Move Image operations and convolver to the skia namespace and clean up a few ...

Move Image operations and convolver to the skia namespace and clean up a few (but not all) base types.

Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=6921

[brettw, 2008-12-10 23:28:42]

Third try for this patch. This is more minimal than before, trying to avoid
weird changes in behavior I was seeing before. I ran the background layout tests
which the last patch had problems with and they're all fine. Also, I didn't
break mac this time.

[M-A Ruel, 2008-12-11 00:39:02]

Mostly nits so you can check-in with the items fixed.

http://codereview.chromium.org/13726/diff/1/11
File skia/ext/convolver.cc (right):

http://codereview.chromium.org/13726/diff/1/11#newcode16
Line 16: if (static_cast<unsigned>(a) < 256)
// Clamp for < 1
i = ~(((unsigned int)a >> (sizeof(int)*8-1)) * -1) & a
maybe it's slower after all. I'm pretty sure there's a way to write this
function without a condition. Since it must be called often in loops, it's worth
optimizing.

Ok I digress. Ignore me.

http://codereview.chromium.org/13726/diff/1/11#newcode233
Line 233: 
You don't want to use SkASSERT?

http://codereview.chromium.org/13726/diff/1/12
File skia/ext/convolver.h (right):

http://codereview.chromium.org/13726/diff/1/12#newcode38
Line 38: static inline Fixed FloatToFixed(float f) {
Remove inline. I hate redundancy.

http://codereview.chromium.org/13726/diff/1/9
File skia/ext/image_operations.cc (right):

http://codereview.chromium.org/13726/diff/1/9#newcode20
Line 20: using namespace gfx;
Can you put the using namespace gfx _inside_ the skia namespace?

http://codereview.chromium.org/13726/diff/1/9#newcode364
Line 364: }  // namespace gfx
skia

http://codereview.chromium.org/13726/diff/1/10
File skia/ext/image_operations.h (right):

http://codereview.chromium.org/13726/diff/1/10#newcode60
Line 60: }  // namespace gfx
skia

http://codereview.chromium.org/13726/diff/1/15
File webkit/port/platform/graphics/TransparencyHelperWin.cpp (right):

http://codereview.chromium.org/13726/diff/1/15#newcode48
Line 48: inline void IntRectToRECT(const IntRect& ir, RECT* r)
Does these functions need to be inline?

http://codereview.chromium.org/13726/diff/1/15#newcode102
Line 102: //                             -(-subset.y() + devices[i].y));
you can remove this

http://codereview.chromium.org/13726/diff/1/15#newcode186
Line 186: // is fine fur us.
for

http://codereview.chromium.org/13726/diff/1/15#newcode305
Line 305: row[x] |= 0xFF000000;
Endianness fun.

http://codereview.chromium.org/13726/diff/1/17
File webkit/port/platform/graphics/TransparencyHelperWin.h (right):

http://codereview.chromium.org/13726/diff/1/17#newcode54
Line 54: // it. This result must be opaque. When complete, the requly will be
requly? opqaue? (3 times) mkes? trmporary?

http://codereview.chromium.org/13726/diff/1/17#newcode84
Line 84: // is unknown and we can't do the text trick (since we know its color),
since we don't know its color?

http://codereview.chromium.org/13726/diff/1/17#newcode91
Line 91: // There are no changes to the transform. Use this when drawing
There is no change?

http://codereview.chromium.org/13726/diff/1/17#newcode102
Line 102: // the coordinate space, and
and I rock so much, you know.

http://codereview.chromium.org/13726/diff/1/17#newcode106
Line 106: // makeRectilinear indicates that the new layer should "undo" the
transform
makeRectilinear?

[brettw, 2008-12-12 05:45:18]

I removed TransparencyHelper (it was in there by mistake) from this CL and
addressed the rest of your issues.