Use a smarter strategy to speed up text run shaping #229

burg · 2012-11-21T17:25:38Z

Gecko, for example, caches shaping results on a per-word basis and then copies the words into a glyph store when shaping is requested. The cache is stored with gfx::Font instances (i.e., tied to a particular size).

This caching reduces the size of inputs and outputs to harfbuzz, and reduces the number of shaping calls. In gecko this mainly is useful for scaling shaping with the number of unique words; in Servo it would be even more useful while we are still doing non-incremental layout (shaping entire text run over and over).

mbrubeck · 2016-05-24T21:49:40Z

Servo now caches shaping results per word.


        Auto merge of servo#229 - glennw:fix-black-anim, r=pcwalton

Fix render target cache. Ensure that the render target cache doesn't return a render target that was allocated on this frame. Fixes servo#184.

mbrubeck closed this May 24, 2016

glennw pushed a commit to glennw/servo that referenced this issue Jan 16, 2017

Auto merge of servo#229 - glennw:fix-black-anim, r=pcwalton

cec868a

Fix render target cache. Ensure that the render target cache doesn't return a render target that was allocated on this frame. Fixes servo#184.

jdm mentioned this issue Nov 5, 2019

Crash loading CSS painting #24658

Closed

servo / servo

Use a smarter strategy to speed up text run shaping #229

Use a smarter strategy to speed up text run shaping #229

burg commented Nov 21, 2012

mbrubeck commented May 24, 2016

servo / servo

Join GitHub today

Use a smarter strategy to speed up text run shaping #229

Use a smarter strategy to speed up text run shaping #229

Comments

burg commented Nov 21, 2012

mbrubeck commented May 24, 2016