Implement Number.prototype.clz()

[Dániel Bátyai]

There is a Number.prototype.clz() specification in the ES6 standard, which is not yet implemented in JSC.

[Dániel Bátyai]

Created attachment 225453 [details]
Proposed patch

[Anders Carlsson]

Comment on attachment 225453 [details]
Proposed patch

View in context: https://bugs.webkit.org/attachment.cgi?id=225453&action=review

> Source/JavaScriptCore/runtime/NumberPrototype.cpp:472
> +    int zeroCount = 0;
> +    for (int i = 31; i >= 0; i--) {
> +        if (!(number >> i))
> +            zeroCount++;
> +        else
> +            break;
> +    }
> +    return JSValue::encode(jsString(exec, String::numberToStringECMAScript(zeroCount)));

I think this should be a separate function and also use __builtin_clz when building with compilers that support it (GCC, clang).

[Geoffrey Garen]

Comment on attachment 225453 [details]
Proposed patch

Is coz really supposed to return a string? That seems impossible, given that its goal is performance.

[Geoffrey Garen]

*clz

[Sam Weinig]

(In reply to comment #3)
> (From update of attachment 225453 [details])
> Is [clz] really supposed to return a string? That seems impossible, given that its goal is performance.

http://people.mozilla.org/~jorendorff/es6-draft.html#sec-number.prototype.clz

Seems like the answer is it is not supposed to return a string.

[Dániel Bátyai]

Created attachment 225661 [details]
Proposed patch

Modified according to comments.

Unfortunately, __builtin_clz(0) was returning 31 instead of the 32 specified in the standard, so I had to add a check for that.

[WebKit Commit Bot]

Comment on attachment 225661 [details]
Proposed patch

Clearing flags on attachment: 225661

Committed r165047: <http://trac.webkit.org/changeset/165047>

[WebKit Commit Bot]

All reviewed patches have been landed.  Closing bug.