1068612 - toPrecision accepts arguments that are larger than 21

Status: RESOLVED WONTFIX

(Core :: JavaScript Engine, defect)

Description

[Boris Zbarsky [:bzbarsky]]

Per spec, this should throw RangeError:

  (123.445).toPrecision(55)

though it _is_ kinda useful that it works... ;)

Comment 1

[André Bargull [:anba]]

This is allowed per ES5.1, ch. 16 and ES6-draft, ch. 16 [1]:

> An implementation may define behaviour other than throwing RangeError for
> toFixed, toExponential, and toPrecision when the fractionDigits or
> precision argument is outside the specified range.


[1] https://people.mozilla.org/~jorendorff/es6-draft.html#sec-error-handling-and-language-extensions

Comment 2

[Boris Zbarsky [:bzbarsky]]

Gah.  Then why is that not specified in the toPrecision algorithm itself?  This sort of spec monkeypatching is ridiculous.  :(

Comment 3

[André Bargull [:anba]]

The allowed extensions are also mentioned after the algorithms for toPrecision() and friends [1]. Maybe an additional in-line note in the actual algorithm should be added?

[1] https://people.mozilla.org/~jorendorff/es6-draft.html#sec-number.prototype.toprecision

Comment 4

[Boris Zbarsky [:bzbarsky]]

Yeah.  That would be a lot better than this COMEFROM style of specification.

Comment 5

[André Bargull [:anba]]

Filed https://bugs.ecmascript.org/show_bug.cgi?id=3230

Comment 6

[André Bargull [:anba]]

Resolving as Won't Fix - Higher precision is allowed per spec.

Comment 7

[Daniel Ehrenberg]

I think it could still make sense to revisit this; posted a little more at https://bugzilla.mozilla.org/show_bug.cgi?id=90551#c8 (sorry, not sure how to mark it as a duplicate here). Does anyone have an idea what the motivation for the larger range beyond integer precision is? If the larger range is useful, it could be implemented in other engines and standardized that way.