Fix Decimal.floor() + Decimal.ceiling() for most non-integral values.

Fix Decimal.floor() + Decimal.ceiling() for many non-integral values.

Extend the check for an integral value to consider all the digits
being truncated, and not just the rightmost/most-significant one, when
rounding the result towards positive and negative infinity for floor()
and ceiling(), respectively.

R=yosin,mkwst
BUG=234323
Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=162197

[sof, 2013-11-15 16:01:23]

At your leisure, please take a look.

[yosin_UTC9, 2013-11-18 03:48:28]

Thank for improvment of Decimal package!

https://codereview.chromium.org/59603006/diff/1/Source/platform/Decimal.cpp
File Source/platform/Decimal.cpp (right):

https://codereview.chromium.org/59603006/diff/1/Source/platform/Decimal.cpp#n...
Source/platform/Decimal.cpp:617: static bool digitsAllZero(uint64_t coefficient,
int n)
How about this?

static bool isMultipleOfTenPowerN(uint64_t coefficient, int n) {
    uint64_t tenPowerN = scaleUp(1, n);
    return !coefficient % tenPowerN;
}

log(N, 2) multiplication + one division may be faster than 2xN division.

https://codereview.chromium.org/59603006/diff/1/Source/platform/Decimal.cpp#n...
Source/platform/Decimal.cpp:625: n--;
nit: --n
Other parts of this file use prefix decrement/increment.

https://codereview.chromium.org/59603006/diff/1/Source/platform/Decimal.cpp#n...
Source/platform/Decimal.cpp:648: result++;
nit: ++result
Other parts of this file use prefix decrement/increment.

https://codereview.chromium.org/59603006/diff/1/Source/platform/Decimal.cpp#n...
Source/platform/Decimal.cpp:689: result++;
nit: ++result
Other parts of this file use prefix decrement/increment.

[sof, 2013-11-18 06:48:27]

Thanks Yoshi, the Decimal code is v. clean & clear.

https://codereview.chromium.org/59603006/diff/1/Source/platform/Decimal.cpp
File Source/platform/Decimal.cpp (right):

https://codereview.chromium.org/59603006/diff/1/Source/platform/Decimal.cpp#n...
Source/platform/Decimal.cpp:617: static bool digitsAllZero(uint64_t coefficient,
int n)
On 2013/11/18 03:48:28, Yoshi wrote:
> How about this?
> 
> static bool isMultipleOfTenPowerN(uint64_t coefficient, int n) {
>     uint64_t tenPowerN = scaleUp(1, n);
>     return !coefficient % tenPowerN;
> }
> 
> log(N, 2) multiplication + one division may be faster than 2xN division.
> 

Yes, clearer too. Done + took the opportunity to spot the zero special case.

https://codereview.chromium.org/59603006/diff/1/Source/platform/Decimal.cpp#n...
Source/platform/Decimal.cpp:648: result++;
On 2013/11/18 03:48:28, Yoshi wrote:
> nit: ++result
> Other parts of this file use prefix decrement/increment.

Done.

https://codereview.chromium.org/59603006/diff/1/Source/platform/Decimal.cpp#n...
Source/platform/Decimal.cpp:689: result++;
On 2013/11/18 03:48:28, Yoshi wrote:
> nit: ++result
> Other parts of this file use prefix decrement/increment.

Done.

[Mike West, 2013-11-18 07:09:58]

LGTM, but you'll need a platform/ owner to sign off on it. Jochen, maybe?

On 2013/11/18 06:48:27, sof wrote:
> Thanks Yoshi, the Decimal code is v. clean & clear.
> 
> https://codereview.chromium.org/59603006/diff/1/Source/platform/Decimal.cpp
> File Source/platform/Decimal.cpp (right):
> 
>
https://codereview.chromium.org/59603006/diff/1/Source/platform/Decimal.cpp#n...
> Source/platform/Decimal.cpp:617: static bool digitsAllZero(uint64_t
coefficient,
> int n)
> On 2013/11/18 03:48:28, Yoshi wrote:
> > How about this?
> > 
> > static bool isMultipleOfTenPowerN(uint64_t coefficient, int n) {
> >     uint64_t tenPowerN = scaleUp(1, n);
> >     return !coefficient % tenPowerN;
> > }
> > 
> > log(N, 2) multiplication + one division may be faster than 2xN division.
> > 
> 
> Yes, clearer too. Done + took the opportunity to spot the zero special case.
> 
>
https://codereview.chromium.org/59603006/diff/1/Source/platform/Decimal.cpp#n...
> Source/platform/Decimal.cpp:648: result++;
> On 2013/11/18 03:48:28, Yoshi wrote:
> > nit: ++result
> > Other parts of this file use prefix decrement/increment.
> 
> Done.
> 
>
https://codereview.chromium.org/59603006/diff/1/Source/platform/Decimal.cpp#n...
> Source/platform/Decimal.cpp:689: result++;
> On 2013/11/18 03:48:28, Yoshi wrote:
> > nit: ++result
> > Other parts of this file use prefix decrement/increment.
> 
> Done.

[jochen (gone - plz use gerrit), 2013-11-18 11:50:29]

rubberstamp lgtm

[commit-bot: I haz the power, 2013-11-18 11:53:18]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/sigbjornf@opera.com/59603006/50001

[commit-bot: I haz the power, 2013-11-18 12:50:57]

Change committed as 162197