fix(es/resolver): Resolve top-level undefined, NaN, and Infinity correctly

[Austaras]

…aration correctly in es module

Description:

For following code

var NaN = 1
console.log(NaN)

Result would be

Envirnoment	Result
Non strict script(browser, nodejs repl)	NaN
Non strict script(nodejs script)	1
Strict script(browser, nodejs repl)	runtime error
Strict script(nodejs script)	1
ESM	1

So SWC choose to behave like browser in script mode and confirm to esm standard.

BREAKING CHANGE:

Related issue (if exists):
Closes #8465

[kdy1]

I think the problem here may be the fact that we use ESM mode for terser test suites

[Austaras]

I think the problem here may be the fact that we use ESM mode for terser test suites

Yes. I would open a separate PR for this. Doesn't seem like to have any problem.

[Austaras]

Some investigation, for following code

var NaN = 1
console.log(NaN)

Result would be

Envirnoment	Result
Non strict script(browser, nodejs repl)	NaN
Non strict script(nodejs script)	1
Strict script(browser, ndoejs repl)	runtime error
Strict script(nodejs script)	1
ESM	1

Seems like the benefit of this mechanism is very limited. Maybe just remove it?

[kdy1]

I agree that the benefit is small, but I also think code is trivial enough to justify the size of benefit

[Austaras]

Fine. Let's proceed this way.

[kdy1]

Thank you all!

---

swc-bump:

• swc_ecma_ast
• swc_ecma_transforms_base

[magic-akari]

Maybe just remove it?

What does this mean? Should SWC skip them when renaming?

[Austaras]

Maybe just remove it?

What does this mean? Should SWC skip them when renaming?

Maybe just remove the special treatment for undefined etc like what babel do. But that's not very good for minifier

[kdy1]

Thank you!

outdated

[swc-bot]

Automated review comment generated by auto-rebase script