ES5 question: Error object property 'message' - enumerable or not?

Allen Wirfs-Brock allen at
Sat Sep 3 12:58:27 PDT 2011

On Sep 3, 2011, at 11:59 AM, Thomas L. Shinnick wrote:

> An issue popped up where V8 recently made everything in Error objects non-enumerable.  This was surprising to some apps.  More surprising is the insistence that this is required by the ES5 spec.
> As a simplistic reader, when "own property" is mentioned here
> new Error (message)
>    If the argument message is not undefined, the message own property
>    of the newly constructed object is set to ToString(message).

This isn't the clearest possible language but it is clear enough when read in conjunction with the last paragraph of the introduction to section 15 which says: "Every other property described in this clause has the attributes { [[Writable]]: true, [[Enumerable]]: false, [[Configurable]]: true } unless otherwise specified".  Language, with the same effect also existed in the ES3 spec.

> I just naturally think this property it is not supposed to be mysteriously unseen.  Yet other general statements in the spec have people think everything not explicitly labeled enumerable _can't_ be.
>    "But the spec is the spec"

Which it is.  The exception is if the spec. differs from what has historically been interoperably implemented by browsers.  If most cases if the spec. differs from that then it is probably wrong. Standards such as the ES spec. exist to make sure that all implementations behave the same.  Sometimes, not everybody will like the standardized behavior but the important thing is that it is the same across all implementations 

>    "...this consequence may have been unintended."
> Does the ES5 spec intend for 'message' property of Error objects to be non-enumerable?

That's what it says and that's what you should implement unless you can demonstrate that doing so creates creates a new interoperability problem.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <>

More information about the es-discuss mailing list