510537 - ES5: Array.isArray

Status: RESOLVED FIXED

(Core :: JavaScript Engine, enhancement)

Description

[Jeff Walden [:Waldo]]

Attachment: Patch

Recent array method discussion reminded me of this addition in ES5.  Trivial patch, in-browser users want it, out-of-browser users want it, about as no-risk a feature as they get -- nice driveby pickup for 3.6.  We should skim the spec for other ten-minute features...

Test fodder:

assertEq(Array.isArray.length, 1);
var empty = [];
var fastSmall = [1];
var slow = [];
slow.slow = 0;
var slowSmall = [];
slowSmall[999999] = 0;
slowSmall.length = 0;
var all = [empty, fastSmall, slow, slowSmall];
all.forEach(function(a) { assertEq(Array.isArray(a), true); });
assertEq(Array.isArray(), false);
assertEq(Array.isArray({}), false);
assertEq(Array.isArray(null), false);
assertEq(Array.isArray(undefined), false);
assertEq(Array.isArray(17), false);
assertEq(Array.isArray("Array"), false);
assertEq(Array.isArray(Math.PI), false);
assertEq(Array.isArray(true), false);
assertEq(Array.isArray(false), false);

Comment 1

[Jason Orendorff [:jorendorff]]

Comment on attachment 394528 [details] [diff] [review]
Patch

Looks good.

Comment 2

[Jeff Walden [:Waldo]]

http://hg.mozilla.org/tracemonkey/rev/95ef863925f9

Comment 3

[Brendan Eich [:brendan]]

Any more like this missing from the es5 buglist?

Why not a traceable native?

/be

Comment 4

[Jeff Walden [:Waldo]]

I don't know if anything else is missing -- I've been too busy diving into Object.defineProperties and all those object meta-hooks at the moment to really look lately (and those are much orders of magnitude less trivial than this).  I probably should remedy that, maybe I'll do a spec skim over the weekend at some point.

As for why not a traceable native,

1) It's a driveby patch, didn't want to spend any more time on it than necessary.
2) We can do better than a traceable native; this really cries out for a bug 497173-based solution.
3) I think we get better ROI if I spend time on something like Object.defineProperties than if I spend time wringing out a couple dozen instructions' execution time for one method that nobody uses (yet).

I'm working on a web-tech post on this at the moment, although to be honest I'm less up on the reasons why someone would want to know if a value is an array than I'd like to be.  I'm probably going to lean a little on this URL for rationales and use cases for doing so, maybe ask a few people around Mozilla as well.  No rush, especially since this wants to be on branch before anything would actually be published.

http://thinkweb2.com/projects/prototype/instanceof-considered-harmful-or-how-to-write-a-robust-isarray/

Comment 5

[Brendan Eich [:brendan]]

Just askin', no sweat. Indeed bug 497173 approach could do a lot better than a traceable native!

You might ask jresig, bgalbraith, and dalmaer about isArray use cases.

/be

Comment 6

[Jeff Walden [:Waldo]]

Draft post:

http://web.mit.edu/jwalden/www/isArray.html

(Mother Superior) jumping the gun a bit as this hasn't made 192 yet, but that'll happen eventually enough to make it worth posting this now...

Comment 7

[Robert Sayre]

http://hg.mozilla.org/mozilla-central/rev/95ef863925f9

Comment 8

[Eric Shepherd [:sheppy]]

Would appreciate it if someone would let me know when this officially lands on 1.9.2.

Comment 9

[Eric Shepherd [:sheppy]]

Is this going to land on 1.9.2?

Comment 11

[Juriy "kangax" Zaytsev]

I wonder why searching for "Array.isArray" in bugzilla (before I created bug 537548) returned 0 results. Sigh.

Comment 12

[Fabian (Crash) Grutschus]

dev-doc under https://developer.mozilla.org/en/Core_JavaScript_1.5_Reference/Global_Objects/Array/isArray

Comment 13

[Jeff Walden [:Waldo]]

The eventual blog post following on comment 6 was made here:

https://developer.mozilla.org/web-tech/2010/07/26/determining-with-absolute-accuracy-whether-or-not-a-javascript-object-is-an-array/