Remove streamSpawnUri and mangler.

Remove streamSpawnUri and mangler.

Add more tests.
Make dart2js send sinks.

Committed: https://code.google.com/p/dart/source/detail?r=20213

[floitsch, 2013-03-18 18:47:44]

All tests are new. The code-review tool will show them as modified, but no
existing test has been modified.

https://codereview.chromium.org/12919011/diff/3001/sdk/lib/isolate/isolate_st...
File sdk/lib/isolate/isolate_stream.dart (right):

https://codereview.chromium.org/12919011/diff/3001/sdk/lib/isolate/isolate_st...
sdk/lib/isolate/isolate_stream.dart:120: /** Closing multiple times is allowed.
*/
Changed to allow multiple closings.

[Lasse Reichstein Nielsen, 2013-03-19 12:01:55]

lgtm

https://codereview.chromium.org/12919011/diff/3001/sdk/lib/_internal/compiler...
File sdk/lib/_internal/compiler/implementation/lib/isolate_helper.dart (right):

https://codereview.chromium.org/12919011/diff/3001/sdk/lib/_internal/compiler...
sdk/lib/_internal/compiler/implementation/lib/isolate_helper.dart:1082:
IsolateSink deserializeIsolateSink(List x) {
Better name than 'x', please.

https://codereview.chromium.org/12919011/diff/3001/sdk/lib/_internal/compiler...
File sdk/lib/_internal/compiler/implementation/lib/isolate_patch.dart (right):

https://codereview.chromium.org/12919011/diff/3001/sdk/lib/_internal/compiler...
sdk/lib/_internal/compiler/implementation/lib/isolate_patch.dart:37: //
TODO(floitsch): can we compare against const CloseToken()?
If you can, at it seems you can, please do.

https://codereview.chromium.org/12919011/diff/3001/tests/isolate/count_stream...
File tests/isolate/count_stream_test.dart (right):

https://codereview.chromium.org/12919011/diff/3001/tests/isolate/count_stream...
tests/isolate/count_stream_test.dart:13: stream.listen((msg) {
Where is "stream" declared?

https://codereview.chromium.org/12919011/diff/3001/tests/isolate/count_stream...
tests/isolate/count_stream_test.dart:37: remote.add(count);
Don't send "count" here. Just send null. It looks confusing, and it could hide
an error if the value is just looped back instead of being handled.

https://codereview.chromium.org/12919011/diff/3001/tests/isolate/count_stream...
tests/isolate/count_stream_test.dart:44: remote.add(0);
Send null.

https://codereview.chromium.org/12919011/diff/3001/tests/isolate/cross_isolat...
File tests/isolate/cross_isolate_message_stream_test.dart (right):

https://codereview.chromium.org/12919011/diff/3001/tests/isolate/cross_isolat...
tests/isolate/cross_isolate_message_stream_test.dart:11: 
Add documentation for what is expected to happen in which order.

https://codereview.chromium.org/12919011/diff/3001/tests/isolate/illegal_msg_...
File tests/isolate/illegal_msg_stream_test.dart (right):

https://codereview.chromium.org/12919011/diff/3001/tests/isolate/illegal_msg_...
tests/isolate/illegal_msg_stream_test.dart:29: box.stream.single.then((msg) {
expectAsync?

[floitsch, 2013-03-19 13:18:59]

https://codereview.chromium.org/12919011/diff/3001/sdk/lib/_internal/compiler...
File sdk/lib/_internal/compiler/implementation/lib/isolate_helper.dart (right):

https://codereview.chromium.org/12919011/diff/3001/sdk/lib/_internal/compiler...
sdk/lib/_internal/compiler/implementation/lib/isolate_helper.dart:1082:
IsolateSink deserializeIsolateSink(List x) {
On 2013/03/19 12:01:55, Lasse Reichstein Nielsen wrote:
> Better name than 'x', please.

Done.

https://codereview.chromium.org/12919011/diff/3001/sdk/lib/_internal/compiler...
File sdk/lib/_internal/compiler/implementation/lib/isolate_patch.dart (right):

https://codereview.chromium.org/12919011/diff/3001/sdk/lib/_internal/compiler...
sdk/lib/_internal/compiler/implementation/lib/isolate_patch.dart:37: //
TODO(floitsch): can we compare against const CloseToken()?
On 2013/03/19 12:01:55, Lasse Reichstein Nielsen wrote:
> If you can, at it seems you can, please do.

Done.

https://codereview.chromium.org/12919011/diff/3001/tests/isolate/count_stream...
File tests/isolate/count_stream_test.dart (right):

https://codereview.chromium.org/12919011/diff/3001/tests/isolate/count_stream...
tests/isolate/count_stream_test.dart:13: stream.listen((msg) {
On 2013/03/19 12:01:55, Lasse Reichstein Nielsen wrote:
> Where is "stream" declared?
Top level from async.

https://codereview.chromium.org/12919011/diff/3001/tests/isolate/count_stream...
tests/isolate/count_stream_test.dart:37: remote.add(count);
On 2013/03/19 12:01:55, Lasse Reichstein Nielsen wrote:
> Don't send "count" here. Just send null. It looks confusing, and it could hide
> an error if the value is just looped back instead of being handled.

Done.

https://codereview.chromium.org/12919011/diff/3001/tests/isolate/count_stream...
tests/isolate/count_stream_test.dart:44: remote.add(0);
On 2013/03/19 12:01:55, Lasse Reichstein Nielsen wrote:
> Send null.

Done.

https://codereview.chromium.org/12919011/diff/3001/tests/isolate/cross_isolat...
File tests/isolate/cross_isolate_message_stream_test.dart (right):

https://codereview.chromium.org/12919011/diff/3001/tests/isolate/cross_isolat...
tests/isolate/cross_isolate_message_stream_test.dart:11: 
On 2013/03/19 12:01:55, Lasse Reichstein Nielsen wrote:
> Add documentation for what is expected to happen in which order.

Done.

https://codereview.chromium.org/12919011/diff/3001/tests/isolate/illegal_msg_...
File tests/isolate/illegal_msg_stream_test.dart (right):

https://codereview.chromium.org/12919011/diff/3001/tests/isolate/illegal_msg_...
tests/isolate/illegal_msg_stream_test.dart:29: box.stream.single.then((msg) {
On 2013/03/19 12:01:55, Lasse Reichstein Nielsen wrote:
> expectAsync?

This test doesn't import the unit-test library.
In theory expectAsync1 is not needed, since box.stream.single will either throw
an exception (no single message), or keep the stream open (thus keeping the main
running).

[floitsch, 2013-03-19 16:12:39]

Committed patchset #3 manually as r20213 (presubmit successful).

[Ivan Posva, 2013-03-19 20:50:48]

> Make dart2js send sinks.

I would still like to understand how you would like to solve the issue of having
to crack open the whole message to hunt down the embedded sinks when crossing
process boundaries. It used to be that SendPorts were being send on a different
"channel" making it easy to handle this without having to store the entire
message, decode it and reencode it before sending it on to the eventual
receiver.

Thanks,
-Ivan