commit | de8474c5d27d279828fe45973c28b6d2af085a9f | [log] [tgz] |
---|---|---|
author | luigi-rosso <hello@rive.app> | Mon Sep 19 23:52:57 2022 +0000 |
committer | luigi-rosso <hello@rive.app> | Mon Sep 19 23:52:57 2022 +0000 |
tree | 2444765a67cece718f3c0589562cd397822fe2bd | |
parent | 0746c98c8b0ed9ee291c17f56d256be69796317d [diff] |
Implement a dynamic array (simple vector for marshaling). While working on marshaling the RenderFont abstraction to Dart FFI and WASM, it became clear that dealing with vectors inside of vectors gets tricky. We can wrap one single vector into a return struct like: ``` struct EasyToMarshal { std::vector<float>* ptrToVector; float* dataInVector; size_t size; }; ``` This is what we currently do and FFI/WASM is expected to then call in to C++ to delete ptrToVector when it's done with the data. This gets problematic to do when ptrToVector contains types that internally contain more vectors. The size and memory layout of vector isn't guaranteed across platforms with different STL implementations. The result from the shaper was a vector of RenderGlyphRuns: ``` struct RenderGlyphRun { rcp<RenderFont> font; float size; std::vector<GlyphID> glyphs; // [#glyphs] std::vector<uint32_t> textOffsets; // [#glyphs] std::vector<float> xpos; // [#glyphs + 1] }; std::vector<RenderGlyphRun> shapeText(...){} ``` This PR introduces a DynamicArray (totally up for renaming this to something less lame) which makes it really easy to marshal DynamicArrays inside of other DynamicArrays as we know the memory layout is just like rive::Span (a pointer and a size). Separately, this does make me wonder if we should use uint64_t for size in DynamicArrays so we can guarantee those are always 64 bit (the FFI and WASM tooling doesn't make it easy to marhsal runtime determined sizes). Diffs= e62737cf9 include cstring for memcpy 468dd3a2b Add growToFit, assert we didn’t overflow size_t, and in place constructor calls f0947d32f Add realloc down to actual used size from capacity. 516d0fc12 Move resize to builder and rename to simple array. 091904d22 malloc/realloc version e574043f7 More windows fixes c1f5b96ec Fix windows compiler warning. 164911445 Implement a dynamic array (simple vector for marshaling).
C++ runtime for Rive. Provides these runtime features:
We use premake5. The Rive dev team primarily works on MacOS. There is some work done by the community to also support Windows and Linux. PRs welcomed for specific platforms you wish to support! We encourage you to use premake as it's highly extensible and configurable for a variety of platforms.
In the rive-cpp
directory, run build.sh
to debug build and build.sh release
for a release build.
If you've put the premake5
executable in the rive-cpp/build
folder, you can run it with PATH=.:$PATH ./build.sh
Rive makes use of clang vector builtins, which are, as of 2022, still a work in progress. Please use clang and ensure you have the latest version.
cd skia/dependencies ./make_skia.sh // this will invoke get_skia.sh
To build viewer (plus you'll needed CMake installed)
./make_viewer_dependencies.sh
Uses the Catch2 testing framework.
cd dev ./test.sh
In the dev
directory, run test.sh
to compile and execute the tests.
(if you've installed premake5
in rive-cpp/build
, you can run it with PATH=../../build:$PATH ./test.sh
)
The tests live in rive/test
. To add new tests, create a new xxx_test.cpp
file here. The test harness will automatically pick up the new file.
There's a VSCode command provided to run tests
from the Tasks: Run Task command palette.
rive-cpp uses clang-format, you can install it with brew on MacOS: brew install clang-format
.
Note that if you‘re on MacOS you’ll want to install valgrind, which is somewhat complicated these days. This is the easiest solution (please PR a better one when it becomes available).
brew tap LouisBrunner/valgrind brew install --HEAD LouisBrunner/valgrind/valgrind
You can now run the all the tests through valgrind by running test.sh memory
.
If you want to examine the generated assembly code per cpp file, install Disassembly Explorer in VSCode.
A disassemble
task is provided to compile and preview the generated assembly. You can reach it via the Tasks: Run Task command palette or you can bind it to a shortcut by editing your VSCode keybindings.json:
[ { "key": "cmd+d", "command": "workbench.action.tasks.runTask", "args": "disassemble" } ]