504957 - Assignment-to-const is not removed with rhs of "a.@b::c"

Status: RESOLVED WORKSFORME

(Core :: JavaScript Engine, defect)

Description

[Jesse Ruderman]

js> f = (function x() { x = a.@b::c; });
function x() {
    x = a.@[b::c];
}

js> eval(uneval(f));
function x() {
    a.@[b::c];
}

Gary found this bug (the round-trip change tripped jsfunfuzz).

Comment 1

[Jason Orendorff [:jorendorff]]

This doesn't seem to happen anymore.

Anyway, it would be a bug to remove the rhs since it can throw (if a is undefined, for example).

Comment 2

[Jesse Ruderman]

Here's what happens now:

js> f = (function x() { x = a.@b::c; });
(function x() {x = a.@[b::c];})
js> eval(uneval(f));
(function x() {x = a.@[b::c];})

I think the "remove assignment to const" optimization no longer kicks in when it's an assignment to the name of the function.  In contrast, assigning to a local const does get removed.

js> f = (function () { const x; x = a.@b::c; });
(function () {const x;a.@[b::c];})

The round-trip problem seems to be gone in all cases :)