Always use border texels on layer tiles. We already did that on certain platforms. This patch exten…

Always use border texels on layer tiles. We already did that on certain platforms. This patch extends it to all platforms.

BUG=160072
Committed: http://src.chromium.org/viewvc/chrome?view=rev&revision=167870

[danakj, 2012-11-12 21:29:32]

lgtm if it lgtenne

[enne (OOO), 2012-11-12 21:39:42]

The patch lgtm, with the caveat that you check linux_layout_rel results before
CQing.

[alokp, 2012-11-12 21:57:11]

compositing/checkerboard.html failed but I think it just needs to be
rebaselined.

[enne (OOO), 2012-11-12 22:02:16]

On 2012/11/12 21:57:11, Alok Priyadarshi wrote:
> compositing/checkerboard.html failed but I think it just needs to be
> rebaselined.

How did it fail?

[alokp, 2012-11-12 22:20:32]

The images look off by one pixel. If we draw tiles with and without borders
differently it might have caused the difference. I will look into it in more
detail before landing it.

[enne (OOO), 2012-11-12 22:25:54]

Checkerboard shouldn't be off by any pixels.  It's an image on the root layer
with an integer translation.  We should be able to display that pixel for pixel.

I think this has come up before when we tried to change the NCCH from nearest to
linear and floating point inaccuracies caused this to not look right.

[jamesr, 2012-11-12 23:42:25]

http://codereview.chromium.org/11360205/diff/1/cc/tiled_layer.cc
File cc/tiled_layer.cc (left):

http://codereview.chromium.org/11360205/diff/1/cc/tiled_layer.cc#oldcode286
cc/tiled_layer.cc:286: setBorderTexelOption(borderTexelOption);
If you're ripping this out then a lot of other code in here, layer_tiling_data,
tiled_layer_impl and tile_draw_quad are dead code as well.  Please remove those
too.

[alokp, 2012-11-14 16:58:25]

jamesr: I tried getting rid of dead code, but many unit-tests are written based
on NoBorderTexels. I was updating the tests when I found a bug in
cc::TilingData::ComputeNumTiles. The number of tiles returned is more than
required in a few cases. So this is turning out to be trickier than just
deleting a bunch of dead code.

I will prefer to delete dead code and fix unit tests in a separate patch.

[commit-bot: I haz the power, 2012-11-14 17:11:12]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/11360205/1

[commit-bot: I haz the power, 2012-11-14 18:33:37]

Retried try job too often for step(s) content_browsertests

[commit-bot: I haz the power, 2012-11-14 18:51:59]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/11360205/1

[commit-bot: I haz the power, 2012-11-15 00:29:19]

Retried try job too often for step(s) browser_tests

[commit-bot: I haz the power, 2012-11-15 03:46:53]

CQ is trying da patch. Follow status at
https://chromium-status.appspot.com/cq/[email protected]/11360205/1

[commit-bot: I haz the power, 2012-11-15 05:37:46]

Change committed as 167870