/frog/leg/compile_time_consta...
File frog/leg/compile_time_constants.dart (right):

http://codereview.chromium.org/9211005/diff/2001/frog/leg/compile_time_consta...
frog/leg/compile_time_constants.dart:57: new CompileTimeConstantEvaluator(this,
definitions, compiler);
4 space indent.

http://codereview.chromium.org/9211005/diff/2001/frog/leg/compile_time_consta...
frog/leg/compile_time_constants.dart:106: final CompileTimeConstantHandler
compileTimeConstantHandler;
Maybe shorten the name to constantHandler? The context makes it obvious that
we're talking about compile time.

[ngeoffray, 2012-01-16 14:03:35]

LGTM!

http://codereview.chromium.org/9211005/diff/8001/frog/leg/compile_time_consta...
File frog/leg/compile_time_constants.dart (right):

http://codereview.chromium.org/9211005/diff/8001/frog/leg/compile_time_consta...
frog/leg/compile_time_constants.dart:131: // TODO(floitsch): make sure the field
is final.
You have the modifiers, this should be easy to check.

[floitsch, 2012-01-16 15:07:16]

PTAL.

http://codereview.chromium.org/9211005/diff/2001/frog/leg/compile_time_consta...
File frog/leg/compile_time_constants.dart (right):

http://codereview.chromium.org/9211005/diff/2001/frog/leg/compile_time_consta...
frog/leg/compile_time_constants.dart:57: new CompileTimeConstantEvaluator(this,
definitions, compiler);
On 2012/01/16 10:44:04, kasperl wrote:
> 4 space indent.

Done.

http://codereview.chromium.org/9211005/diff/2001/frog/leg/compile_time_consta...
frog/leg/compile_time_constants.dart:106: final CompileTimeConstantHandler
compileTimeConstantHandler;
On 2012/01/16 10:44:04, kasperl wrote:
> Maybe shorten the name to constantHandler? The context makes it obvious that
> we're talking about compile time.

Done.

http://codereview.chromium.org/9211005/diff/8001/frog/leg/compile_time_consta...
File frog/leg/compile_time_constants.dart (right):

http://codereview.chromium.org/9211005/diff/8001/frog/leg/compile_time_consta...
frog/leg/compile_time_constants.dart:131: // TODO(floitsch): make sure the field
is final.
On 2012/01/16 14:03:35, ngeoffray wrote:
> You have the modifiers, this should be easy to check.

done.

[ngeoffray, 2012-01-16 15:22:35]

LGTM with comments.

http://codereview.chromium.org/9211005/diff/13001/frog/leg/compile_time_const...
File frog/leg/compile_time_constants.dart (right):

http://codereview.chromium.org/9211005/diff/13001/frog/leg/compile_time_const...
frog/leg/compile_time_constants.dart:126: visitSend(Send send) {
Maybe add a helper in Elements:
Elements.isNonInstanceFinalField().

http://codereview.chromium.org/9211005/diff/13001/frog/leg/compile_time_const...
frog/leg/compile_time_constants.dart:133: error(element);
I believe this can be tested in the resolver.

http://codereview.chromium.org/9211005/diff/13001/frog/leg/warnings.dart
File frog/leg/warnings.dart (right):

http://codereview.chromium.org/9211005/diff/13001/frog/leg/warnings.dart#newc...
frog/leg/warnings.dart:122: class CompileTimeConstantError {
I would remove this. This looks like a ResolutionError.

[floitsch, 2012-01-18 12:56:19]

https://chromiumcodereview.appspot.com/9211005/diff/13001/frog/leg/compile_ti...
File frog/leg/compile_time_constants.dart (right):

https://chromiumcodereview.appspot.com/9211005/diff/13001/frog/leg/compile_ti...
frog/leg/compile_time_constants.dart:126: visitSend(Send send) {
On 2012/01/16 15:22:35, ngeoffray wrote:
> Maybe add a helper in Elements:
> Elements.isNonInstanceFinalField().

Since I want to check for non-final accesses keeping as is.

https://chromiumcodereview.appspot.com/9211005/diff/13001/frog/leg/compile_ti...
frog/leg/compile_time_constants.dart:133: error(element);
On 2012/01/16 15:22:35, ngeoffray wrote:
> I believe this can be tested in the resolver.

this one yes, but the resolver would need a state. Since the CTC needs to check
anyways for the cases the resolver cannot handle it seems easier to keep all
checking in the CTC.
leaving as is.

https://chromiumcodereview.appspot.com/9211005/diff/13001/frog/leg/warnings.dart
File frog/leg/warnings.dart (right):

https://chromiumcodereview.appspot.com/9211005/diff/13001/frog/leg/warnings.d...
frog/leg/warnings.dart:122: class CompileTimeConstantError {
On 2012/01/16 15:22:35, ngeoffray wrote:
> I would remove this. This looks like a ResolutionError.

as discussed: keeping it, since there are CTC errors the resolver cannot catch.