[strong] More scoping related errors: object literal methods.

[strong] More scoping related errors: object literal methods.

This CL adds errors for illegal references which occur inside object literal
methods inside computed properrty names.

BUG=v8:3948, v8:3956
LOG=N

Committed: https://crrev.com/6ad87e4f869523f2cfe048f4538dfe568ad04b5d
Cr-Commit-Position: refs/heads/master@{#27118}

[marja, 2015-03-10 15:38:00]

marja@chromium.org changed reviewers:
+ dslomov@gmail.com

[marja, 2015-03-10 15:38:01]

dslomov, ptal

[marja, 2015-03-10 15:38:58]

marja@chromium.org changed reviewers:
+ dslomov@chromium.org
- dslomov@gmail.com

[Dmitry Lomov (no reviews), 2015-03-10 16:06:01]

lgtm % comment

https://codereview.chromium.org/994043003/diff/40001/src/preparser.h
File src/preparser.h (right):

https://codereview.chromium.org/994043003/diff/40001/src/preparser.h#newcode2152
src/preparser.h:2152: kind = static_cast<FunctionKind>(kind |
FunctionKind::kInObjectLiteral);
Add function:
  FunctionKind WithObjectLiteralBit(FunctionKind kind)
that does this and asserts validity.

https://codereview.chromium.org/994043003/diff/40001/src/preparser.h#newcode2188
src/preparser.h:2188: kind = static_cast<FunctionKind>(kind |
FunctionKind::kInObjectLiteral);
Ditto.

[marja, 2015-03-10 16:17:38]

t

https://codereview.chromium.org/994043003/diff/40001/src/preparser.h
File src/preparser.h (right):

https://codereview.chromium.org/994043003/diff/40001/src/preparser.h#newcode2152
src/preparser.h:2152: kind = static_cast<FunctionKind>(kind |
FunctionKind::kInObjectLiteral);
On 2015/03/10 16:06:01, Dmitry Lomov (chromium) wrote:
> Add function:
>   FunctionKind WithObjectLiteralBit(FunctionKind kind)
> that does this and asserts validity.

Done.

https://codereview.chromium.org/994043003/diff/40001/src/preparser.h#newcode2188
src/preparser.h:2188: kind = static_cast<FunctionKind>(kind |
FunctionKind::kInObjectLiteral);
On 2015/03/10 16:06:01, Dmitry Lomov (chromium) wrote:
> Ditto.

Done.

[marja, 2015-03-10 16:17:41]

The CQ bit was checked by marja@chromium.org

[marja, 2015-03-10 16:17:42]

The patchset sent to the CQ was uploaded after l-g-t-m from dslomov@chromium.org
Link to the patchset: https://codereview.chromium.org/994043003/#ps60001 (title:
"moar code reviews")

[commit-bot: I haz the power, 2015-03-10 16:17:55]

CQ is trying da patch. Follow status at
 https://chromium-cq-status.appspot.com/patch-status/994043003/60001

[commit-bot: I haz the power, 2015-03-10 16:47:56]

Committed patchset #4 (id:60001)

[commit-bot: I haz the power, 2015-03-10 16:48:14]

Patchset 4 (id:??) landed as
https://crrev.com/6ad87e4f869523f2cfe048f4538dfe568ad04b5d
Cr-Commit-Position: refs/heads/master@{#27118}

[arv (Not doing code reviews), 2015-03-10 16:49:05]

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

[arv (Not doing code reviews), 2015-03-10 16:49:06]

Would it not make more sense to track that we are inside the method/accessor of
a class instead of tracking when we are in an object literal outside a class?

https://codereview.chromium.org/994043003/diff/60001/src/hydrogen-instructions.h
File src/hydrogen-instructions.h (right):

https://codereview.chromium.org/994043003/diff/60001/src/hydrogen-instruction...
src/hydrogen-instructions.h:7584: class PretenureField : public BitField<bool,
8, 1> {};
Maybe use kNext here?

https://codereview.chromium.org/994043003/diff/60001/src/preparser.h
File src/preparser.h (right):

https://codereview.chromium.org/994043003/diff/60001/src/preparser.h#newcode2185
src/preparser.h:2185: if (!in_class) kind = WithObjectLiteralBit(kind);
It is not clear to me that we need this extra bit. Why do we need to distinguish
if the accessor is part of a ClassLiteral vs ObjectLiteral?

https://codereview.chromium.org/994043003/diff/60001/test/mjsunit/strong/decl...
File test/mjsunit/strong/declaration-after-use.js (right):

https://codereview.chromium.org/994043003/diff/60001/test/mjsunit/strong/decl...
test/mjsunit/strong/declaration-after-use.js:167: "if (false) {\n" +
unindent

https://codereview.chromium.org/994043003/diff/60001/test/mjsunit/strong/decl...
test/mjsunit/strong/declaration-after-use.js:167: "if (false) {\n" +
unindent

[marja, 2015-03-10 17:08:13]

My first approach (patch set 2) was to add the reverse "is in class literal"
bit, but we concluded w/ dslomov@ that this makes more sense, since there are
less function types allowed in an object literal (those are a subset of the
function types allowed in a class literal).

https://codereview.chromium.org/994043003/diff/60001/src/preparser.h
File src/preparser.h (right):

https://codereview.chromium.org/994043003/diff/60001/src/preparser.h#newcode2185
src/preparser.h:2185: if (!in_class) kind = WithObjectLiteralBit(kind);
On 2015/03/10 16:49:06, arv wrote:
> It is not clear to me that we need this extra bit. Why do we need to
distinguish
> if the accessor is part of a ClassLiteral vs ObjectLiteral?

Scope-wise, a method (or an accessor) of an object literal inside a computed
property name will look the same as a method (or an accessor) of the class - but
only the latter is allowed to refer to the class name.

[marja, 2015-03-10 17:09:30]

(I'm planning to do the changes you suggested as separate follow-ups; I'm
planning to do a round of test cleanup anyhow, and the kNext thing can be a
separate CL.)