Optimize glyph positioning for SVG <textPath> layout

Optimize glyph positioning for SVG <textPath> layout

Introduce a new class Path::PositionCalculator, that keeps traversal
state related to path position/normal calculations, and allows
exploiting locality for such queries.

Using this new class for the layout of glyphs on a <textPath>. This
means it possible to achieve O(n) runtimes in many cases. The worst
case remains the same, O(n^2), as when using state-less queries.

Committed: https://src.chromium.org/viewvc/blink?view=rev&revision=166813

[Stephen Chennney, 2014-02-07 16:16:05]

Nice. The inefficiency of SVG text layout actually came up in a discussion with
the Skia folks yesterday.

Just one question about pointer ownership ...

https://codereview.chromium.org/140053006/diff/40001/Source/core/rendering/sv...
File Source/core/rendering/svg/SVGTextLayoutEngine.h (right):

https://codereview.chromium.org/140053006/diff/40001/Source/core/rendering/sv...
Source/core/rendering/svg/SVGTextLayoutEngine.h:99:
OwnPtr<Path::PositionCalculator> m_textPathCalculator;
The lifetime of this appears to be within the
beginTextPathLayout/endTextPathLayout, whereas OwnPtr implies something that
should be automatically deleted on destruction of the owner. I presume that the
SVGTextLayoutEngine will not be destructed within the begin/end. So this should
be a raw pointer.

[fs, 2014-02-07 17:08:58]

On 2014/02/07 16:16:05, Stephen Chenney wrote:
> Nice. The inefficiency of SVG text layout actually came up in a discussion
with
> the Skia folks yesterday.

In the TC I looked at for this, SkGlyphCache::VisitCache ended up quite hot
(~60% of samples) - I guess that's not what you talked about though, since I
guess that's only easily produced with (non-trivial) <textPath>s. (And the
<textPath> was actually only the second highest before this patch...)

[fs, 2014-02-07 17:09:19]

https://codereview.chromium.org/140053006/diff/40001/Source/core/rendering/sv...
File Source/core/rendering/svg/SVGTextLayoutEngine.h (right):

https://codereview.chromium.org/140053006/diff/40001/Source/core/rendering/sv...
Source/core/rendering/svg/SVGTextLayoutEngine.h:99:
OwnPtr<Path::PositionCalculator> m_textPathCalculator;
On 2014/02/07 16:16:05, Stephen Chenney wrote:
> The lifetime of this appears to be within the
> beginTextPathLayout/endTextPathLayout, whereas OwnPtr implies something that
> should be automatically deleted on destruction of the owner. I presume that
the
> SVGTextLayoutEngine will not be destructed within the begin/end. So this
should
> be a raw pointer.

Right. Changed.

[fs, 2014-02-07 17:12:00]

https://codereview.chromium.org/140053006/diff/120001/Source/core/rendering/s...
File Source/core/rendering/svg/SVGTextLayoutEngine.cpp (right):

https://codereview.chromium.org/140053006/diff/120001/Source/core/rendering/s...
Source/core/rendering/svg/SVGTextLayoutEngine.cpp:230: m_textPathCalculator = 0;
Crap, it's too late here...

[fs, 2014-02-07 17:12:39]

https://codereview.chromium.org/140053006/diff/120001/Source/core/rendering/s...
File Source/core/rendering/svg/SVGTextLayoutEngine.cpp (right):

https://codereview.chromium.org/140053006/diff/120001/Source/core/rendering/s...
Source/core/rendering/svg/SVGTextLayoutEngine.cpp:230: m_textPathCalculator = 0;
On 2014/02/07 17:12:01, fs wrote:
> Crap, it's too late here...

Fixed...

[Stephen Chennney, 2014-02-07 18:20:15]

Get some rest. :-)

LGTM

[fs, 2014-02-10 08:30:51]

The CQ bit was checked by fs@opera.com

[commit-bot: I haz the power, 2014-02-10 08:31:07]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/fs@opera.com/140053006/180001

[commit-bot: I haz the power, 2014-02-10 09:30:59]

Change committed as 166813