The issue was Skia would be in the middle of drawing when it
needed to lazy-load the image's texture. This would mess up the
bound texture in WebGL, but Skia didn't know that. As a result,
it would not realize that some of the work had been undone.
Calling resetContext() makes Skia aware of this fact, so it can
adjust its behavior.
Change-Id: I5e62987546be8d63d13c906dfab3bc92cf3120cd
Bug: skia:12740
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/485016
Reviewed-by: Brian Osman <brianosman@google.com>
Also clean up some more findMarkedCTM references.
Change-Id: I29a52e1157691c86992e70a3774b984e5383e3ee
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/482005
Reviewed-by: Brian Osman <brianosman@google.com>
Reviewed-by: Yegor Jbanov <yjbanov@google.com>
This entire API existed, but was unused (no longer connected to
drawVertices or runtime effects).
In theory, we could further simplify some of the matrix providers, but
more importantly - I have serious doubts about the correctness of
localToDeviceHitsPixelsCenters for most of them.
Change-Id: If5af182015dd96e5ed3353a117223e8dbbe17097
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/481683
Reviewed-by: Michael Ludwig <michaelludwig@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
To make the atomic rules a bit easier to work with, in many
of the folders, this adds in cc_library rules to group
together the sources from that folder (and subfolders
where prudent). We only needs sources because those atoms
should have their headers as deps.
One issue that was pointed out is that there is currently
no way to restrict the inclusion of certain packages,
a la, `gn check`. For example, there is no mechanism from
stopping a dev from adding
#include "modules/canvaskit/WasmCommon.h"
to something in //src/core (except circular dependencies).
We can probably address that using Bazel's visibility
rules as needed:
https://docs.bazel.build/versions/main/visibility.htmlhttps://docs.bazel.build/versions/main/be/functions.html#package_group
It is recommended to look at this CL patchset by patchset.
PS1: Update gazelle command to generate rules in more folders.
PS2: A few changes to make generation work better.
PS3: The result of running make generate in //bazel
PS4: Adding the rules to build sksllex, the simplest binary I
could find in the Skia repo.
PS5: Adding the rules to build skdiff, a more complex binary.
I tried a few approaches, but ended up gravitating back
towards the layout where we have each folder/package
group up the sources. I imagine at some point, we'll have
skdiff depend on skia_core or something, which will
have things like //src/core, //src/codecs, //src/pathops
all bundled together.
PS7: Added in the groupings of sources, similar to what we had
earlier. I liked these for readability. These helped fix
up the //:skia_core build, and by extension, the CanvasKit
build.
Change-Id: I3faa7c4e821c876b243617aacf0246efa524cbde
Bug: skia:12541
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/476219
Reviewed-by: Ben Wagner <bungeman@google.com>
Reviewed-by: Leandro Lovisolo <lovisolo@google.com>
This re-works src/ports/BUILD.bazel to work like our other
BUILD files, i.e. one rule "srcs" that brings in the necessary
private filegroups.
To work around an abort with LLVM [1], we have to go back to an
earlier version of emscripten (temporarily?).
Future work should look at using transitions [2] to allow various
executables (e.g. CanvasKit, DM) to set their own set of Bazel
flags, w/o the build invokers having to specify them.
These transitions might be able to handle more complex cases
that we currently use if statements in GN to deal with.
The Freetype build rule was created by taking the BUILD.gn
rule, adding in all the sources listed there and then playing
compile-whack-a-mole to add in all the headers and included
.c files.
Suggested Review Order:
- third_party/BUILD.bazel to see freetype build rules
- bazel/common_config_settings/ to see treatment of fontmgr
like codecs (many possible) and fontmgr_factory (only one).
- src/ports/BUILD.bazel
- BUILD.bazel
- modules/canvaskit/BUILD.bazel. Take note of the gen_rule that
calls tools/embed_resources.py to produce the .cpp file
containing the embedded font data.
- Everything else.
[1] https://github.com/emscripten-core/emscripten/issues/15528
[2] https://github.com/bazelbuild/examples/tree/main/rules/starlark_configurations/cc_binary_selectable_copts
Bug: skia:12541
Change-Id: I08dab82a901d80507007b354ca20cbfad2c2388f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/471636
Reviewed-by: Ben Wagner <bungeman@google.com>
Reviewed-by: Leandro Lovisolo <lovisolo@google.com>
Many things are not enabled currently (e.g. Skottie, Paragraph),
but we can render many APIs using WebGL.
To turn on Paragraph, etc, we'll need to tackle fonts, which
is a separate effort.
This also changes where the build artifacts go. ./build/ is
easier to deal with than the old way of sticking them in
./npm_build/bin
Change-Id: Ia377360af580a887d03630670438fea2e3157e90
Bug: skia:12541
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/470682
Reviewed-by: Leandro Lovisolo <lovisolo@google.com>
Reviewed-by: Ben Wagner <bungeman@google.com>
Change-Id: Ibeea56ebd5dce53af1252ca3ecf6cc6f010bd461
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/469902
Reviewed-by: John Stiles <johnstiles@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
We'll switch to the correct context when necessary (e.g. before
calls that talk to the GPU). This is achieved by adding in
calls at the JS layer to switch the context before making a call
that is known to talk to the GPU (e.g. draw calls on SkCanvas).
Another implementation that was considered was to add a C++
shim in GrGLInterface that would switch the context before
every call in the GPU - however, that seemed too difficult
and would add extra overhead if a single draw* call talks
to the GPU multiple times.
Bug: skia:12255
Change-Id: I96e4c6b41a5bfcc9913aeaca7ccb125358048ad3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/432136
Reviewed-by: Brian Salomon <bsalomon@google.com>
(it had already been abandoned by Skia)
MakeMatrixTransform now takes FilterOptions | CubicResampler
(just like drawAtlas) so that it can have a single entry-point
that picks either variant of SamplingOptions.
Proposal : migrate this pattern (one name, 2 parameter types)
to methods like drawImage, drawImageRect, etc.
Change-Id: I309fbd8be94e642a57ab3016dad7252537ded7d3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/429496
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Commit-Queue: Mike Reed <reed@google.com>
Unused by flutter
Would need variants for sampling options
Easy for clients to achieve this with other APIs
Change-Id: Ia8b29f5183f5b729b62ef35e87897d5312b38da5
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/427197
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Commit-Queue: Mike Reed <reed@google.com>
Callers should use asBlendMode or getBlendMode_or
Bug: skia:12173
Change-Id: I8b62527a2ae11c9bf0c1473c4fffa53b38aa7017
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/425756
Commit-Queue: Mike Reed <reed@google.com>
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Reviewed-by: Yegor Jbanov <yjbanov@google.com>
This will replace needing to call FontMgr.RefDefault()
just to call MakeTypefaceFromData().
Change-Id: I72a8c3be62267f6c54c48a5309ef9351acb16768
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/421916
Reviewed-by: Ben Wagner <bungeman@google.com>
Reviewed-by: Mike Reed <reed@google.com>
It's entirely unused, and trivial for clients to create with SkSL.
Change-Id: I197986232d3706f5af3a197f0fb8e744e1009e5f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/419796
Commit-Queue: Brian Osman <brianosman@google.com>
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Reviewed-by: Mike Reed <reed@google.com>
This includes a helper Surface.makeFromTextureSource that will
make the gl calls with some sensible defaults.
Change-Id: I31bb2f3118b4749e77fb1fc2cb013d35ccb7cfce
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/413336
Reviewed-by: Brian Osman <brianosman@google.com>
This approach is 5 to 20x faster than the existing version,
mostly by minimizing the number of back-and-forths, especially
in the vanilla JS Array case.
Change-Id: Icb4212f781b80ef743f3deb1c17ea6af80ea828b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/414517
Reviewed-by: Mike Reed <reed@google.com>
Grants read-access to a JS array, either ...
- direct peek into the (malloc'd buffer)
- quick copy (the src was a typed_array)
- slow copy (had to perform a per-element lookup)
Change-Id: Id389f244039d35aefd84e0d95d598cbe15cd28c4
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/409397
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Reviewed-by: Herb Derby <herb@google.com>
Also some includes that I noticed along the way.
Change-Id: If48a3a9ff8bd5daea21065051c811e8b3e8a701f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/402136
Reviewed-by: Kevin Lubick <kjlubick@google.com>
1. Fold "origin" into the positions (simplification)
2. Extend positions and offset arrays by 1, to include 1-past last glyph
3. Add flags field to run (with just WS flag for now)
Change-Id: I93e38df808a5e9e4e5bcedbc62bc5a7aa433ef2e
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/399876
Commit-Queue: Mike Reed <reed@google.com>
Reviewed-by: Julia Lavrova <jlavrova@google.com>
Reviewed-by: Yegor Jbanov <yjbanov@google.com>
The old factory is deprecated. The new factory does stricter checking on
the signature of main and calls to sample - verifying that the SkSL is
valid as an SkShader (vs SkColorFilter) at effect creation time.
Bug: skia:11813
Change-Id: I8d70f81b5c3fa78b05add1b591caf1d26dd70402
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/399077
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Font was missing getMetrics()
However, seems like getShapedRuns() needs to return its
particular line spacing choices...
Change-Id: I574ebf789fb03b79c8e09198f1cb6e09dac3441b
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/397916
Commit-Queue: Mike Reed <reed@google.com>
Reviewed-by: Julia Lavrova <jlavrova@google.com>
These were added to SkRuntimeEffects, but we need to do a bit of fixup
in the CK bindings. Note that 'getUniformFloatCount' is now poorly
named, but the idea still stands: it's how many total scalar values are
required.
Bug: skia:11803
Change-Id: If464156d8e6240736e324ef833e57ba7d53f55a0
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/394476
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Reviewed-by: Joe Gregorio <jcgregorio@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
Have the canvaskit bindings provide SkData since only the bindings know
the correct release procedure for the underlying data.
Bug: skia:11778
Change-Id: I0fac15287b014ea01dfc0d4b87dd2fe7987fe1d3
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/388496
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Commit-Queue: Ben Wagner <bungeman@google.com>
Also rename //modules/canvaskit/canvaskit to //modules/canvaskit/npm_build
to make it more clear the purpose of that folder (what we ship to
npm and stage our builds for local testing).
Bug: skia:11203
Change-Id: I4299ded97d14f4155c36798d60e88a660ce6fe6a
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/372392
Reviewed-by: Kevin Lubick <kjlubick@google.com>
This had been previously deprecated.
Bug: skia:10717
Change-Id: Ic57ed835c13cfa7812099a3ef20ed7ff5aa62f7f
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/371339
Reviewed-by: Nathaniel Nifong <nifong@google.com>
The Paragraph API is what should be used.
Bug: skia:10717
Change-Id: I135aff09bffae0718045b5c744f8e774e2ee1bce
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/371338
Reviewed-by: Ben Wagner <bungeman@google.com>
Reviewed-by: Nathaniel Nifong <nifong@google.com>
Written to use the exact same names and semantics as particles.
Cq-Include-Trybots: luci.skia.skia.primary:Test-Debian10-EMCC-GCE-CPU-AVX2-wasm-Release-All-CanvasKit,Test-Debian10-EMCC-GCE-GPU-AVX2-wasm-Release-All-CanvasKit
Change-Id: I4031efbce06527a519f1ce8c261f79231554e1ef
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/365701
Reviewed-by: Joe Gregorio <jcgregorio@google.com>
Reviewed-by: Kevin Lubick <kjlubick@google.com>
Commit-Queue: Brian Osman <brianosman@google.com>
It would otherwise be difficult/tedious/not-backwards-compatible
to return the RuntimeEffect AND the error message.
Change-Id: I9bdbafb653398ccbb72e6e762ec4b6af294f9110
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/365483
Reviewed-by: Joe Gregorio <jcgregorio@google.com>
This will make cleaning up the PosTan value easier.
Change-Id: I13aa12f94c560bf8539aa2edb1f9e13779635692
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/362458
Reviewed-by: Nathaniel Nifong <nifong@google.com>
Using value_array (and value_object), while convenient,
adds a measurable overhead. This removes the Point value_object
and replaces it as a return value with Float32Array
(similar to rects). For inputs of a single point, I just
split it into x and y. For inputs with two points, I used
a _scratchFourFloats (formerly _scratchRect) bit of memory.
Two subtle decisions here:
- Why not use scratch memory for a single point? The cost of
having one extra param is a small/negligible price to pay
for less complex code.
- Why not accept Malloc objects? Again, simplicity. Accommodating
Malloc would make the code harder to read and require more
checks. I don't know if anyone wants to have malloced points;
if they do, we can probably accommodate that.
Change-Id: I1b1c29f62e01c2f1c8c1218f58e3bad642214322
Reviewed-on: https://skia-review.googlesource.com/c/skia/+/362097
Reviewed-by: Nathaniel Nifong <nifong@google.com>
