Update test262 suite to latest and fix using/await using issues

[lahma]

Summary

• Fix IsUsingKeyword to correctly handle of as a variable name in for-loops with using/await using declarations
• Add opt-in AllowCallExpressionAsLhs parser option for AnnexB B.3.9 (Runtime Errors for Function Call Assignment Targets)
• Update test262 suite SHA and allow-list

IsUsingKeyword fix

The isFor && id is "of" check was too aggressive — it rejected of as a binding name without looking ahead. Now disambiguates:

Syntax	Interpretation
for (using of = null;;)	Regular for-loop, of is variable name
for (using of of [])	For-of loop, of is binding name
for (await using of of [])	Await-using, of is binding name
for (using of <expr>)	Not a using keyword, using is expression

AllowCallExpressionAsLhs

Per AnnexB B.3.7, in non-strict mode f() = g() should parse successfully and throw ReferenceError at runtime (not SyntaxError at parse time). The new option (defaults to false) allows CallExpression as LHS in:

• Simple and compound assignments
• Prefix/postfix update expressions
• For-in/for-of loop left-hand side

Test plan

• All edge cases verified (using of =, using of of, await using of of, using as expression)
• Acornima unit tests pass (10908/10909, 1 pre-existing net10.0 failure in CanHandleDeepRecursion on Windows fixed by adjusting limits)
• test262 suite: all tests now pass, 0 regressions

🤖 Generated with Claude Code

UPDATE

The IsUsingKeyword fix has been corrected as it didn't handle some edge cases (e.g. for (using of == null;;) or for (using ofx of []), etc.)

The AnnexB B.3.9 stuff has been extracted to a separate PR (#33).

[adams85]

Thank you for submitting these corrections. I'll need some time to review them thoroughly, but I have some preliminary remarks:

• The current implementation of (await) using declarations in for loops looks buggy indeed. AFAICS, these problems are also present in the upstream project.

• However, I have some concerns about the "Runtime Errors for Function Call Assignment Targets" stuff:

  On one hand, acorn directly supports Annex B.3 features in non-strict mode, without putting those behind some kind of feature flag. To follow this convention, we probably shouldn't introduce the AllowCallExpressionAsLhs switch.

  On the other hand, this feature looks like a recent addition, which is only present in the draft of ES2026. At least, I couldn't find this in the spec of earlier versions. And, as you're surely aware Acornima is ES2023-compliant at the moment. (BTW, the reference is wrong, it's B.3.9, not B.3.7).

  So I'm a bit puzzled as to what to do with this. Not even the upstream project has implemented this yet. Maybe we'd better wait for that.

[lahma]

Thank you for the feedback these were issues that were somewhat blocking Jint's test suite, but not a pressing concern or anything. Feel free to adjust and throw away things (or create a separate change set that you feel comfortable with).

[adams85]

FYI, I made some corrections to the using/await using fix (see the updated PR description) and merged #32 into this PR. I plan to merge and release it soon.

I also extracted the AnnexB B.3.9 stuff into a separate a PR, which I'm going to leave as a draft for now, until I make up my mind what to do with it.

Is this okay with you?

[lahma]

Is this okay with you?

I'm happy if I'm of any help, even with the AI tooling. Feel free to do whatever you feel is best, it's your codebase, I'm not looking for attribution either. This pushes Jint forward so that's just my selfish goal 👍

[adams85]

These bugfixes are tremendously useful, keep 'em coming! 👍

In general, draft/experimental features are welcome too, but those may be put on hold as I want to keep the project as close to upstream as possible to avoid extra work and ensure maintainability in the long run.

So, in the case of features, it's best to have a discussion beforehand to find out whether the time is ripe.

[adams85]

@lahma FYI, I backported the using and Annex B.3.9 fixes to acornjs. The latter hasn't been accepted yet (might never be), however I decided to release this pack anyway so you can move forward with Jint.

[lahma]

Thank you for the update, Jint has integrated the version ( sebastienros/jint#2374 ) and again we got rid of more exclusions 🥳