Implement logical assigment operators ||= and &&=

[andreabergia]

Turned out to be pretty simple 🙂

I had to re-number all the entries in Token because the number for the new tokens has to land between FIRST_ASSIGN and LAST_ASSIGN

Also, we get two regression in test262 (lgcl-and-assignment-operator-non-simple-lhs.js and lgcl-or-assignment-operator-non-simple-lhs.js), where the tests does something like this:

// It is a Syntax Error if AssignmentTargetType of LeftHandSideExpression is not simple.
function test() {}
test() ||= 1;

Before, the code was throwing a SyntaxError because it did not recognize the ||= operator. Now, it does, and it does not handle correctly the fact that there is a non-simple expression on the left-hand side. However, this behavior is shared with all other assignments operator (e.g., test() += 32 does not throw a SyntaxError either), so I don't think this is a real "regression".

Partially resolves #936

[p-bakker]

Great stuff!

What about ??= through? That is also part of the same proposal, see #938

As for the 'regression': Can you create a case for that (assuming you don't (directly) go off and provide a PR for that one as well :-)

[p-bakker]

BEsides the 'regression': seems not all lgcl-or/and-assignment-opearator-* are passing.

I just looked at language/expressions/logical-assignment/lgcl-or-assignment-operator-lhs-before-rhs.js for example

Aren't those fixable?

[andreabergia]

What about ??= through? That is also part of the same proposal, see #938

I know, but IMHO that is better implemented after #1591 is merged. So, I'd prefer opening a new PR after this and that one are merged.

BEsides the 'regression': seems not all lgcl-or/and-assignment-opearator-* are passing.

I just looked at language/expressions/logical-assignment/lgcl-or-assignment-operator-lhs-before-rhs.js for example

All of these problems happen also for existing operators, like +=. They should be tracked in proper issues (I'll try to open some), but they are not strictly related to the new operators that this PR adds.

[andreabergia]

Opened #1598 #1599 and #1600 which should track the test262 failures for the new operators

[gbrail]

I'm OK with merging this and opening new issues and PRs for the other features if others think this is a good PR. (looks good to me, BTW). Out of time today but it won't be long. Thanks!

[p-bakker]

This PR looks good to me as well: have no clue how the limited set of changes actually made things work 🤔, but the relevant tests in test262 pass where they can, so apparently they work 🙂 so I'm all for merging this!

[gbrail]

Thanks for doing this!

[andreabergia]

This PR looks good to me as well: have no clue how the limited set of changes actually made things work 🤔, but the relevant tests in test262 pass where they can, so apparently they work 🙂 so I'm all for merging this!

Well, very simply: we already have a lot of logic for operators such as += or |=. It was super easy to extend it to ||= because it has the same semantic.