Idea: use buffers to reuse hunks of data
diff --git a/include/rive/renderer.hpp b/include/rive/renderer.hpp
index bcbdaff..5b35d21 100644
--- a/include/rive/renderer.hpp
+++ b/include/rive/renderer.hpp
@@ -17,6 +17,19 @@
namespace rive {
class Vec2D;
+ // A render buffer holds an immutable array of values
+ class RenderBuffer : public RefCnt {
+ const size_t m_Count;
+ public:
+ RenderBuffer(size_t count) : m_Count(count) {}
+
+ size_t count() const { return m_Count; }
+ };
+
+ extern rcp<RenderBuffer> makeBufferU16(const uint16_t[], size_t count);
+ extern rcp<RenderBuffer> makeBufferU32(const uint32_t[], size_t count);
+ extern rcp<RenderBuffer> makeBufferF32(const float[], size_t count);
+
enum class RenderPaintStyle { stroke, fill };
enum class RenderTileMode {
diff --git a/skia/renderer/src/skia_renderer.cpp b/skia/renderer/src/skia_renderer.cpp
index 14fc57f..96eb5d8 100644
--- a/skia/renderer/src/skia_renderer.cpp
+++ b/skia/renderer/src/skia_renderer.cpp
@@ -7,6 +7,38 @@
using namespace rive;
+class SkiaBuffer : public RenderBuffer {
+ const size_t m_ElemSize;
+ void* m_Buffer;
+public:
+ SkiaBuffer(const void* src, size_t count, size_t elemSize)
+ : RenderBuffer(count)
+ , m_ElemSize(elemSize)
+ {
+ size_t bytes = count * elemSize;
+ m_Buffer = malloc(bytes);
+ memcpy(m_Buffer, src, bytes);
+ }
+
+ ~SkiaBuffer() {
+ free(m_Buffer);
+ }
+
+ const float* f32s() {
+ assert(m_ElemSize == sizeof(float));
+ return static_cast<const float*>(m_Buffer);
+ }
+
+ const uint16_t* u16s() const {
+ assert(m_ElemSize == sizeof(uint16_t));
+ return static_cast<const uint16_t*>(m_Buffer);
+ }
+};
+
+template <typename T> rcp<RenderBuffer> make_buffer(const T src[], size_t count) {
+ return rcp<RenderBuffer>(new SkiaBuffer(src, count, sizeof(T)));
+}
+
class SkiaRenderShader : public RenderShader {
public:
SkiaRenderShader(sk_sp<SkShader> sh) : shader(std::move(sh)) {}
@@ -114,6 +146,16 @@
}
namespace rive {
+ rcp<RenderBuffer> makeBufferU16(const uint16_t src[], size_t count) {
+ return make_buffer<uint16_t>(src, count);
+ }
+ rcp<RenderBuffer> makeBufferU32(const uint32_t src[], size_t count) {
+ return make_buffer<uint32_t>(src, count);
+ }
+ rcp<RenderBuffer> makeBufferF32(const float src[], size_t count) {
+ return make_buffer<float>(src, count);
+ }
+
RenderPath* makeRenderPath() { return new SkiaRenderPath(); }
RenderPaint* makeRenderPaint() { return new SkiaRenderPaint(); }
RenderImage* makeRenderImage() { return new SkiaRenderImage(); }
