[Webkit-unassigned] [Bug 48308] Too precise serialization from floating point number to string for "number" input elements

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=48308





--- Comment #27 from Dai Mikurube <dmikurube at google.com>  2010-11-02 04:50:04 PST ---
(In reply to comment #26)
Hi,

For the first one, I think we have no special reason to give up for small values with exponential expression like "1.02E3" or "3.3013E-5". I tried handling values with exponential expressions in the Patch, but I will discard it if it results worse than nothing.

For the second one, I agree with your suggestion. Another reason is that digits around zero are not significant when newValue > 10^21. Because 2^53 < 10^21. (Actually 2^53 < 10^17 < 10^23.)

Then, may I ask your opinion on the upper limit of DP? My thought :

* Naturally, it must be smaller than 21 because of the same ("another") reason above. Multiplying by 10^21 occurs unnecessary rounding error.
* Actually, I guess at most 16 is better for the same reason. 2^53 < 10^17. (I used 16 in the Patch. Multiplying 10^16 for numbers with DP > 16.)
* But it makes "step=0.000000000000000001" incorrect though it is a valid number > FLT_MIN.
* Doing no truncation for any number with DP > 16 (like 0.123456789012345678) is another option.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.