Improve web-socket implementation by using making _WebSocketProtocolProcessor a StreamTransformer.

Improve web-socket implementation by using making _WebSocketProtocolProcessor a StreamTransformer.


BUG=

Committed: https://code.google.com/p/dart/source/detail?r=19454

[Anders Johnsen, 2013-03-04 16:38:09]

This is now ready for review (fixed some more corner-cases found in production).

[Søren Gjesse, 2013-03-05 09:32:29]

lgtm

https://codereview.chromium.org/12387085/diff/8001/sdk/lib/io/websocket_impl....
File sdk/lib/io/websocket_impl.dart (right):

https://codereview.chromium.org/12387085/diff/8001/sdk/lib/io/websocket_impl....
sdk/lib/io/websocket_impl.dart:221: } catch (e, s) {
s not used.

https://codereview.chromium.org/12387085/diff/8001/sdk/lib/io/websocket_impl....
sdk/lib/io/websocket_impl.dart:351: int closeCode = 1005;
Use WebSocketStatus.NO_STATUS_RECEIVED instead of 1005.

https://codereview.chromium.org/12387085/diff/8001/sdk/lib/io/websocket_impl....
sdk/lib/io/websocket_impl.dart:352: String closeReason = "";
Shouldn't we use null instead of ""?

https://codereview.chromium.org/12387085/diff/8001/sdk/lib/io/websocket_impl....
sdk/lib/io/websocket_impl.dart:525: bool clean = true;
Looks as if clean is not used.

https://codereview.chromium.org/12387085/diff/8001/tests/standalone/io/web_so...
File tests/standalone/io/web_socket_protocol_processor_test.dart (right):

https://codereview.chromium.org/12387085/diff/8001/tests/standalone/io/web_so...
tests/standalone/io/web_socket_protocol_processor_test.dart:207: for (int i = 0;
i < messageLength; i++) message[i] = i & 0x7F;
Maybe keep the FF for binary messages.

https://codereview.chromium.org/12387085/diff/8001/tests/standalone/io/web_so...
File tests/standalone/io/web_socket_test.dart (right):

https://codereview.chromium.org/12387085/diff/8001/tests/standalone/io/web_so...
tests/standalone/io/web_socket_test.dart:126: closeReason == null ? "" :
closeReason, webSocket.closeReason);
Long line

[Anders Johnsen, 2013-03-05 09:38:34]

Thanks, landing.

https://codereview.chromium.org/12387085/diff/8001/sdk/lib/io/websocket_impl....
File sdk/lib/io/websocket_impl.dart (right):

https://codereview.chromium.org/12387085/diff/8001/sdk/lib/io/websocket_impl....
sdk/lib/io/websocket_impl.dart:221: } catch (e, s) {
On 2013/03/05 09:32:29, Søren Gjesse wrote:
> s not used.

Done.

https://codereview.chromium.org/12387085/diff/8001/sdk/lib/io/websocket_impl....
sdk/lib/io/websocket_impl.dart:351: int closeCode = 1005;
On 2013/03/05 09:32:29, Søren Gjesse wrote:
> Use WebSocketStatus.NO_STATUS_RECEIVED instead of 1005.

Done.

https://codereview.chromium.org/12387085/diff/8001/sdk/lib/io/websocket_impl....
sdk/lib/io/websocket_impl.dart:352: String closeReason = "";
On 2013/03/05 09:32:29, Søren Gjesse wrote:
> Shouldn't we use null instead of ""?

By using "" we are always sure that we never send a 'null' to the user. Note
that this is the internal state, not exposed to the user until close.

https://codereview.chromium.org/12387085/diff/8001/sdk/lib/io/websocket_impl....
sdk/lib/io/websocket_impl.dart:525: bool clean = true;
On 2013/03/05 09:32:29, Søren Gjesse wrote:
> Looks as if clean is not used.

Done.

https://codereview.chromium.org/12387085/diff/8001/tests/standalone/io/web_so...
File tests/standalone/io/web_socket_protocol_processor_test.dart (right):

https://codereview.chromium.org/12387085/diff/8001/tests/standalone/io/web_so...
tests/standalone/io/web_socket_protocol_processor_test.dart:207: for (int i = 0;
i < messageLength; i++) message[i] = i & 0x7F;
On 2013/03/05 09:32:29, Søren Gjesse wrote:
> Maybe keep the FF for binary messages.

Done.

https://codereview.chromium.org/12387085/diff/8001/tests/standalone/io/web_so...
File tests/standalone/io/web_socket_test.dart (right):

https://codereview.chromium.org/12387085/diff/8001/tests/standalone/io/web_so...
tests/standalone/io/web_socket_test.dart:126: closeReason == null ? "" :
closeReason, webSocket.closeReason);
On 2013/03/05 09:32:29, Søren Gjesse wrote:
> Long line

Done.

[Anders Johnsen, 2013-03-05 09:41:01]

Committed patchset #4 manually as r19454 (presubmit successful).