add getTypes() to SkMatrix44, to cache how complex the matrix is.

add getTypes() to SkMatrix44, to cache how complex the matrix is.
add bench

optimize operator== by performing 4 compares in a row before checking
optimize setconcat by noting when we can write the answer directly into this

At least on this macbook, I had to mark helpers like isIdentity() as inline to get them inlined.
Committed: https://code.google.com/p/skia/source/detail?r=6655

[bsalomon, 2012-12-03 19:33:30]

Really weird that you needed to explicitly say inline. Maybe that warrants some
more investigation. Maybe we need a macro with platform-specific implementation
to be used here and in other classes.

https://codereview.appspot.com/6863053/diff/1/bench/Matrix44Bench.cpp
File bench/Matrix44Bench.cpp (right):

https://codereview.appspot.com/6863053/diff/1/bench/Matrix44Bench.cpp#newcode66
bench/Matrix44Bench.cpp:66: //            always_do(m0 == m1);
not needed?

https://codereview.appspot.com/6863053/diff/1/include/utils/SkMatrix44.h
File include/utils/SkMatrix44.h (right):

https://codereview.appspot.com/6863053/diff/1/include/utils/SkMatrix44.h#newc...
include/utils/SkMatrix44.h:106: SkMatrix44() { this->setIdentity(); }
This is different than SkMatrix.

https://codereview.appspot.com/6863053/diff/1/include/utils/SkMatrix44.h#newc...
include/utils/SkMatrix44.h:157: void setIdentity();
Should we add setIdentity() to SkMatrix for parity? Why have both reset and
setIdentity()?

https://codereview.appspot.com/6863053/diff/1/src/utils/SkMatrix44.cpp
File src/utils/SkMatrix44.cpp (right):

https://codereview.appspot.com/6863053/diff/1/src/utils/SkMatrix44.cpp#newcode22
src/utils/SkMatrix44.cpp:22: const SkMScalar* SK_RESTRICT b) {
space

https://codereview.appspot.com/6863053/diff/1/src/utils/SkMatrix44.cpp#newcode74
src/utils/SkMatrix44.cpp:74: if (1 != scaleX() || 1 != scaleY() || 1 !=
scaleX()) {
last one should be scaleZ()

"this->"?

[reed1, 2012-12-03 20:39:54]

https://codereview.appspot.com/6863053/diff/1/bench/Matrix44Bench.cpp
File bench/Matrix44Bench.cpp (right):

https://codereview.appspot.com/6863053/diff/1/bench/Matrix44Bench.cpp#newcode66
bench/Matrix44Bench.cpp:66: //            always_do(m0 == m1);
On 2012/12/03 19:33:30, bsalomon wrote:
> not needed?

Done.

https://codereview.appspot.com/6863053/diff/1/include/utils/SkMatrix44.h
File include/utils/SkMatrix44.h (right):

https://codereview.appspot.com/6863053/diff/1/include/utils/SkMatrix44.h#newc...
include/utils/SkMatrix44.h:106: SkMatrix44() { this->setIdentity(); }
On 2012/12/03 19:33:30, bsalomon wrote:
> This is different than SkMatrix. 

Yes, though this is not new with this CL. I just inlined what it was doing.

I think a separate thread should be started to get rid of this...

https://codereview.appspot.com/6863053/diff/1/include/utils/SkMatrix44.h#newc...
include/utils/SkMatrix44.h:157: void setIdentity();
On 2012/12/03 19:33:30, bsalomon wrote:
> Should we add setIdentity() to SkMatrix for parity? Why have both reset and
> setIdentity()?

I didn't add that either. But sure, we can add it to SkMatrix.h as well. I like
reset() since it follows lots of other skia classes, but having a
domain-specific one seems ok too.

https://codereview.appspot.com/6863053/diff/1/src/utils/SkMatrix44.cpp
File src/utils/SkMatrix44.cpp (right):

https://codereview.appspot.com/6863053/diff/1/src/utils/SkMatrix44.cpp#newcode74
src/utils/SkMatrix44.cpp:74: if (1 != scaleX() || 1 != scaleY() || 1 !=
scaleX()) {
On 2012/12/03 19:33:30, bsalomon wrote:
> last one should be scaleZ()

Oops, good catch.

> 
> "this->"?

Yea, I was trying to not line-break

[bsalomon, 2012-12-03 21:02:57]

On 2012/12/03 20:39:54, reed1 wrote:
> https://codereview.appspot.com/6863053/diff/1/bench/Matrix44Bench.cpp
> File bench/Matrix44Bench.cpp (right):
> 
>
https://codereview.appspot.com/6863053/diff/1/bench/Matrix44Bench.cpp#newcode66
> bench/Matrix44Bench.cpp:66: //            always_do(m0 == m1);
> On 2012/12/03 19:33:30, bsalomon wrote:
> > not needed?
> 
> Done.
> 
> https://codereview.appspot.com/6863053/diff/1/include/utils/SkMatrix44.h
> File include/utils/SkMatrix44.h (right):
> 
>
https://codereview.appspot.com/6863053/diff/1/include/utils/SkMatrix44.h#newc...
> include/utils/SkMatrix44.h:106: SkMatrix44() { this->setIdentity(); }
> On 2012/12/03 19:33:30, bsalomon wrote:
> > This is different than SkMatrix. 
> 
> Yes, though this is not new with this CL. I just inlined what it was doing.
> 
> I think a separate thread should be started to get rid of this...
> 
>
https://codereview.appspot.com/6863053/diff/1/include/utils/SkMatrix44.h#newc...
> include/utils/SkMatrix44.h:157: void setIdentity();
> On 2012/12/03 19:33:30, bsalomon wrote:
> > Should we add setIdentity() to SkMatrix for parity? Why have both reset and
> > setIdentity()?
> 
> I didn't add that either. But sure, we can add it to SkMatrix.h as well. I
like
> reset() since it follows lots of other skia classes, but having a
> domain-specific one seems ok too.
> 
> https://codereview.appspot.com/6863053/diff/1/src/utils/SkMatrix44.cpp
> File src/utils/SkMatrix44.cpp (right):
> 
>
https://codereview.appspot.com/6863053/diff/1/src/utils/SkMatrix44.cpp#newcode74
> src/utils/SkMatrix44.cpp:74: if (1 != scaleX() || 1 != scaleY() || 1 !=
> scaleX()) {
> On 2012/12/03 19:33:30, bsalomon wrote:
> > last one should be scaleZ()
> 
> Oops, good catch.
> 
> > 
> > "this->"?
> 
> Yea, I was trying to not line-break

lgtm