Remove some extra calls in date.js.

Remove some extra calls in date.js.

A few hot Date.prototype functions used to be implemented as
DateGetFoo calling GetFooFrom(this), and other callers of GetFooFrom
did repeated date value extraction.  In this change GetFooFrom
functions were inlined into the functions on the prototype and the
other callers switched to using date values directly.

Committed: http://code.google.com/p/v8/source/detail?r=5027

[Vyacheslav Egorov (Chromium), 2010-07-05 08:46:35]

LGTM with comment addressed.

http://codereview.chromium.org/2841041/diff/1/2
File src/date.js (right):

http://codereview.chromium.org/2841041/diff/1/2#newcode578
src/date.js:578: function DateValueGetDate(t) {
I don't like that (1) helper functions are now kinda mixed with "interface" of
Date specified by ECMA. (2) comments referencing ECMA disappeared from functions
like DateGetFullYear() etc.

[Lasse Reichstein, 2010-07-05 10:35:22]

Drive-by whining.

http://codereview.chromium.org/2841041/diff/5001/6001
File src/date.js (right):

http://codereview.chromium.org/2841041/diff/5001/6001#newcode897
src/date.js:897: date = %_ArgumentsLength() < 2 ? %_CallFunction(this,
DateGetUTCDate) : ToNumber(date);
The DateGetUTCDate function is externally available, and is also called
internally.
I consider that a code smell, because it means that the function serves two
purposes - satisfying the interface and being a sub-computation of another
function (which is also why you need to use CallFunction - it doesn't have the
signature of a helper function, so you need to circumvent that).
However, I also don't like code duplication, so inlining it in both places isn't
a good idea either.
I think I actually prefer the helper function used in both places.

(For the functions that aren't reused, inlining is fine).

[Vitaly Repeshko, 2010-07-05 14:34:45]

o Restored the original order and references to ECMA.

o Moved shared code to macros. It turned out we need less of them than I
originally thought.

Please take another look.


Thanks,
Vitaly

[Vyacheslav Egorov (Chromium), 2010-07-05 14:52:18]

There is still a slight duplication of NaN handling but I thinks it is
unavoidable. 

LGTM

http://codereview.chromium.org/2841041/diff/9001/10002
File src/macros.py (right):

http://codereview.chromium.org/2841041/diff/9001/10002#newcode157
src/macros.py:157: macro NAN_OR_MS_FROM_TIME(time) = (NUMBER_IS_NAN(time) ?
time: (Modulo(time, 1000)));
define in terms of MS_FROM_TIME?

whitespace before colon