Implement the function.sent proposal.

Implement the function.sent proposal.

The body of a generator function can now refer to the generator's input value via a new
"function.sent" expression.  We extend the proposal at
https://github.com/allenwb/ESideas/blob/master/Generator%20metaproperty.md
in the obvious way to also apply to GeneratorResumeAbrupt.
This will enable us to desugar yield*.

The new syntax is behind a new --harmony-function-sent flag.

BUG=v8:4700
LOG=n

Committed: https://crrev.com/e2466bb5ff5288610383cffb5ab698bed1da75c8
Cr-Commit-Position: refs/heads/master@{#33574}

[neis, 2016-01-22 13:59:21]

Description was changed from

==========
Implement the function.sent proposal.

The body of a generator function can now refer to its input value via a new
"function.sent" expression.  We extend the proposal at
https://github.com/allenwb/ESideas/blob/master/Generator%20metaproperty.md
in the obvious way to also apply to GeneratorResumeAbrupt.
This will enable us to desugar yield*.

The new syntax is behind a new --harmony-function-sent flag.

BUG=
==========

to

==========
Implement the function.sent proposal.

The body of a generator function can now refer to its input value via a new
"function.sent" expression.  We extend the proposal at
https://github.com/allenwb/ESideas/blob/master/Generator%20metaproperty.md
in the obvious way to also apply to GeneratorResumeAbrupt.
This will enable us to desugar yield*.

The new syntax is behind a new --harmony-function-sent flag.

BUG=
==========

[neis, 2016-01-22 13:59:21]

neis@chromium.org changed reviewers:
+ rossberg@chromium.org

[neis, 2016-01-22 14:00:10]

PTAL.  If this looks good, I will modify the other parts of full-codegen.

[neis, 2016-01-22 14:01:12]

Description was changed from

==========
Implement the function.sent proposal.

The body of a generator function can now refer to its input value via a new
"function.sent" expression.  We extend the proposal at
https://github.com/allenwb/ESideas/blob/master/Generator%20metaproperty.md
in the obvious way to also apply to GeneratorResumeAbrupt.
This will enable us to desugar yield*.

The new syntax is behind a new --harmony-function-sent flag.

BUG=
==========

to

==========
Implement the function.sent proposal.

The body of a generator function can now refer to the generator's input value
via a new
"function.sent" expression.  We extend the proposal at
https://github.com/allenwb/ESideas/blob/master/Generator%20metaproperty.md
in the obvious way to also apply to GeneratorResumeAbrupt.
This will enable us to desugar yield*.

The new syntax is behind a new --harmony-function-sent flag.

BUG=
==========

[rossberg, 2016-01-22 14:13:07]

https://codereview.chromium.org/1620253003/diff/1/src/messages.h
File src/messages.h (right):

https://codereview.chromium.org/1620253003/diff/1/src/messages.h#newcode498
src/messages.h:498: T(UnexpectedFunctionSent, "function.sent expression is not
allowed here")    \
"...not allowed outside a generator"

https://codereview.chromium.org/1620253003/diff/1/src/parsing/parser-base.h
File src/parsing/parser-base.h (right):

https://codereview.chromium.org/1620253003/diff/1/src/parsing/parser-base.h#n...
src/parsing/parser-base.h:2596: if (!is_generator()) {
This may not be good enough. Like other magic constants, function.sent is
probably intended to be accessible in nested arrow functions, e.g.

  function *g() { yield; return (() => function.sent))() }

I think it's fine to leave it like that for now, until there is a more detailed
spec, but at least add a TODO.

https://codereview.chromium.org/1620253003/diff/1/test/mjsunit/harmony/functi...
File test/mjsunit/harmony/function-sent.js (right):

https://codereview.chromium.org/1620253003/diff/1/test/mjsunit/harmony/functi...
test/mjsunit/harmony/function-sent.js:6: 
Add tests using yield*.

https://codereview.chromium.org/1620253003/diff/1/test/mjsunit/harmony/functi...
test/mjsunit/harmony/function-sent.js:20: return function.sent;
Maybe also have a test that yields in the finally clause?

https://codereview.chromium.org/1620253003/diff/1/test/mjsunit/harmony/functi...
test/mjsunit/harmony/function-sent.js:39: 
Once there is 'return', we should probably have more tests here, especially some
that yield after return. Perhaps add a TODO.

[neis, 2016-01-22 14:51:58]

Description was changed from

==========
Implement the function.sent proposal.

The body of a generator function can now refer to the generator's input value
via a new
"function.sent" expression.  We extend the proposal at
https://github.com/allenwb/ESideas/blob/master/Generator%20metaproperty.md
in the obvious way to also apply to GeneratorResumeAbrupt.
This will enable us to desugar yield*.

The new syntax is behind a new --harmony-function-sent flag.

BUG=
==========

to

==========
Implement the function.sent proposal.

The body of a generator function can now refer to the generator's input value
via a new
"function.sent" expression.  We extend the proposal at
https://github.com/allenwb/ESideas/blob/master/Generator%20metaproperty.md
in the obvious way to also apply to GeneratorResumeAbrupt.
This will enable us to desugar yield*.

The new syntax is behind a new --harmony-function-sent flag.

BUG=v8:4700
LOG=n
==========

[adamk, 2016-01-23 00:55:15]

adamk@chromium.org changed reviewers:
+ adamk@chromium.org

[adamk, 2016-01-23 00:55:15]

According to https://github.com/tc39/ecma262, function.sent is only at stage 2.
I'd prefer to avoid implementing things before stage 3 unless there's a
particular impetus, especially since the spec for such features isn't complete
yet (as noted by rossberg in his TODO suggestion).

Is there a reason we want to do this before stage 3?

[rossberg, 2016-01-25 11:16:11]

On 2016/01/23 00:55:15, adamk wrote:
> According to https://github.com/tc39/ecma262, function.sent is only at stage
2.
> I'd prefer to avoid implementing things before stage 3 unless there's a
> particular impetus, especially since the spec for such features isn't complete
> yet (as noted by rossberg in his TODO suggestion).
> 
> Is there a reason we want to do this before stage 3?

Yes: we discovered that this was exactly the missing bit for desugaring yield*
(which in turn is needed to make `return` fully working). Since it is a fairly
trivial addition, we decided to just go for it to unblock the other generator
stuff.

[adamk, 2016-01-25 19:16:07]

Thanks for the background/ Not sure how I missed this in the CL description.
Carry on...

[neis, 2016-01-26 10:13:01]

The CQ bit was checked by neis@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-01-26 10:13:21]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1620253003/40001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1620253003/40001

[neis, 2016-01-26 10:27:18]

On 2016/01/22 14:13:07, rossberg wrote:
>
https://codereview.chromium.org/1620253003/diff/1/src/parsing/parser-base.h#n...
> src/parsing/parser-base.h:2596: if (!is_generator()) {
> This may not be good enough. Like other magic constants, function.sent is
> probably intended to be accessible in nested arrow functions, e.g.
> 
>   function *g() { yield; return (() => function.sent))() }
> 
> I think it's fine to leave it like that for now, until there is a more
detailed
> spec, but at least add a TODO.

I think now that it's not intended for such use, just like yield itself.  The
extension to the parser that we discussed (walking up the scope to find the
magic generator variable) does not make much sense because if function.sent is
captured in a closure then what value should it return when the closure gets
called?  The generator may not be running anymore at that point.

[rossberg, 2016-01-26 19:13:39]

On 2016/01/26 10:27:18, neis wrote:
> On 2016/01/22 14:13:07, rossberg wrote:
> >
>
https://codereview.chromium.org/1620253003/diff/1/src/parsing/parser-base.h#n...
> > src/parsing/parser-base.h:2596: if (!is_generator()) {
> > This may not be good enough. Like other magic constants, function.sent is
> > probably intended to be accessible in nested arrow functions, e.g.
> > 
> >   function *g() { yield; return (() => function.sent))() }
> > 
> > I think it's fine to leave it like that for now, until there is a more
> detailed
> > spec, but at least add a TODO.
> 
> I think now that it's not intended for such use, just like yield itself.  The
> extension to the parser that we discussed (walking up the scope to find the
> magic generator variable) does not make much sense because if function.sent is
> captured in a closure then what value should it return when the closure gets
> called?  The generator may not be running anymore at that point.

The last value. That would be consistent with other mutable state you close
over.

I agree this is a dubious feature. But never forget this is JavaScript, where
you can even do

let f;
class C extends Object {
  constructor() { f = () => super(); f() }
}
let o = new C
f()  // Say what?

Wanna be consistently insane.

[neis, 2016-01-27 16:07:53]

The CQ bit was checked by neis@chromium.org to run a CQ dry run

[commit-bot: I haz the power, 2016-01-27 16:08:00]

Dry run: CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1620253003/140001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1620253003/140001

[commit-bot: I haz the power, 2016-01-27 16:34:40]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-01-27 16:34:41]

Dry run: This issue passed the CQ dry run.

[neis, 2016-01-27 16:49:40]

Please have another look.

When this lands, I'll immediately upload the yield* desugaring.

https://codereview.chromium.org/1620253003/diff/1/src/messages.h
File src/messages.h (right):

https://codereview.chromium.org/1620253003/diff/1/src/messages.h#newcode498
src/messages.h:498: T(UnexpectedFunctionSent, "function.sent expression is not
allowed here")    \
On 2016/01/22 14:13:06, rossberg wrote:
> "...not allowed outside a generator"

Done.

https://codereview.chromium.org/1620253003/diff/1/src/parsing/parser-base.h
File src/parsing/parser-base.h (right):

https://codereview.chromium.org/1620253003/diff/1/src/parsing/parser-base.h#n...
src/parsing/parser-base.h:2596: if (!is_generator()) {
On 2016/01/22 14:13:06, rossberg wrote:
> This may not be good enough. Like other magic constants, function.sent is
> probably intended to be accessible in nested arrow functions, e.g.
> 
>   function *g() { yield; return (() => function.sent))() }
> 
> I think it's fine to leave it like that for now, until there is a more
detailed
> spec, but at least add a TODO.

Done.

https://codereview.chromium.org/1620253003/diff/1/test/mjsunit/harmony/functi...
File test/mjsunit/harmony/function-sent.js (right):

https://codereview.chromium.org/1620253003/diff/1/test/mjsunit/harmony/functi...
test/mjsunit/harmony/function-sent.js:6: 
On 2016/01/22 14:13:06, rossberg wrote:
> Add tests using yield*.

Done.

https://codereview.chromium.org/1620253003/diff/1/test/mjsunit/harmony/functi...
test/mjsunit/harmony/function-sent.js:20: return function.sent;
On 2016/01/22 14:13:06, rossberg wrote:
> Maybe also have a test that yields in the finally clause?

Done.

[rossberg, 2016-01-27 17:13:10]

lgtm

[neis, 2016-01-28 07:53:43]

The CQ bit was checked by neis@chromium.org

[commit-bot: I haz the power, 2016-01-28 07:53:53]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1620253003/140001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1620253003/140001

[commit-bot: I haz the power, 2016-01-28 07:56:25]

The CQ bit was unchecked by commit-bot@chromium.org

[commit-bot: I haz the power, 2016-01-28 07:56:26]

Try jobs failed on following builders:
  v8_presubmit on tryserver.v8 (JOB_FAILED,
http://build.chromium.org/p/tryserver.v8/builders/v8_presubmit/builds/10250)

[neis, 2016-01-28 07:58:33]

neis@chromium.org changed reviewers:
+ jarin@chromium.org

[neis, 2016-01-28 07:59:27]

On 2016/01/28 07:56:26, commit-bot: I haz the power wrote:
> Try jobs failed on following builders:
>   v8_presubmit on tryserver.v8 (JOB_FAILED,
> http://build.chromium.org/p/tryserver.v8/builders/v8_presubmit/builds/10250)

Hi Jaro, missing a full-codegen owner.  Can you have a look at the changes? 
They are minimal.

[Jarin, 2016-01-28 08:52:19]

full-code stuff lgtm.

[neis, 2016-01-28 08:52:57]

The CQ bit was checked by neis@chromium.org

[commit-bot: I haz the power, 2016-01-28 08:53:05]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/1620253003/140001
View timeline at
 https://chromium-cq-status.appspot.com/patch-timeline/1620253003/140001

[commit-bot: I haz the power, 2016-01-28 08:54:32]

Description was changed from

==========
Implement the function.sent proposal.

The body of a generator function can now refer to the generator's input value
via a new
"function.sent" expression.  We extend the proposal at
https://github.com/allenwb/ESideas/blob/master/Generator%20metaproperty.md
in the obvious way to also apply to GeneratorResumeAbrupt.
This will enable us to desugar yield*.

The new syntax is behind a new --harmony-function-sent flag.

BUG=v8:4700
LOG=n
==========

to

==========
Implement the function.sent proposal.

The body of a generator function can now refer to the generator's input value
via a new
"function.sent" expression.  We extend the proposal at
https://github.com/allenwb/ESideas/blob/master/Generator%20metaproperty.md
in the obvious way to also apply to GeneratorResumeAbrupt.
This will enable us to desugar yield*.

The new syntax is behind a new --harmony-function-sent flag.

BUG=v8:4700
LOG=n
==========

[commit-bot: I haz the power, 2016-01-28 08:54:33]

Committed patchset #8 (id:140001)

[commit-bot: I haz the power, 2016-01-28 08:55:03]

Description was changed from

==========
Implement the function.sent proposal.

The body of a generator function can now refer to the generator's input value
via a new
"function.sent" expression.  We extend the proposal at
https://github.com/allenwb/ESideas/blob/master/Generator%20metaproperty.md
in the obvious way to also apply to GeneratorResumeAbrupt.
This will enable us to desugar yield*.

The new syntax is behind a new --harmony-function-sent flag.

BUG=v8:4700
LOG=n
==========

to

==========
Implement the function.sent proposal.

The body of a generator function can now refer to the generator's input value
via a new
"function.sent" expression.  We extend the proposal at
https://github.com/allenwb/ESideas/blob/master/Generator%20metaproperty.md
in the obvious way to also apply to GeneratorResumeAbrupt.
This will enable us to desugar yield*.

The new syntax is behind a new --harmony-function-sent flag.

BUG=v8:4700
LOG=n

Committed: https://crrev.com/e2466bb5ff5288610383cffb5ab698bed1da75c8
Cr-Commit-Position: refs/heads/master@{#33574}
==========

[commit-bot: I haz the power, 2016-01-28 08:55:04]

Patchset 8 (id:??) landed as
https://crrev.com/e2466bb5ff5288610383cffb5ab698bed1da75c8
Cr-Commit-Position: refs/heads/master@{#33574}