Specify the with-fractional-digits protocol #177
Conversation
jessealama
force-pushed
the
remove-trailing-zeroes
branch
from
fe2f84e
to
3481c4d
Compare
| 1. If _digits_ is not a Number, throw a *TypeError* exception. | ||
| 1. If _digits_ is *NaN*<sub>𝔽</sub>, *+∞<sub>𝔽</sub>*, or *+∞<sub>𝔽</sub>*, throw a *RangeError* exception. | ||
| 1. If ℝ(_digits_) is not an integer, throw a *TypeError* exception. | ||
| 1. IF ℝ(_digits_) < 0, throw a *RangeError* exception. |
There was a problem hiding this comment.
The name .withFractionalDigits might not be ideal, but it'd be great to allow negative numbers here.
1234..withFractionalDigits(1) is 1234.0, and 1234..withFractionalDigits is 123e1.
Let's see if this is desired though. If nobody cares it's fine to throw for simmetry with .toFixed().
| <h1>get %WithFractionalDigitsPrototype%.precision</h1> | ||
| <p>The <q>precision</q> that the number is understood to have.</p> |
There was a problem hiding this comment.
The reason I called this .fractionalDigis is because it's a different concept of "precision" from the .toPrecision() method (once starts counting from the units, the other one from the most-significant digit).
Also, I really dislike .number exactly because it might not be a .number value. I would make it explicit that the naming is very much open for bikeshed between Stage 2 and 2.7/
There was a problem hiding this comment.
Agree. I'm not happy with the names, either. I chose "precision" because that's what we use instead of "quantum". I'm happy to just roll with these names and bikeshed later.
Co-authored-by: Nicolò Ribaudo <[email protected]>
Co-authored-by: Nicolò Ribaudo <[email protected]>
Co-authored-by: Nicolò Ribaudo <[email protected]>
Co-authored-by: Nicolò Ribaudo <[email protected]>
Co-authored-by: Nicolò Ribaudo <[email protected]>
Co-authored-by: Nicolò Ribaudo <[email protected]>
Co-authored-by: Nicolò Ribaudo <[email protected]>
| 1. Perform ? RequireInternalSlot(_O_, [[Prototype]]). | ||
| 1. If _O_.[[Prototype]] is not %WithFractionalDigitsPrototype%, throw an *TypeError* exception. |
There was a problem hiding this comment.
Also change the brand check here
| 1. If _options_ is not an Object, throw a *TypeError* exception. | ||
| 1. Let _N_ be ! Get(_O_, *"number"*). | ||
| 1. Let _P_ be ! Get(_O_, *"precision"*). | ||
| 1. Let _toLocaleString_ be ! Get(_N_, *"toLocaleString"*). |
| 1. Set _options_.[[minimumFractionalDigits]] to _P_. | ||
| 1. Set _options_.[[maximumFractionalDigits]] to _P_. |
There was a problem hiding this comment.
These should be properties, not internal slots.
Also, I am thinking that probably it would be better to not install extra properties on a user-provided config object. Instead, we should leave the implementation of this .toLocaleString to be ECMA-402 defined, and then in ECMA-402 we do we use this.[[Precision]] as a fallback for when there is no minimum/maximumFractionalDigits specified in the options.
This fallback should happen in NumberFormat.prototype.format (which will receive the WithFractionalDigits object as an argument), and it should check if the constructor had the option explicitly set.
| 1. IF ℝ(_digits_) < 0, throw a *RangeError* exception. | ||
| 1. Let _obj_ be OrdinaryObjectCreate(%WithFractionalDigitsPrototype%, « [[Number]], [[Precision]] »). | ||
| 1. Set _obj_.[[Number]] to _O_. | ||
| 1. Set _obj_.[[Precision]] to _digits_. |
There was a problem hiding this comment.
| 1. Set _obj_.[[Precision]] to _digits_. | |
| 1. Set _obj_.[[Precision]] to ℝ(_digits_). |
No description provided.