Normative: Add Logical Assignment Operators

[jridgewell]

https://github.com/tc39/proposal-logical-assignment

This includes a few new changes:

• NamedEvaluation is used, given today's tentative consensus
• LogicalAssignmentOperator has been removed, and instead each @@= operator is defined individually

---

Outdated discussion

I need help getting it to compile correctly, though. When I run build, I get:

Could not find a production matching RHS in syntax-directed operation (undefined-nonterminal)

This is likely from defining the grammar as LeftHandSideExpression ShortCircuitAssignmentOperator AssignmentExpression, but defining individual algorithms as AssignmentExpression : LeftHandSideExpression `||=` AssignmentExpression. We could define each grammar individually, or do a bit of metaprogramming for the algorithms, like AssignmentOperator did. I just don't know how to metaprogram in spec-ese.

[bakkot]

This is likely from defining the grammar as LeftHandSideExpression ShortCircuitAssignmentOperator AssignmentExpression, but defining individual algorithms as AssignmentExpression : LeftHandSideExpression `||=` AssignmentExpression.

That is indeed the reason! That's the linter doing its job. I'm glad the error message was clear enough for you to figure it out. (We're all going to be beta-testing it for a bit.) Apologies for not catching this during the review.

We could define each grammar individually, or do a bit of metaprogramming for the algorithms, like AssignmentOperator did. I just don't know how to metaprogram in spec-ese.

In this instance I think it will be clearest to just do away with ShortCircuitAssignmentOperator and instead add each of the three cases individually to AssignmentExpression.

[jmdyck]

In this instance I think it will be clearest to just do away with ShortCircuitAssignmentOperator and instead add each of the three cases individually to AssignmentExpression.

With the existing production, you could do something like:

<emu-grammar>AssignmentExpression : LeftHandSideExpression ShortCircuitAssignmentOperator AssignmentExpression</emu-grammar>
<emu-alg>
  1. Let _lref_ be the result of evaluating |LeftHandSideExpression|.
  1. Let _lval_ be ? GetValue(_lref_).
  1. Let _opText_ be the source text matched by |ShortCircuitAssignmentOperator|.
  1. If _opText_ is `??=`, then
    1. If _lval_ is neither *undefined* nor *null*, return _lval_.
  1. Else,
    1. Let _lbool_ be ! ToBoolean(_lval_).
    1. If _opText_ is `||=`, then
      1. If _lbool_ is *true*, return _lval_.
    1. Else,
      1. Assert: _opText_ is `&amp;&amp=`.
      1. If _lbool_ is *false*, return _lval_.
  1. If IsAnonymousFunctionDefinition(|AssignmentExpression|) and IsIdentifierRef of |LeftHandSideExpression| are both *true*, then
    1. Let _rval_ be NamedEvaluation of |AssignmentExpression| with argument GetReferencedName(_lref_).
  1. Else,
    1. Let _rref_ be the result of evaluating |AssignmentExpression|.
    1. Let _rval_ be ? GetValue(_rref_).
  1. Perform ? PutValue(_lref_, _rval_).
  1. Return _rval_.
</emu-alg>

(I started it before @bakkot's comment and didn't want to just throw it away.)

[jmdyck]

If you do keep ShortCircuitAssignmentOperator, you'll need to reference it from Annex A.

And in any event, HasCallInTailPosition will need to be defined for the new right-hand-side(s).

[jridgewell]

Updated.

[jridgewell]

If you do keep ShortCircuitAssignmentOperator, you'll need to reference it from Annex A.

It's removed now, so I think this is done.

And in any event, HasCallInTailPosition will need to be defined for the new right-hand-side(s).

Whoops, not sure how I missed that. Updated.

[jridgewell]

This is going for Stage 4 at the July meeting, and still requires Spec Editor approvals.

/cc @syg @michaelficarra @ljharb

[ljharb]

We plan to review it during next week's editor call, if not before.

[syg]

lgtm with step linking added back