fix(es/minifier): Don't track usage assign

[Austaras]

Description:
Don't know if it's right, but it generate correct output.

BREAKING CHANGE:

Related issue (if exists):

[kdy1]

Thank you

[Austaras]

Oh, sorry

[kdy1]

I'm not sure if we are proving these facts.
I added two-way infection, which is clearly not optimal, because it prevents bugs

[Austaras]

Yeah, if isObject is a normal function it would work fine. Maybe there's some more complex mechanism.

[Austaras]

The difference can be run down to https://github.com/terser/terser/blob/master/lib/compress/tighten-body.js#L887. It checks if it's a AST_SymbolDefun(definition of function) or defined in current scope, which I believe is just a convenient implementation.

[kdy1]

Oh... It's bit shocking but if so I think this PR is fine

[Austaras]

The difference can be run down to https://github.com/terser/terser/blob/master/lib/compress/tighten-body.js#L887. It checks if it's a AST_SymbolDefun(definition of function) or defined in current scope, which I believe is just a convenient implementation.

So basically function isObject is fine, but const isObject = function is fine only if it's defined in same scope of newValue.

[kdy1]

This PR seems innocent.
I'll merge this after merging #6053

---

swc-bump:

• swc_ecma_minifier

[kdy1]

Thank you!