1462897 - The "better solution" is simply a more bad one

Status: RESOLVED FIXED

(Developer Documentation Graveyard :: JavaScript, defect, P2)

Description

[wiimm]

:: Developer Documentation Request

      Request Type: Correction
     Gecko Version: unspecified
 Technical Contact: 

:: Details

The number 1.005 cant be stored exact and is internally stored as 1.0049999999xxxx. So the mathematical rounding to 1.00 is absolutely correct. The "better solution" simply ignores these fact and produces other errors. So the result of round(1.0049999999999999,2) is 1.01 too -- definitely a bug!

This is a general problem of understanding, how floating point number are stored (based on power of 2; but humans write numbers based on power of 10). And so I think, than an explanation, why 1.005 id rounded to 1.00 and not to 1.01 is much  better than providing an bad solution that have other mathematical issues and is simply buggy.

Comment 1

[Tooru Fujisawa [:arai]]

I think the page is explaining completely different thing than API itself.

it's better just changing the demo code not to implement precisionRound but directly use the Math.round API itself,
and remove the "better solution" section.

Comment 2

[wiimm]

Agree, removing the "better solution" section is the best option.

Comment 3

[Tooru Fujisawa [:arai]]

Removed unrelated sections.
https://developer.mozilla.org/en-US/docs/Web/JavaScript/Reference/Global_Objects/Math/round$compare?locale=en-US&to=1385078&from=1383484

also created PR to change the demo
https://github.com/mdn/interactive-examples/pull/956

Comment 4

[Tooru Fujisawa [:arai]]

The demo is also updated.