WIP 6

TODO.md

@@ -1,28 +1,25 @@
 # TODO
 
-* Handle setting `binding.isFrozenName` on `this` better. Better to set it, but just not to delete internal var trails? Actually I don't think it makes much odds. There's nowhere in a class constructor where `eval()` can go where it can't access `this$0`, and if `eval()` in nested function, `this$0` still needs to be frozen as it's in a scope above the `eval()`. So the implementation as is already optimal.
-* No point setting `isFrozenName` on external vars passed to serializer? (as it depends entirely on `containsEval` anyway)
-  * Right now, it is pointless, but in future when scope tree is built dynamically, it would allow
-    pulling vars which aren't accessible by `eval()` up to a higher level in the tree so they can be renamed.
-  * `serialize/blocks` + `serialize/functions` + `serialize/parseFunction` could have changes stripped out if so.
-* TODO comment in `serialize/parseFunction`
+* Tests
 * Deal with duplicate bindings:
   * Hoisted sloppy functions - actually no need, because `isFunction` means both are frozen anyway
   * `for (let x of (() => x, [])) eval('x');` - done, but write test
   * `function (x, _ = eval('x')) { var x; return x; }` - done (apart from `arguments`), but write test
-* Tests
 * Tests for binding in higher scope which is shadowed still not being renamed
   e.g. `() => { let x = 1; { let x; eval('typeof a !== "undefined" && a = 2'); } return x; }` - make sure param `x` not renamed to `a`
   * Make sure this includes vars which are illegal to access inside `eval()` if `eval()` is strict mode.
 * Test for const violations with frozen vars when internal to function:
+
 ```js
 module.exports = function(module, exports) {
 	const x = 1;
 	eval('x');
 	return () => { x = 2; };
 };
 ```
+
 Also where another binding which isn't frozen:
+
 ```js
 module.exports = function(module, exports) {
 	const x = 1;