193555 - 1.5R4 regression: function expression has no access to its name

Status: VERIFIED FIXED

(Rhino Graveyard :: Core, defect)

Description

[Igor Bukanov]

In 1.5R4 a function expression can not access its function name. To test, run
the following test case in the Rhino shell with arbitrary optimization:

var x = function f() { return f.toString(); };

var ok = (x.toString() === x());

print(ok ? "OK" : "FAILED");


In Rhino 1.5R3 it prints OK while running 1.5R4 produces:

js: "/home/igor/js/x/f_rec.js", line 2: uncaught JavaScript exception:
ReferenceError: "f" is not defined. (/home/igor/js/x/f_rec.js; line 2)

Comment 1

[Igor Bukanov]

Attachment: Fix: replacing vars.hasLocal(name) by !vars.hasLocal(name)

I wish I would be more careful when made the changes to VraibaleTable. In
NodeTransformer.addVariables() in 1.4R3 had a condition "vars.getLocal(name) ==
null" which I replaced by "vars.hasLocal(name)" thus disabling access to the
function name. On the other hand even the test suite does not have a check for
that, so it seems this feature quite unknown.

In any case, this one line patch to NodeTransformer.addVariables() restores the
functionality.

Comment 2

[Phil Schwartau]

Testcase added to JS testsuite:

      mozilla/js/tests/ecma_3/Function/regress-193555.js

Comment 3

[Phil Schwartau]

Testcase passes in SpiderMonkey; and as Igor reports, it passes if I run
version 1.5R3 of Rhino. Fails in 1.5RC4 with the error Igor reports -

Comment 4

[Igor Bukanov]

Marking fixed as I commited the fix some time ago

Comment 5

[Phil Schwartau]

Verified FIXED. 

The above testcase now passes in the Rhino shell, in both 
compiled and interpreted mode -

Comment 6

[Igor Bukanov]

Targeting as resolved against 1.5R4.1