We need a test here, I think it is unnecessary, because we already renamed them before

Just try

const renderedModule = this.renderModule(
					m,
					chunkRenderContext,
					hooks,
					false
				);

And the output the same

We need a test here, I think it is unnecessary, because we already renamed them before

It's necessary. The initial value of ⁠renamedInlinedModule is false. If ⁠avoidEntryIife is enabled, ⁠getRenamedInlineModule will check for other reasons to bail out of the IIFE. If any exist, ⁠renamedInlinedModule will remain false because the inlined module will still be wrapped in an IIFE for those reasons. Therefore, we skip the optimization here.

I ask this because of

Do we really need source.replace? We need a test here, because our tests don't need renaming, so i thought we could avoid this here

If we don't need renaming, we can simplify our code quite a bit.

@fi3ework I still think we have wrong logic, because if I remove

for (const variable of usedInNonInlined) {
				const references = getAllReferences(variable);
				const allIdentifiers = new Set(
					references.map(r => r.identifier).concat(variable.identifiers)
				);

				const newName = this.findNewName(
					variable.name,
					usedNames,
					m.readableIdentifier(runtimeTemplate.requestShortener)
				);
				usedNames.add(newName);
				for (const identifier of allIdentifiers) {
					const r = /** @type {Range} */ (identifier.range);
					const path = getPathInAst(ast, identifier);
					if (path && path.length > 1) {
						const maybeProperty =
							path[1].type === "AssignmentPattern" && path[1].left === path[0]
								? path[2]
								: path[1];
						if (maybeProperty.type === "Property" && maybeProperty.shorthand) {
							source.insert(r[1], `: ${newName}`);
							continue;
						}
					}
					source.replace(r[0], r[1] - 1, newName);
				}
			}

Everything works fine

As I said before, we already doing renaming before (in concatination modules) and I don't think we need to do it here

Just put console.log or use debugger for fast debug and we never reach this branches

We already continue here:

if (usedInNonInlined.size === 0) {
  renamedInlinedModules.set(m, source);
  continue;
}

The case in test/configCases/output-module/iife-entry-module-with-others/index.js It's not working as well as it should. I'm fixing it. I checked the 'test/cases/entry-inline' case, and it is still working.

I made https://stackblitz.com/edit/node-ojwwax, you could check localVar in dist-5.91.0 and dist-5.94.0 to see the difference. In 5.94.0, localVar is renamed to src_localVar by renaming.

@alexander-akait Updated, please check the new test case now.
The current test cases serve the following purposes, for instance with iife-entry-module-with-others:

1. Assert the correct IIFE comment in the bundle.
2. Assert the correct runtime behavior of the bundle.
3. If the reason branch it need to be isolated against other entry modules. is deleted, the test will fail

The other test cases are similar, iife-multiple-entry-modules has an extra avoidEntryIife configuration toggle test, which can be seen as a supplement to test/cases/entry-inline.