why resolver ignore NamedExport which has source , keep ctxt as zero

[stormslowly]

// case 1
let a= 1;
export {a as b}

// case 2
export {a as b } from "c"

in case 1, b has been marked under unresolved marks.
but
in case 2 , both a and b's ctxt are all 0

swc/crates/swc_ecma_transforms_base/src/resolver/mod.rs

Lines 1079 to 1085 in 2113bb3

	fn visit_mut_named_export(&mut self, e: &mut NamedExport) {
	if e.src.is_some() {
	return;
	}
	e.visit_mut_children_with(self);
	}

the code shows that, when NamedExport with source, resolver will not mark its ctxt.

anyone know why ?

[stormslowly]

why i care about the ctxt, because i have a requirement, for example
if b is a not defined variable, a stmt will be inject into the module eg var b = require("some-mock-b")

i use ctxt.outer() == unresolved_mark, to check if it defined or not, in case 1, b is treated as unresolved; so an un-expected require inject to code.

in case 2, the ctxt is undermark root, so inject not happened.

[kdy1]

It's not a thing that the core is responsible for. Write your own one.

[stormslowly]

yes, i has wrote a own implementation for injecting. my question is why

1. export {a as b}, b is marked as unresolved
2. export {a as b } from "c" all idents' ctxt are zero .

all these by design or by mistake?

[kdy1]

By design 2023년 12월 9일 (토) 오전 8:38, pshu ***@***.***>님이 작성:

…

yes, i has wrote a own implementation for injecting. my question is why 1. export {a as b}, b is marked as unresolved 2. export {a as b } from "c" all idents' ctxt are zero . all these by design or by mistake? — Reply to this email directly, view it on GitHub <#8408 (reply in thread)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AHELSJ3QGPDXQGQAD2LSB3TYIOQGDAVCNFSM6AAAAABAM5KRWOVHI2DSMVQWIX3LMV43SRDJONRXK43TNFXW4Q3PNVWWK3TUHM3TQMBUGI3TK> . You are receiving this because you commented.Message ID: ***@***.***>

[stormslowly]

could you plz give me some hint why design like this.
current ctxt makes me a little confused : (

thank you so much : P