Add Isolate.kill().

Add Isolate.kill().

This sends a kill signal to the isolate, which is allowed to terminate
the isolate as soon as possible, and at the latest before the next event.

R=ajohnsen@google.com, floitsch@google.com, sgjesse@google.com

Committed: https://code.google.com/p/dart/source/detail?r=34234

Committed: https://code.google.com/p/dart/source/detail?r=34960

[Lasse Reichstein Nielsen, 2014-03-07 07:22:10]

Is my assumption, that control messages only happen at the event loop top,
correct?

[floitsch, 2014-03-07 19:38:53]

I think we will need different types like for ping.

webworkers can be killed in the middle of whatever they are doing.
The VM can kill something at any point.
Basically there are three ways to kill:
- immediate (in the middle of a loop).
- at next event (do microtasks count?)
- after all received events have been handled.

[Lasse Reichstein Nielsen, 2014-03-12 10:31:39]

This sounds like the same times as the ping. Suggests reusing the same
constants.

[floitsch, 2014-03-12 11:45:05]

On 2014/03/12 10:31:39, Lasse Reichstein Nielsen wrote:
> This sounds like the same times as the ping. Suggests reusing the same
> constants.

I think that makes sense.

[Lasse Reichstein Nielsen, 2014-03-13 15:01:50]

PTAL.
Names for priorities of ping/kill now generic, and bad.
Suggestions welcome.

[floitsch, 2014-03-13 15:52:11]

LGTM after we have found better names for the priorities.

[Lasse Reichstein Nielsen, 2014-03-14 12:32:20]

PTAL

[floitsch, 2014-03-14 13:01:44]

LGTM.
Ask others if they have better names, but I could live with these.

https://codereview.chromium.org/190013005/diff/40001/sdk/lib/_internal/lib/is...
File sdk/lib/_internal/lib/isolate_helper.dart (right):

https://codereview.chromium.org/190013005/diff/40001/sdk/lib/_internal/lib/is...
sdk/lib/_internal/lib/isolate_helper.dart:346:
!identical(_globalState.currentContext, this))) {
Create a helper function, so that it's more obvious what this is doing.

https://codereview.chromium.org/190013005/diff/40001/sdk/lib/_internal/lib/is...
sdk/lib/_internal/lib/isolate_helper.dart:374: assert(pingType ==
Isolate.BEFORE_NEXT_EVENT);
priority.

[Lasse Reichstein Nielsen, 2014-03-19 13:07:50]

+sgjesse, +ajohnsen: Any suggestions for better names for the contants?

https://codereview.chromium.org/190013005/diff/40001/sdk/lib/_internal/lib/is...
File sdk/lib/_internal/lib/isolate_helper.dart (right):

https://codereview.chromium.org/190013005/diff/40001/sdk/lib/_internal/lib/is...
sdk/lib/_internal/lib/isolate_helper.dart:374: assert(pingType ==
Isolate.BEFORE_NEXT_EVENT);
Ack.

[Anders Johnsen, 2014-03-19 13:15:34]

lgtm

https://codereview.chromium.org/190013005/diff/60001/sdk/lib/isolate/isolate....
File sdk/lib/isolate/isolate.dart (right):

https://codereview.chromium.org/190013005/diff/60001/sdk/lib/isolate/isolate....
sdk/lib/isolate/isolate.dart:33: static const int OUT_OF_BAND = 0;
IMMEDIATE?

People should not need to know about OOB.

https://codereview.chromium.org/190013005/diff/60001/sdk/lib/isolate/isolate....
sdk/lib/isolate/isolate.dart:35: static const int BEFORE_NEXT_EVENT = 1;
Not sure if I have another name for this one.

https://codereview.chromium.org/190013005/diff/60001/sdk/lib/isolate/isolate....
sdk/lib/isolate/isolate.dart:37: static const int AS_EVENT = 2;
END_OF_QUEUE?

[Søren Gjesse, 2014-03-20 08:08:32]

lgtm

https://codereview.chromium.org/190013005/diff/60001/sdk/lib/isolate/isolate....
File sdk/lib/isolate/isolate.dart (right):

https://codereview.chromium.org/190013005/diff/60001/sdk/lib/isolate/isolate....
sdk/lib/isolate/isolate.dart:33: static const int OUT_OF_BAND = 0;
On 2014/03/19 13:15:34, Anders Johnsen wrote:
> IMMEDIATE?
> 
> People should not need to know about OOB.

Agree with Anders. ASAP is also an option, but that's probably to strange a
name.

https://codereview.chromium.org/190013005/diff/60001/sdk/lib/isolate/isolate....
sdk/lib/isolate/isolate.dart:35: static const int BEFORE_NEXT_EVENT = 1;
HEAD_OF_QUEUE?

https://codereview.chromium.org/190013005/diff/60001/sdk/lib/isolate/isolate....
sdk/lib/isolate/isolate.dart:37: static const int AS_EVENT = 2;
On 2014/03/19 13:15:34, Anders Johnsen wrote:
> END_OF_QUEUE?

Using HEAD_OF_QUEUE / END_OF_QUEUE has some symmetry to it.

[Anders Johnsen, 2014-03-20 08:48:00]

https://codereview.chromium.org/190013005/diff/60001/sdk/lib/isolate/isolate....
File sdk/lib/isolate/isolate.dart (right):

https://codereview.chromium.org/190013005/diff/60001/sdk/lib/isolate/isolate....
sdk/lib/isolate/isolate.dart:37: static const int AS_EVENT = 2;
On 2014/03/20 08:08:33, Søren Gjesse wrote:
> On 2014/03/19 13:15:34, Anders Johnsen wrote:
> > END_OF_QUEUE?
> 
> Using HEAD_OF_QUEUE / END_OF_QUEUE has some symmetry to it.

+1

[Lasse Reichstein Nielsen, 2014-03-20 13:12:33]

https://codereview.chromium.org/190013005/diff/60001/sdk/lib/isolate/isolate....
File sdk/lib/isolate/isolate.dart (right):

https://codereview.chromium.org/190013005/diff/60001/sdk/lib/isolate/isolate....
sdk/lib/isolate/isolate.dart:33: static const int OUT_OF_BAND = 0;
I tried ASAP, and it didn't really work. 
IMMEDIATE is fine.

https://codereview.chromium.org/190013005/diff/60001/sdk/lib/isolate/isolate....
sdk/lib/isolate/isolate.dart:35: static const int BEFORE_NEXT_EVENT = 1;
I prefer to talk about "events" rather than "queue". 
We talk about "the event loop" in several places, and the "event queue" only a
few times.

https://codereview.chromium.org/190013005/diff/60001/sdk/lib/isolate/isolate....
sdk/lib/isolate/isolate.dart:37: static const int AS_EVENT = 2;
HEAD_OF_QUEUE sounds like it is in the queue, which it isn't. The control
message will be executed, even if the event loop is paused.

I think I prefer the current names to X_OF_QUEUE.

[Lasse Reichstein Nielsen, 2014-03-21 08:56:19]

Committed patchset #6 manually as r34234 (presubmit successful).

[Lasse Reichstein Nielsen, 2014-04-10 12:21:14]

Commit was reverted. Trying again with fixes.

[Lasse Reichstein Nielsen, 2014-04-10 12:32:43]

Very simple change to test. PTAL

[Anders Johnsen, 2014-04-10 12:40:13]

diff to previous commit, LGTM

[Lasse Reichstein Nielsen, 2014-04-11 07:23:01]

Committed patchset #8 manually as r34960 (presubmit successful).