The existing code didn't work properly with half types since the $mat
type encompassed both halfNxM and floatNxM. This was fixed by splitting
the half types out of $mat into a separate $matH generic.
Unit tests now compile properly for GLSL, but generate errors in SPIR-V
and generate Metal code which attempts to call a non-existent intrinsic.
Change-Id: I2fff10f0dd7f00534bf6b1d5b13354543694194e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342926
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Just like for fills, we use the normal Wang's formula on the conic's
down-projected control points until we can formalize on a better
formula.
Bug: skia:10419
Change-Id: Ifd735534a2e793f79f4f5d5b7e7acf50db81fe5e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341156
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Reviewed-by: Tyler Denniston <tdenniston@google.com>
Just the class/files. variable names and additional comments to follow.
Change-Id: Ic03d07fd5009eaf3d706c2536486a117328963fc
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342617
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
SkDrawIter goes away, as well as the DRAW_BEGIN, DRAW_END, and
FOR_EACH_TOP_DEVICE macros. A new macro TOP_DEVICE is added that
simplifies checking for a null device (which is currently only possible
if SkCanvas::init() was called with null). The intent is that this will
be replaced with just "this->getTopDevice()->" in the future.
The rest of the auto-layer magic that happened in DRAW_BEGIN is now
explicitly written out in the draw functions. Additionally, it has
been updated to check for an image filter. When there's no image
filter, it completely bypasses the construction of the AutoLayer object.
This extra check maybe isn't worth the code complexity, but some draw
methods had previously followed this pattern, so presumably stack
allocating the AutoLayer had some overhead in tight draw loops.
Bug: skia:10986, skia:10987
Change-Id: I9c48fe4645a6d2b46e1d0006a669fd073c0e1478
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341161
Reviewed-by: Mike Reed <reed@google.com>
Commit-Queue: Michael Ludwig <michaelludwig@google.com>
This reverts commit 53f69f1539.
Change-Id: I374b016c8a08d83c99cbab800f30b882244b87f1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342919
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
Initial data suggests that the non-rational Wang's formula does not
form an upper bound for the rational case, so additional work is needed
to handle them. Note this new formula is not actually due to Wang,
but is an analogue.
The version added in this CL is for cubic rationals because the current
tessellation code promotes all curve segments to cubic rationals for
ease. If we end up using this approach, it would be likely simpler and
faster to implement the degree-2 version and handle conics separately
during tessellation.
From:
J. Zheng, T. Sederberg. "Estimating Tessellation Parameter Intervals for
Rational Curves and Surfaces." ACM Transactions on Graphics 19(1). 2000.
Bug: skia:10419
Change-Id: Ie02e229c089541ece05c7502217b1ef5d4799b52
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/337720
Commit-Queue: Tyler Denniston <tdenniston@google.com>
Reviewed-by: Chris Dalton <csmartdalton@google.com>
This intrinsic multiplies the vertex against the fixed-function
gl_ModelViewProjectionMatrix matrix, which is not a thing in Skia.
Change-Id: Ie361d9d631762b1397ee7be062bda317b415c7c9
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342923
Commit-Queue: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Initially we restricted tessellated stroking to opaque solid colors.
This CL adds support for transparency by enabling a stencil test. The
stencil test also allows us to use mixed samples.
Bug: skia:10419
Change-Id: Ie40f3099d2b009f92ed49f7f43e5f269b1a479af
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/340798
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Expand existing SkottieColorizeGM to also handle text properties and
add text-focused instance.
Update layer names in one of the json assets to match expected demo
prefix ($).
TBR=
Change-Id: I076229067523fe597be66c611a8653897f995bc8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342916
Reviewed-by: Florin Malita <fmalita@google.com>
Commit-Queue: Florin Malita <fmalita@google.com>
Commit-Queue: Florin Malita <fmalita@chromium.org>
playback methods would call on mAnimation before checking if it was null
builder constructor created unneeded confusion, as setSource() currently acts a build method
Change-Id: I90a36e39cab1ed76a12c49fb53b4518bba9bf4ef
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/340521
Reviewed-by: Derek Sollenberger <djsollen@google.com>
Reviewed-by: Florin Malita <fmalita@chromium.org>
Commit-Queue: Jorge Betancourt <jmbetancourt@google.com>
Bug: webp:490
Depending on the encoded image, SkWebpCodec may need to blend the
the output from libwebp with the prior frame. It does so using the
method blend_line, which expects the libwebp output to be unpremul.
Prior to this commit, SkWebpCodec sometimes told libwebp to premultiply,
and then passed that premultiplied data to blend_line, which
premultiplied again.
Use webpInfo's alphaType to decide whether to premultiply. Consolidate
choosing its alphaType into one block. The functional difference is that
if (blendWithPrevFrame), we no longer premul if the dst is kPremul.
Move declaration of webDst to where it's used.
Add a test.
Change-Id: Ic0cfb4d918c2ab434c6787ed5a532c4d1e67fa17
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342618
Reviewed-by: Derek Sollenberger <djsollen@google.com>
Commit-Queue: Derek Sollenberger <djsollen@google.com>
Auto-Submit: Leon Scroggins <scroggo@google.com>
Change-Id: I7af94d89d349b67b2c070179324fcad7b62e0d1e
Bug: skia:11071
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342758
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Change-Id: I1d5a056e08ba6e67016e45c52518da6074a62c8f
Bug: skia:11071
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342759
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
floatBitsToUint was missing from our intrinsic list entirely, and
u?intBitsToFloat were misspelled.
These intrinsics aren't implemented in SPIR-V or Metal either, but that
will be handled in followup CLs.
Change-Id: Iaf9b9d5a2e46e25d41eef71903fad8bd1c177d4e
Bug: skia:11071
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342757
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Flutter is seeing an issue where we're trying to create a
blitCommandEncoder but we're still in the process of encoding. This
doesn't seem to happen with renderCommandEncoders so trying to mirror
what we do there.
Change-Id: I281509116a5666ebd67c19e4d8a21a83076861e8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342621
Reviewed-by: Greg Daniel <egdaniel@google.com>
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Remove more limited tests that exercise the same code.
Reduce the number of test cases to improve run time.
Bug: skia:11019
Change-Id: I68b5b18ebf077e825a260285223b4a31243dcb4d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341456
Commit-Queue: Brian Salomon <bsalomon@google.com>
Reviewed-by: Greg Daniel <egdaniel@google.com>
This reverts commit a3b8ac76e5.
Reason for revert: Need to revert again, red tree.
Original change's description:
> Revert "Revert "Initial land of SkSL DSL.""
>
> This reverts commit dd213e9d46.
>
> Change-Id: I43be020dd1b07dc13862150a9d95493f8c48b3b1
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342622
> Auto-Submit: Ethan Nicholas <ethannicholas@google.com>
> Reviewed-by: John Stiles <johnstiles@google.com>
> Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
TBR=brianosman@google.com,ethannicholas@google.com,johnstiles@google.com
# Not skipping CQ checks because this is a reland.
No-Presubmit: true
No-Try: true
Change-Id: I8e967ef8ecb7f01dc578d38264e2600b04e9b62d
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342917
Reviewed-by: Jorge Betancourt <jmbetancourt@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
Change-Id: I887e700a7bf11bf2d5359c9721798f72f00e53f3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342756
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
This reverts commit dd213e9d46.
Change-Id: I43be020dd1b07dc13862150a9d95493f8c48b3b1
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342622
Auto-Submit: Ethan Nicholas <ethannicholas@google.com>
Reviewed-by: John Stiles <johnstiles@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
This reverts commit ff7bfea4ab.
Reason for revert: several broken bots, e.g.
../SRC/skia/modules/skparagraph/src/TextLine.cpp:309:21: error: non-const lvalue reference to type 'skia::textlayout::TextLine::TextBlobRecord' cannot bind to a temporary of type 'void'
TextBlobRecord& record = fTextBlobCache.emplace_back();
^ ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Original change's description:
> Cache text blobs computed by TextLine::paintText
>
> Change-Id: I7d4414fb64b14a714623e4c20199cab06e6ed18c
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342297
> Commit-Queue: Jason Simmons <jsimmons@google.com>
> Reviewed-by: Julia Lavrova <jlavrova@google.com>
TBR=jsimmons@google.com,jlavrova@google.com
Change-Id: I467c61eead7b0d8c190cb2f883b05537de97a08e
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342856
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
27e52ca05d..067a12f059
2020-12-09 cclao@google.com Reland "Vulkan: Use specialization constant for halfRenderAreaWidth"
2020-12-09 cclao@google.com Vulkan: Add dirty bits instead of direct access FBO state
2020-12-09 syoussefi@chromium.org Suppress test requiring VK_EXT_transform_feedback on Nvidia/win7
2020-12-09 courtneygo@google.com Add VK-GL-CTS autoroller link to wrangler doc
2020-12-09 syoussefi@chromium.org Fix varying packing of I/O block fields
2020-12-09 xinghua.cao@intel.com D3D: throw a perf warning for uniform block
2020-12-09 angle-autoroll@skia-public.iam.gserviceaccount.com Roll Chromium from a3984b7ec4ae to 3fc522ff6579 (498 revisions)
2020-12-09 syoussefi@chromium.org Don't output matrix qualifiers in I/O blocks
2020-12-09 syoussefi@chromium.org Vulkan: Fix query pause on framebuffer binding change
2020-12-09 jiawei.shao@intel.com Enable ComputeShaderTest/DrawDispatchDispatchDraw on Intel/Win/Vulkan
If this roll has caused a breakage, revert this CL and stop the roller
using the controls here:
https://autoroll.skia.org/r/angle-skia-autoroll
Please CC jmbetancourt@google.com on the revert to ensure that a human
is aware of the problem.
To report a problem with the AutoRoller itself, please file a bug:
https://bugs.chromium.org/p/skia/issues/entry?template=Autoroller+Bug
Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+doc/master/autoroll/README.md
Cq-Include-Trybots: skia/skia.primary:Build-Debian10-Clang-x86_64-Release-ANGLE;skia/skia.primary:Test-Win10-Clang-AlphaR2-GPU-RadeonR9M470X-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-Golo-GPU-QuadroP400-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC5i7RYH-GPU-IntelIris6100-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC6i5SYK-GPU-IntelIris540-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUC8i5BEK-GPU-IntelIris655-x86_64-Debug-All-ANGLE;skia/skia.primary:Test-Win10-Clang-NUCD34010WYKH-GPU-IntelHD4400-x86_64-Debug-All-ANGLE
Tbr: jmbetancourt@google.com
Change-Id: I1c230955ddd16fdd56976012553d32f14069fbc8
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342748
Reviewed-by: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
Commit-Queue: skia-autoroll <skia-autoroll@skia-public.iam.gserviceaccount.com>
Adds a vertex shader that maps a variable-length triangle strip to a
stroke and its preceding join. Adds a new op that generates stroke
instances from a path, bins them by log2 triangle strip length (using
SIMD for the calculations), and renders them with indirect draws.
Bug: skia:10419
Change-Id: I6d52df02cffe97d14827c6d66136957f1859f53b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/339716
Commit-Queue: Chris Dalton <csmartdalton@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
Follow-on to https://skia-review.googlesource.com/c/skia/+/342616
Assuming this survives, we can probably simplify other code in SkPath
(e.g. Iterators) that today worry about consing up moves after close.
Change-Id: I913d6be9c20547267230c287d394fd300b985d31
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342619
Commit-Queue: Mike Reed <reed@google.com>
Reviewed-by: Florin Malita <fmalita@chromium.org>
Change-Id: I674d758c11071582e9fbedcda5596c540bfb5f71
Bug: skia:11054
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342558
Commit-Queue: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
This does not give us 100% coverage of intrinsics yet, but it is a
pretty good start.
Change-Id: I97d49324db1afd9f2975c2eeafbacdead710d4aa
Bug: skia:11054
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341977
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
We're passing in QuadEdge attributes as half precision, but
kStableLargeNegativeValue is well past any value that can be handled
by half float. Increasing the precision used fixes the issues we're
seeing on iOS 14.
Bug: skia:10594
Change-Id: I650a0930cf8dd9a2369bfeeffc132c0f7f0fa79e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342303
Commit-Queue: Jim Van Verth <jvanverth@google.com>
Reviewed-by: Brian Salomon <bsalomon@google.com>
These are somewhat the first presentation attributes of their kind,
in that they are non-inherited but also not applied via canvas layers.
Implementation-wise the main difference is that these attributes are
not propagated through the fInherited field of the render context's
presentation attribute list.
Change-Id: I0909507b0ecbd21732b3f80c46a343f5a0a9bf7a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/340661
Reviewed-by: Florin Malita <fmalita@chromium.org>
Commit-Queue: Tyler Denniston <tdenniston@google.com>
This color type will be shared in paint, stop-color, and flood-color
attributes (mainly so they can use 'currentColor').
Change-Id: Ib4200ea729a91a0db5da069c68d64e5e8e3f5010
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/340617
Reviewed-by: Florin Malita <fmalita@chromium.org>
Commit-Queue: Tyler Denniston <tdenniston@google.com>
We now insert helper functions which defer the assignment of out-
parameters back into their original variables to the end of the
function call. This allows us to match the semantics listed the GLSL
spec in section 6.1.1:
"All arguments are evaluated at call time, exactly once, in order, from
left to right. [...] Evaluation of an out parameter results in an
l-value that is used to copy out a value when the function returns.
Evaluation of an inout parameter results in both a value and an l-value;
the value is copied to the formal parameter at call time and the lvalue
is used to copy out a value when the function returns."
This technique also allows us to support swizzled out-parameters in
Metal, by reading the swizzle into a temp variable, calling the original
function, and then re-assigning the result back into the original
swizzle expression.
At present, we don't deduplicate these helper functions, so in theory
there could be a fair amount of redundant code generated if a function
with out parameters is called many times in a row. The cost of properly
deduplicating them is probably larger than the benefit in the 99% case.
Change-Id: Iefc922ac9e2b24ef2ff1e9dacb17a735a75ec8ea
Bug: skia:10855, skia:11052
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341162
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
This is a reland of eca04eeeb8
Problem was that any error during image flattening left the json encoder
in an invalid state. this closes the object before all early returns.
Original change's description:
> Always attempt to flatten images with mskp image ids
>
> b:skia:11041
>
> Change-Id: I9eeac07b5cad7809d1e8ec9f9b12149f016227e0
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341798
> Reviewed-by: Leandro Lovisolo <lovisolo@google.com>
> Commit-Queue: Nathaniel Nifong <nifong@google.com>
Change-Id: I622dbaa3ce5db85c9f790f822d51be247af528bf
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342463
Reviewed-by: Leandro Lovisolo <lovisolo@google.com>
Commit-Queue: Nathaniel Nifong <nifong@google.com>
Change-Id: Id2135834062fed316d53e1419389a664e8d39c64
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342677
Reviewed-by: Mike Klein <mtklein@google.com>
Commit-Queue: Mike Klein <mtklein@google.com>
This reverts commit 6e599511d4.
Reason for revert: Breaking bots: https://logs.chromium.org/logs/skia/5061fbd134144011/+/steps/dm/0/stdout
Original change's description:
> Initial land of SkSL DSL.
>
> This is not 100% complete: it lacks support for several kinds of nodes
> and supports only a bare handful of builtin functions, but it
> demonstrates the core functionality and it should be relatively
> straightforward to fill in the missing pieces.
>
> Change-Id: I3058089338e20eebc3da18ac5571801abcaab564
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/331177
> Reviewed-by: John Stiles <johnstiles@google.com>
> Reviewed-by: Brian Osman <brianosman@google.com>
> Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
TBR=brianosman@google.com,ethannicholas@google.com,johnstiles@google.com
Change-Id: Iee77e5322a0b1efb0f3718ec1f5976a4d4e7323a
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342620
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
This is not 100% complete: it lacks support for several kinds of nodes
and supports only a bare handful of builtin functions, but it
demonstrates the core functionality and it should be relatively
straightforward to fill in the missing pieces.
Change-Id: I3058089338e20eebc3da18ac5571801abcaab564
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/331177
Reviewed-by: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Commit-Queue: Ethan Nicholas <ethannicholas@google.com>
In crbug.com/1155871, the audit trail is used during destruction of the
drawing manager.
Bug: chromium:1155871
Change-Id: I74015a033c7620bdad03ac7a3cf5a7ee671482f4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342305
Reviewed-by: Brian Salomon <bsalomon@google.com>
Commit-Queue: Adlai Holler <adlai@google.com>
This reverts commit 540b51ce8f.
Reason for revert: RPI disk space
Original change's description:
> Reland "[infra] Switch from isolate to RBE-CAS"
>
> Bug: skia:10883
> Change-Id: Iec82a07fdf3c0807a9bb1870309eded85d4f0b1e
> Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341679
> Commit-Queue: Eric Boren <borenet@google.com>
> Reviewed-by: Weston Tracey <westont@google.com>
TBR=borenet@google.com,westont@google.com
Change-Id: I0156842b995d275d8deb5bd73c6877de7f8af72e
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Bug: skia:10883
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342596
Reviewed-by: Eric Boren <borenet@google.com>
Commit-Queue: Eric Boren <borenet@google.com>
Change-Id: Iac8096f6c225258b430858bad90199ec00b93b6c
Bug: skia:10998
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/342304
Commit-Queue: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Prefer the simplified form half4(N) instead of half4(N, N, N, N) where
we can. Slightly easier to humans to read and for computers to parse.
Change-Id: Ibfe6d1c5ef6ebddc157ff39c7bdeddeb27131689
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/337722
Commit-Queue: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Previously, `equal(a, b).x` would emit `a == b.x`.
Change-Id: I311116f45fb275b158a948a0fd165d97688ce8ff
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/341978
Commit-Queue: John Stiles <johnstiles@google.com>
Auto-Submit: John Stiles <johnstiles@google.com>
Reviewed-by: Brian Osman <brianosman@google.com>
Reviewed-by: Ethan Nicholas <ethannicholas@google.com>
