Recommit "Fix issues with Socket shutdown.""

Recommit "Fix issues with Socket shutdown.""

This recommits revision r44178, which was reverted in r44180, with changes to
fix the issues with Process.stdin.  It also adds a test for issue 22636, duplicate READ_CLOSED message.

Changes from r44187
Test tests/standalone/io/issue_22636_test.dart added.
Line "native.closedReadEventSent = true;" added to _RawSocket._writePipe() in runtime/bin/socket_patch.dart.

BUG=http://dartbug.com/22636
BUG=http://dartbug.com/22637
R=sgjesse@google.com

Committed: https://code.google.com/p/dart/source/detail?r=44236

[Bill Hesse, 2015-03-04 12:37:33]

whesse@google.com changed reviewers:
+ sgjesse@google.com

[Bill Hesse, 2015-03-04 12:37:34]

https://codereview.chromium.org/980593002/diff/1/runtime/bin/socket_patch.dart
File runtime/bin/socket_patch.dart (right):

https://codereview.chromium.org/980593002/diff/1/runtime/bin/socket_patch.dar...
runtime/bin/socket_patch.dart:1356: native.closedReadEventSent = true;
Added to CL, to fix broken Process.stdin tests.

[Søren Gjesse, 2015-03-04 13:28:16]

lgtm

https://codereview.chromium.org/980593002/diff/1/tests/standalone/io/issue_22...
File tests/standalone/io/issue_22636_test.dart (right):

https://codereview.chromium.org/980593002/diff/1/tests/standalone/io/issue_22...
tests/standalone/io/issue_22636_test.dart:13: final Duration delay = new
Duration(seconds: 1);
Do you need to wait that long?

https://codereview.chromium.org/980593002/diff/1/tests/standalone/io/issue_22...
tests/standalone/io/issue_22636_test.dart:22: print("server $event");
Remove print?

https://codereview.chromium.org/980593002/diff/1/tests/standalone/io/issue_22...
tests/standalone/io/issue_22636_test.dart:50: server = await
RawServerSocket.bind("localhost", 0);
We have been bitten by 'localhost' resolving to 127.0.0.1 on some platform and
::1 on other platforms.

Please use InternetAddress.IP_V4_LOOPBACK

https://codereview.chromium.org/980593002/diff/1/tests/standalone/io/issue_22...
File tests/standalone/io/issue_22637_test.dart (right):

https://codereview.chromium.org/980593002/diff/1/tests/standalone/io/issue_22...
tests/standalone/io/issue_22637_test.dart:59: server = await
RawServerSocket.bind("localhost", 0);
'localhost' -> InternetAddress.IP_V4_LOOPBACK

[Bill Hesse, 2015-03-04 14:36:42]

https://codereview.chromium.org/980593002/diff/1/tests/standalone/io/issue_22...
File tests/standalone/io/issue_22636_test.dart (right):

https://codereview.chromium.org/980593002/diff/1/tests/standalone/io/issue_22...
tests/standalone/io/issue_22636_test.dart:13: final Duration delay = new
Duration(seconds: 1);
On 2015/03/04 13:28:15, Søren Gjesse wrote:
> Do you need to wait that long?

No.  now 100 ms.

https://codereview.chromium.org/980593002/diff/1/tests/standalone/io/issue_22...
tests/standalone/io/issue_22636_test.dart:22: print("server $event");
On 2015/03/04 13:28:15, Søren Gjesse wrote:
> Remove print?

Done.

https://codereview.chromium.org/980593002/diff/1/tests/standalone/io/issue_22...
tests/standalone/io/issue_22636_test.dart:50: server = await
RawServerSocket.bind("localhost", 0);
On 2015/03/04 13:28:15, Søren Gjesse wrote:
> We have been bitten by 'localhost' resolving to 127.0.0.1 on some platform and
> ::1 on other platforms.
> 
> Please use InternetAddress.IP_V4_LOOPBACK

How well I remember. Changed to LOOPBACK_IP_V4. Done.

https://codereview.chromium.org/980593002/diff/1/tests/standalone/io/issue_22...
File tests/standalone/io/issue_22637_test.dart (right):

https://codereview.chromium.org/980593002/diff/1/tests/standalone/io/issue_22...
tests/standalone/io/issue_22637_test.dart:59: server = await
RawServerSocket.bind("localhost", 0);
On 2015/03/04 13:28:15, Søren Gjesse wrote:
> 'localhost' -> InternetAddress.IP_V4_LOOPBACK

Done.

[Bill Hesse, 2015-03-04 14:37:42]

Committed patchset #2 (id:20001) manually as 44236 (presubmit successful).