FW: use of SameValue in Array.prototype.indexOf and lastIndexOf

Allen Wirfs-Brock Allen.Wirfs-Brock at microsoft.com
Thu Aug 13 12:23:42 PDT 2009


This was sent to es5-discuss even though the original thread was on es-discuss

-----Original Message-----
From: es5-discuss-bounces at mozilla.org [mailto:es5-discuss-bounces at mozilla.org] On Behalf Of Brendan Eich
Sent: Thursday, August 13, 2009 12:07 PM
To: Mark S. Miller
Cc: es5-discuss at mozilla.org
Subject: Re: use of SameValue in Array.prototype.indexOf and lastIndexOf

On Aug 13, 2009, at 9:46 AM, Mark S. Miller wrote:

>> but if during field-testing we subsequently (possibly after the spec
>> has been finalized) find that the change breaks enough web content
>> which is
>> commonplace or otherwise hard to change, we will have to revert.
>
> Yes, exactly. And in case we find sufficient web content
> incompatibilities, we agreed at Redmond that we will change the
> "finalized" spec to match as well. That was the point of my "cross
> examination" (and sorry that it came across that way). I asked my
> question in light of all the talk about "finalizing" the spec before
> we'd begun serious web testing. I was surprised and relieved to find
> that we could still adjust the spec after it is "final" while keeping
> the same edition number. I was sufficiently surprised that I wanted to
> check someone else's recollection before stating my summary of it.
> Have I misunderstood the process?

There's still a point of no return for the ES5 draft to become an Ecma
standard at this December's Ecma GA meeting, after which it JTC1 fast-
tracks to ISO. Changing it later without generating a new edition
number is possible, but that would be later -- much later (a year?).
However, we would have more time to get feedback from a trial
implementation by then. Still, in the worst case it would thrash the
spec around one of its edges, and set a bad precedent.

This is all a bit vague, and I'm still hoping John Neumann will give
more detail here. FWIW, I wish we had made the change a year ago, but
it fell off my radar after that thread with Maciej's testing
recommendation.

As I said to you in private mail, I think Allen (ES5 Editor) gets to
call this shot. I'm not convinced that we could not "get away with
it", but I'm more skeptical than you (due to the -0 vs 0 issue in
particular), and in principle I am less inclined to make a breaking
change to the spec, then test it after the spec is finalized for
Ecma / ISO, and then patch the spec if needed.

That objection in principle is probably more about costs in our
"compatibility bookkeeping" and "standards body protocols", than about
any cost felt by web developers or browser users -- I don't know for
sure, but the policy smell is likely worse than the actual effects
outside of TC39. It seems to be an over-optimization we don't need,
not until we have more SameValue uses in Harmony (Set, Map, egal, ...).

/be
_______________________________________________
es5-discuss mailing list
es5-discuss at mozilla.org
https://mail.mozilla.org/listinfo/es5-discuss



More information about the es-discuss mailing list