Implement SkPoint::Normalize() for SK_SCALAR_IS_FIXED and add performance warning to its declaration

[epoger, 2011-06-13 19:05:17]

rearranged per conversation with Mike

[epoger, 2011-06-13 21:33:46]

back to you... please let me know what you think of this attempt at a new level
of tests

[reed1, 2011-06-13 21:50:26]

Possibly a new CL,

Change the nearlys impl to be <= tolerance (so we can pass 0) and adjust the
assert(tol > 0) accordingly.

http://codereview.appspot.com/4582043/diff/11001/tests/core/SkPointTest.cpp
File tests/core/SkPointTest.cpp (right):

http://codereview.appspot.com/4582043/diff/11001/tests/core/SkPointTest.cpp#n...
tests/core/SkPointTest.cpp:12: point.set(SkIntToScalar(3), SkIntToScalar(4));
might drop these two asserts

http://codereview.appspot.com/4582043/diff/11001/tests/core/SkPointTest.cpp#n...
tests/core/SkPointTest.cpp:17: scalar = SkPoint::Normalize(&point);
SkScalar s1 = point.length();
SkScalar s2 = SkPoint::Normalize(&point);
ASSERT(s1 == s2);

nearly(s1, 5);

[epoger, 2011-06-14 13:07:15]

On 2011/06/13 21:50:26, reed1 wrote:
> Possibly a new CL,
> 
> Change the nearlys impl to be <= tolerance (so we can pass 0) and adjust the
> assert(tol > 0) accordingly.

Yup, made a note to do that in a separate CL later.

> 
> http://codereview.appspot.com/4582043/diff/11001/tests/core/SkPointTest.cpp
> File tests/core/SkPointTest.cpp (right):
> 
>
http://codereview.appspot.com/4582043/diff/11001/tests/core/SkPointTest.cpp#n...
> tests/core/SkPointTest.cpp:12: point.set(SkIntToScalar(3), SkIntToScalar(4));
> might drop these two asserts
> 
>
http://codereview.appspot.com/4582043/diff/11001/tests/core/SkPointTest.cpp#n...
> tests/core/SkPointTest.cpp:17: scalar = SkPoint::Normalize(&point);
> SkScalar s1 = point.length();
> SkScalar s2 = SkPoint::Normalize(&point);
> ASSERT(s1 == s2);
> 
> nearly(s1, 5);

Renamed SkPointTest.cpp to PointTest.cpp and made changes inspired by the above
feedback.

[reed1, 2011-06-15 17:54:10]

LGTM

[epoger, 2011-06-15 18:05:30]

committed as r1602