416353 - Error in a setter under JSOP_SETNAME generates bogus error messages

Status: RESOLVED INVALID

(Core :: JavaScript Engine, defect)

Description

[Igor Bukanov]

[This is a spin of of the bug 385133]

Consider the following test case:

this.x setter = Object.prototype.watch;

function f() {
    x = 1;
}

f();

It is execution under the js shell generates an exception:

~/m/ff/mozilla/js/src $ ./Linux_All_DBG.OBJ/js ~/m/y.js
/home/igor/m/y.js:4: TypeError: x is not a function

An exception is expected here. "x = 1" will invoke the watch function as a setter via watch.call(this, 1) when the function expects the second argument which should be a function.

But the text of the exception is misleading and it would be nice to report the real source of the error.

The error message is generated during decompilation of

bindname "x"
one
setname "x"

when GetOff from jsopcode.c invokes recursively js_DecompileValueGenerator at
http://lxr.mozilla.org/seamonkey/source/js/src/jsopcode.c#794 . The comment before GetOff indicates that such recursion is a recovery procedure and indicates a bug in the decompiler. As such it can be related to bad error message.

Comment 1

[André Bargull [:anba]]

Object.prototype.watch and old setter functions -> INVALID.