skia2/tools/viewer/Viewer.cpp

1598 lines
61 KiB
C++
Raw Normal View History

/*
* Copyright 2016 Google Inc.
*
* Use of this source code is governed by a BSD-style license that can be
* found in the LICENSE file.
*/
#include "Viewer.h"
#include "GMSlide.h"
#include "ImageSlide.h"
#include "Resources.h"
#include "SampleSlide.h"
#include "SKPSlide.h"
#include "GrContext.h"
#include "SkATrace.h"
#include "SkCanvas.h"
#include "SkColorSpace_Base.h"
#include "SkColorSpacePriv.h"
#include "SkColorSpaceXformCanvas.h"
#include "SkCommandLineFlags.h"
#include "SkCommonFlagsGpuThreads.h"
#include "SkCommonFlagsPathRenderer.h"
#include "SkDashPathEffect.h"
#include "SkEventTracingPriv.h"
#include "SkGraphics.h"
#include "SkImagePriv.h"
#include "SkMetaData.h"
#include "SkOnce.h"
#include "SkOSFile.h"
#include "SkOSPath.h"
#include "SkRandom.h"
#include "SkScan.h"
#include "SkStream.h"
#include "SkSurface.h"
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
#include "SkSwizzle.h"
#include "SkTaskGroup.h"
#include "SkTime.h"
#include "SkVertices.h"
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
#include "imgui.h"
#include "ccpr/GrCoverageCountingPathRenderer.h"
#include <stdlib.h>
#include <map>
using namespace sk_app;
static std::map<GpuPathRenderers, std::string> gPathRendererNames;
Application* Application::Create(int argc, char** argv, void* platformData) {
return new Viewer(argc, argv, platformData);
}
static void on_backend_created_func(void* userData) {
Viewer* vv = reinterpret_cast<Viewer*>(userData);
return vv->onBackendCreated();
}
static void on_paint_handler(SkCanvas* canvas, void* userData) {
Viewer* vv = reinterpret_cast<Viewer*>(userData);
return vv->onPaint(canvas);
}
static bool on_touch_handler(intptr_t owner, Window::InputState state, float x, float y, void* userData)
{
Viewer* viewer = reinterpret_cast<Viewer*>(userData);
return viewer->onTouch(owner, state, x, y);
}
static void on_ui_state_changed_handler(const SkString& stateName, const SkString& stateValue, void* userData) {
Viewer* viewer = reinterpret_cast<Viewer*>(userData);
return viewer->onUIStateChanged(stateName, stateValue);
}
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
static bool on_mouse_handler(int x, int y, Window::InputState state, uint32_t modifiers,
void* userData) {
ImGuiIO& io = ImGui::GetIO();
io.MousePos.x = static_cast<float>(x);
io.MousePos.y = static_cast<float>(y);
if (Window::kDown_InputState == state) {
io.MouseDown[0] = true;
} else if (Window::kUp_InputState == state) {
io.MouseDown[0] = false;
}
if (io.WantCaptureMouse) {
return true;
} else {
Viewer* viewer = reinterpret_cast<Viewer*>(userData);
return viewer->onMouse(x, y, state, modifiers);
}
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
}
static bool on_mouse_wheel_handler(float delta, uint32_t modifiers, void* userData) {
ImGuiIO& io = ImGui::GetIO();
io.MouseWheel += delta;
return true;
}
static bool on_key_handler(Window::Key key, Window::InputState state, uint32_t modifiers,
void* userData) {
ImGuiIO& io = ImGui::GetIO();
io.KeysDown[static_cast<int>(key)] = (Window::kDown_InputState == state);
if (io.WantCaptureKeyboard) {
return true;
} else {
Viewer* viewer = reinterpret_cast<Viewer*>(userData);
return viewer->onKey(key, state, modifiers);
}
}
static bool on_char_handler(SkUnichar c, uint32_t modifiers, void* userData) {
ImGuiIO& io = ImGui::GetIO();
if (io.WantTextInput) {
if (c > 0 && c < 0x10000) {
io.AddInputCharacter(c);
}
return true;
} else {
Viewer* viewer = reinterpret_cast<Viewer*>(userData);
return viewer->onChar(c, modifiers);
}
}
static DEFINE_bool2(fullscreen, f, true, "Run fullscreen.");
static DEFINE_string2(match, m, nullptr,
"[~][^]substring[$] [...] of bench name to run.\n"
"Multiple matches may be separated by spaces.\n"
"~ causes a matching bench to always be skipped\n"
"^ requires the start of the bench to match\n"
"$ requires the end of the bench to match\n"
"^ and $ requires an exact match\n"
"If a bench does not match any list entry,\n"
"it is skipped unless some list entry starts with ~");
static DEFINE_string(slide, "", "Start on this sample.");
static DEFINE_bool(list, false, "List samples?");
#ifdef SK_VULKAN
# define BACKENDS_STR "\"sw\", \"gl\", and \"vk\""
#else
# define BACKENDS_STR "\"sw\" and \"gl\""
#endif
#ifdef SK_BUILD_FOR_ANDROID
static DEFINE_string(skps, "/data/local/tmp/skps", "Directory to read skps from.");
static DEFINE_string(jpgs, "/data/local/tmp/resources", "Directory to read jpgs from.");
#else
static DEFINE_string(skps, "skps", "Directory to read skps from.");
static DEFINE_string(jpgs, "jpgs", "Directory to read jpgs from.");
#endif
static DEFINE_string2(backend, b, "sw", "Backend to use. Allowed values are " BACKENDS_STR ".");
static DEFINE_int32(msaa, 0, "Number of subpixel samples. 0 for no HW antialiasing.");
static DEFINE_bool(cachePathMasks, true, "Allows path mask textures to be cached in GPU configs.");
DEFINE_pathrenderer_flag;
static DEFINE_bool(instancedRendering, false, "Enable instanced rendering on GPU backends.");
DECLARE_int32(threads)
const char* kBackendTypeStrings[sk_app::Window::kBackendTypeCount] = {
"OpenGL",
#if SK_ANGLE && defined(SK_BUILD_FOR_WIN)
"ANGLE",
#endif
#ifdef SK_VULKAN
"Vulkan",
#endif
"Raster"
};
static sk_app::Window::BackendType get_backend_type(const char* str) {
#ifdef SK_VULKAN
if (0 == strcmp(str, "vk")) {
return sk_app::Window::kVulkan_BackendType;
} else
#endif
#if SK_ANGLE && defined(SK_BUILD_FOR_WIN)
if (0 == strcmp(str, "angle")) {
return sk_app::Window::kANGLE_BackendType;
} else
#endif
if (0 == strcmp(str, "gl")) {
return sk_app::Window::kNativeGL_BackendType;
} else if (0 == strcmp(str, "sw")) {
return sk_app::Window::kRaster_BackendType;
} else {
SkDebugf("Unknown backend type, %s, defaulting to sw.", str);
return sk_app::Window::kRaster_BackendType;
}
}
static SkColorSpacePrimaries gSrgbPrimaries = {
0.64f, 0.33f,
0.30f, 0.60f,
0.15f, 0.06f,
0.3127f, 0.3290f };
static SkColorSpacePrimaries gAdobePrimaries = {
0.64f, 0.33f,
0.21f, 0.71f,
0.15f, 0.06f,
0.3127f, 0.3290f };
static SkColorSpacePrimaries gP3Primaries = {
0.680f, 0.320f,
0.265f, 0.690f,
0.150f, 0.060f,
0.3127f, 0.3290f };
static SkColorSpacePrimaries gRec2020Primaries = {
0.708f, 0.292f,
0.170f, 0.797f,
0.131f, 0.046f,
0.3127f, 0.3290f };
struct NamedPrimaries {
const char* fName;
SkColorSpacePrimaries* fPrimaries;
} gNamedPrimaries[] = {
{ "sRGB", &gSrgbPrimaries },
{ "AdobeRGB", &gAdobePrimaries },
{ "P3", &gP3Primaries },
{ "Rec. 2020", &gRec2020Primaries },
};
static bool primaries_equal(const SkColorSpacePrimaries& a, const SkColorSpacePrimaries& b) {
return memcmp(&a, &b, sizeof(SkColorSpacePrimaries)) == 0;
}
static Window::BackendType backend_type_for_window(Window::BackendType backendType) {
// In raster mode, we still use GL for the window.
// This lets us render the GUI faster (and correct).
return Window::kRaster_BackendType == backendType ? Window::kNativeGL_BackendType : backendType;
}
const char* kName = "name";
const char* kValue = "value";
const char* kOptions = "options";
const char* kSlideStateName = "Slide";
const char* kBackendStateName = "Backend";
const char* kMSAAStateName = "MSAA";
const char* kPathRendererStateName = "Path renderer";
const char* kInstancedRenderingStateName = "Instanced rendering";
const char* kSoftkeyStateName = "Softkey";
const char* kSoftkeyHint = "Please select a softkey";
const char* kFpsStateName = "FPS";
const char* kON = "ON";
const char* kOFF = "OFF";
const char* kRefreshStateName = "Refresh";
Viewer::Viewer(int argc, char** argv, void* platformData)
: fCurrentMeasurement(0)
, fCumulativeMeasurementTime(0)
, fCumulativeMeasurementCount(0)
, fDisplayStats(false)
, fRefresh(false)
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
, fShowImGuiDebugWindow(false)
, fShowImGuiTestWindow(false)
, fShowZoomWindow(false)
, fLastImage(nullptr)
, fBackendType(sk_app::Window::kNativeGL_BackendType)
, fColorMode(ColorMode::kLegacy)
, fColorSpacePrimaries(gSrgbPrimaries)
// Our UI can only tweak gamma (currently), so start out gamma-only
, fColorSpaceTransferFn(g2Dot2_TransferFn)
, fZoomLevel(0.0f)
, fGestureDevice(GestureDevice::kNone)
{
SkGraphics::Init();
static SkOnce initPathRendererNames;
initPathRendererNames([]() {
gPathRendererNames[GpuPathRenderers::kAll] = "All Path Renderers";
gPathRendererNames[GpuPathRenderers::kDefault] =
"Default Ganesh Behavior (best path renderer, not including CCPR)";
gPathRendererNames[GpuPathRenderers::kStencilAndCover] = "NV_path_rendering";
gPathRendererNames[GpuPathRenderers::kMSAA] = "Sample shading";
gPathRendererNames[GpuPathRenderers::kSmall] = "Small paths (cached sdf or alpha masks)";
gPathRendererNames[GpuPathRenderers::kCoverageCounting] = "Coverage counting";
gPathRendererNames[GpuPathRenderers::kTessellating] = "Tessellating";
gPathRendererNames[GpuPathRenderers::kNone] = "Software masks";
});
memset(fPaintTimes, 0, sizeof(fPaintTimes));
memset(fFlushTimes, 0, sizeof(fFlushTimes));
memset(fAnimateTimes, 0, sizeof(fAnimateTimes));
SkDebugf("Command line arguments: ");
for (int i = 1; i < argc; ++i) {
SkDebugf("%s ", argv[i]);
}
SkDebugf("\n");
SkCommandLineFlags::Parse(argc, argv);
#ifdef SK_BUILD_FOR_ANDROID
SetResourcePath("/data/local/tmp/resources");
#endif
initializeEventTracingForTools();
static SkTaskGroup::Enabler kTaskGroupEnabler(FLAGS_threads);
fBackendType = get_backend_type(FLAGS_backend[0]);
fWindow = Window::CreateNativeWindow(platformData);
DisplayParams displayParams;
displayParams.fMSAASampleCount = FLAGS_msaa;
displayParams.fGrContextOptions.fEnableInstancedRendering = FLAGS_instancedRendering;
displayParams.fGrContextOptions.fGpuPathRenderers = CollectGpuPathRenderersFromFlags();
displayParams.fGrContextOptions.fAllowPathMaskCaching = FLAGS_cachePathMasks;
displayParams.fGrContextOptions.fExecutor = GpuExecutorForTools();
fWindow->setRequestedDisplayParams(displayParams);
// register callbacks
fCommands.attach(fWindow);
fWindow->registerBackendCreatedFunc(on_backend_created_func, this);
fWindow->registerPaintFunc(on_paint_handler, this);
fWindow->registerTouchFunc(on_touch_handler, this);
fWindow->registerUIStateChangedFunc(on_ui_state_changed_handler, this);
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
fWindow->registerMouseFunc(on_mouse_handler, this);
fWindow->registerMouseWheelFunc(on_mouse_wheel_handler, this);
fWindow->registerKeyFunc(on_key_handler, this);
fWindow->registerCharFunc(on_char_handler, this);
// add key-bindings
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
fCommands.addCommand(' ', "GUI", "Toggle Debug GUI", [this]() {
this->fShowImGuiDebugWindow = !this->fShowImGuiDebugWindow;
fWindow->inval();
});
fCommands.addCommand('g', "GUI", "Toggle GUI Demo", [this]() {
this->fShowImGuiTestWindow = !this->fShowImGuiTestWindow;
fWindow->inval();
});
fCommands.addCommand('z', "GUI", "Toggle zoom window", [this]() {
this->fShowZoomWindow = !this->fShowZoomWindow;
fWindow->inval();
});
fCommands.addCommand('s', "Overlays", "Toggle stats display", [this]() {
this->fDisplayStats = !this->fDisplayStats;
fWindow->inval();
});
fCommands.addCommand('0', "Overlays", "Reset stats", [this]() {
this->resetMeasurements();
this->updateTitle();
fWindow->inval();
});
fCommands.addCommand('c', "Modes", "Cycle color mode", [this]() {
switch (fColorMode) {
case ColorMode::kLegacy:
this->setColorMode(ColorMode::kColorManagedSRGB8888_NonLinearBlending);
break;
case ColorMode::kColorManagedSRGB8888_NonLinearBlending:
this->setColorMode(ColorMode::kColorManagedSRGB8888);
break;
case ColorMode::kColorManagedSRGB8888:
this->setColorMode(ColorMode::kColorManagedLinearF16);
break;
case ColorMode::kColorManagedLinearF16:
this->setColorMode(ColorMode::kLegacy);
break;
}
});
fCommands.addCommand(Window::Key::kRight, "Right", "Navigation", "Next slide", [this]() {
int previousSlide = fCurrentSlide;
fCurrentSlide++;
if (fCurrentSlide >= fSlides.count()) {
fCurrentSlide = 0;
}
this->setupCurrentSlide(previousSlide);
});
fCommands.addCommand(Window::Key::kLeft, "Left", "Navigation", "Previous slide", [this]() {
int previousSlide = fCurrentSlide;
fCurrentSlide--;
if (fCurrentSlide < 0) {
fCurrentSlide = fSlides.count() - 1;
}
this->setupCurrentSlide(previousSlide);
});
fCommands.addCommand(Window::Key::kUp, "Up", "Transform", "Zoom in", [this]() {
this->changeZoomLevel(1.f / 32.f);
fWindow->inval();
});
fCommands.addCommand(Window::Key::kDown, "Down", "Transform", "Zoom out", [this]() {
this->changeZoomLevel(-1.f / 32.f);
fWindow->inval();
});
fCommands.addCommand('d', "Modes", "Change rendering backend", [this]() {
sk_app::Window::BackendType newBackend = (sk_app::Window::BackendType)(
(fBackendType + 1) % sk_app::Window::kBackendTypeCount);
// Switching to and from Vulkan is problematic on Linux so disabled for now
#if defined(SK_BUILD_FOR_UNIX) && defined(SK_VULKAN)
if (newBackend == sk_app::Window::kVulkan_BackendType) {
newBackend = (sk_app::Window::BackendType)((newBackend + 1) %
sk_app::Window::kBackendTypeCount);
} else if (fBackendType == sk_app::Window::kVulkan_BackendType) {
newBackend = sk_app::Window::kVulkan_BackendType;
}
#endif
this->setBackend(newBackend);
});
fCommands.addCommand('A', "AAA", "Toggle analytic AA", [this]() {
if (!gSkUseAnalyticAA) {
gSkUseAnalyticAA = true;
} else if (!gSkForceAnalyticAA) {
gSkForceAnalyticAA = true;
} else {
gSkUseAnalyticAA = gSkForceAnalyticAA = false;
}
this->updateTitle();
fWindow->inval();
});
fCommands.addCommand('D', "DAA", "Toggle delta AA", [this]() {
if (!gSkUseDeltaAA) {
gSkUseDeltaAA = true;
} else if (!gSkForceDeltaAA) {
gSkForceDeltaAA = true;
} else {
gSkUseDeltaAA = gSkForceDeltaAA = false;
}
this->updateTitle();
fWindow->inval();
});
// set up slides
this->initSlides();
this->setStartupSlide();
if (FLAGS_list) {
this->listNames();
}
fAnimTimer.run();
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
// ImGui initialization:
ImGuiIO& io = ImGui::GetIO();
io.DisplaySize.x = static_cast<float>(fWindow->width());
io.DisplaySize.y = static_cast<float>(fWindow->height());
// Keymap...
io.KeyMap[ImGuiKey_Tab] = (int)Window::Key::kTab;
io.KeyMap[ImGuiKey_LeftArrow] = (int)Window::Key::kLeft;
io.KeyMap[ImGuiKey_RightArrow] = (int)Window::Key::kRight;
io.KeyMap[ImGuiKey_UpArrow] = (int)Window::Key::kUp;
io.KeyMap[ImGuiKey_DownArrow] = (int)Window::Key::kDown;
io.KeyMap[ImGuiKey_PageUp] = (int)Window::Key::kPageUp;
io.KeyMap[ImGuiKey_PageDown] = (int)Window::Key::kPageDown;
io.KeyMap[ImGuiKey_Home] = (int)Window::Key::kHome;
io.KeyMap[ImGuiKey_End] = (int)Window::Key::kEnd;
io.KeyMap[ImGuiKey_Delete] = (int)Window::Key::kDelete;
io.KeyMap[ImGuiKey_Backspace] = (int)Window::Key::kBack;
io.KeyMap[ImGuiKey_Enter] = (int)Window::Key::kOK;
io.KeyMap[ImGuiKey_Escape] = (int)Window::Key::kEscape;
io.KeyMap[ImGuiKey_A] = (int)Window::Key::kA;
io.KeyMap[ImGuiKey_C] = (int)Window::Key::kC;
io.KeyMap[ImGuiKey_V] = (int)Window::Key::kV;
io.KeyMap[ImGuiKey_X] = (int)Window::Key::kX;
io.KeyMap[ImGuiKey_Y] = (int)Window::Key::kY;
io.KeyMap[ImGuiKey_Z] = (int)Window::Key::kZ;
int w, h;
unsigned char* pixels;
io.Fonts->GetTexDataAsAlpha8(&pixels, &w, &h);
SkImageInfo info = SkImageInfo::MakeA8(w, h);
SkPixmap pmap(info, pixels, info.minRowBytes());
SkMatrix localMatrix = SkMatrix::MakeScale(1.0f / w, 1.0f / h);
auto fontImage = SkImage::MakeFromRaster(pmap, nullptr, nullptr);
auto fontShader = fontImage->makeShader(&localMatrix);
fImGuiFontPaint.setShader(fontShader);
fImGuiFontPaint.setColor(SK_ColorWHITE);
fImGuiFontPaint.setFilterQuality(kLow_SkFilterQuality);
io.Fonts->TexID = &fImGuiFontPaint;
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
auto gamutImage = GetResourceAsImage("gamut.png");
if (gamutImage) {
fImGuiGamutPaint.setShader(gamutImage->makeShader());
}
fImGuiGamutPaint.setColor(SK_ColorWHITE);
fImGuiGamutPaint.setFilterQuality(kLow_SkFilterQuality);
fWindow->attach(backend_type_for_window(fBackendType));
}
void Viewer::initSlides() {
fAllSlideNames = Json::Value(Json::arrayValue);
const skiagm::GMRegistry* gms(skiagm::GMRegistry::Head());
while (gms) {
std::unique_ptr<skiagm::GM> gm(gms->factory()(nullptr));
if (!SkCommandLineFlags::ShouldSkip(FLAGS_match, gm->getName())) {
sk_sp<Slide> slide(new GMSlide(gm.release()));
fSlides.push_back(slide);
}
gms = gms->next();
}
// reverse array
for (int i = 0; i < fSlides.count()/2; ++i) {
sk_sp<Slide> temp = fSlides[i];
fSlides[i] = fSlides[fSlides.count() - i - 1];
fSlides[fSlides.count() - i - 1] = temp;
}
// samples
const SkViewRegister* reg = SkViewRegister::Head();
while (reg) {
sk_sp<Slide> slide(new SampleSlide(reg->factory()));
if (!SkCommandLineFlags::ShouldSkip(FLAGS_match, slide->getName().c_str())) {
fSlides.push_back(slide);
}
reg = reg->next();
}
// SKPs
for (int i = 0; i < FLAGS_skps.count(); i++) {
if (SkStrEndsWith(FLAGS_skps[i], ".skp")) {
if (SkCommandLineFlags::ShouldSkip(FLAGS_match, FLAGS_skps[i])) {
continue;
}
SkString path(FLAGS_skps[i]);
sk_sp<SKPSlide> slide(new SKPSlide(SkOSPath::Basename(path.c_str()), path));
if (slide) {
fSlides.push_back(slide);
}
} else {
SkOSFile::Iter it(FLAGS_skps[i], ".skp");
SkString skpName;
while (it.next(&skpName)) {
if (SkCommandLineFlags::ShouldSkip(FLAGS_match, skpName.c_str())) {
continue;
}
SkString path = SkOSPath::Join(FLAGS_skps[i], skpName.c_str());
sk_sp<SKPSlide> slide(new SKPSlide(skpName, path));
if (slide) {
fSlides.push_back(slide);
}
}
}
}
// JPGs
for (int i = 0; i < FLAGS_jpgs.count(); i++) {
SkOSFile::Iter it(FLAGS_jpgs[i], ".jpg");
SkString jpgName;
while (it.next(&jpgName)) {
if (SkCommandLineFlags::ShouldSkip(FLAGS_match, jpgName.c_str())) {
continue;
}
SkString path = SkOSPath::Join(FLAGS_jpgs[i], jpgName.c_str());
sk_sp<ImageSlide> slide(new ImageSlide(jpgName, path));
if (slide) {
fSlides.push_back(slide);
}
}
}
}
Viewer::~Viewer() {
fWindow->detach();
delete fWindow;
}
void Viewer::updateTitle() {
if (!fWindow) {
return;
}
if (fWindow->sampleCount() < 0) {
return; // Surface hasn't been created yet.
}
SkString title("Viewer: ");
title.append(fSlides[fCurrentSlide]->getName());
if (gSkUseDeltaAA) {
if (gSkForceDeltaAA) {
title.append(" <FDAA>");
} else {
title.append(" <DAA>");
}
} else if (gSkUseAnalyticAA) {
if (gSkForceAnalyticAA) {
title.append(" <FAAA>");
} else {
title.append(" <AAA>");
}
}
switch (fColorMode) {
case ColorMode::kLegacy:
title.append(" Legacy 8888");
break;
case ColorMode::kColorManagedSRGB8888_NonLinearBlending:
title.append(" ColorManaged 8888 (Nonlinear blending)");
break;
case ColorMode::kColorManagedSRGB8888:
title.append(" ColorManaged 8888");
break;
case ColorMode::kColorManagedLinearF16:
title.append(" ColorManaged F16");
break;
}
if (ColorMode::kLegacy != fColorMode) {
int curPrimaries = -1;
for (size_t i = 0; i < SK_ARRAY_COUNT(gNamedPrimaries); ++i) {
if (primaries_equal(*gNamedPrimaries[i].fPrimaries, fColorSpacePrimaries)) {
curPrimaries = i;
break;
}
}
title.appendf(" %s", curPrimaries >= 0 ? gNamedPrimaries[curPrimaries].fName : "Custom");
if (ColorMode::kColorManagedSRGB8888_NonLinearBlending == fColorMode) {
title.appendf(" Gamma %f", fColorSpaceTransferFn.fG);
}
}
title.append(" [");
title.append(kBackendTypeStrings[fBackendType]);
if (int msaa = fWindow->sampleCount()) {
title.appendf(" MSAA: %i", msaa);
}
title.append("]");
GpuPathRenderers pr = fWindow->getRequestedDisplayParams().fGrContextOptions.fGpuPathRenderers;
if (GpuPathRenderers::kDefault != pr) {
title.appendf(" [Path renderer: %s]", gPathRendererNames[pr].c_str());
}
fWindow->setTitle(title.c_str());
}
void Viewer::setStartupSlide() {
if (!FLAGS_slide.isEmpty()) {
int count = fSlides.count();
for (int i = 0; i < count; i++) {
if (fSlides[i]->getName().equals(FLAGS_slide[0])) {
fCurrentSlide = i;
return;
}
}
fprintf(stderr, "Unknown slide \"%s\"\n", FLAGS_slide[0]);
this->listNames();
}
fCurrentSlide = 0;
}
void Viewer::listNames() {
int count = fSlides.count();
SkDebugf("All Slides:\n");
for (int i = 0; i < count; i++) {
SkDebugf(" %s\n", fSlides[i]->getName().c_str());
}
}
void Viewer::resetMeasurements() {
memset(fPaintTimes, 0, sizeof(fPaintTimes));
memset(fFlushTimes, 0, sizeof(fFlushTimes));
memset(fAnimateTimes, 0, sizeof(fAnimateTimes));
fCurrentMeasurement = 0;
fCumulativeMeasurementTime = 0;
fCumulativeMeasurementCount = 0;
}
void Viewer::setupCurrentSlide(int previousSlide) {
if (fCurrentSlide == previousSlide) {
return; // no change; do nothing
}
// prepare dimensions for image slides
fSlides[fCurrentSlide]->load(SkIntToScalar(fWindow->width()), SkIntToScalar(fWindow->height()));
fGesture.reset();
fDefaultMatrix.reset();
Simplify some Viewer code, and fix a few bugs The content rect was always identical to the window rect, so most of the related code did nothing. The translation limit code is always useful (to avoid dragging the slide way off-screen with the mouse), so always include it. The auto-scaling to fit the screen is also still useful, but just base it on the window rect. The zoom code has four state variables, only used two of them, and one was a trivially derived computation. Fold most of that work into computeMatrix. (The translation was always zero -- we never changed the zoom center.) Include fDefaultMatrix in the matrix from computeMatrix, rather than needing to apply it specially to the canvas. Don't apply the inverse default matrix to touch or mouse points. The absolute positions of those touch points is not important, but because that matrix includes scale (and sometimes very large or very small scale), it just had the effect of greatly amplifying or damping the drag speed. Without it, the slide always pans at the speed of the touch/mouse drag -- which seems more desirable. The use of the inverse default matrix was a clever trick, but it caused the translation (applied to the global mtx) to be scaled, so the slide was always pinned incorrectly. Instead, supply the unmodified window rect and the default matrix, so the trans limit code can do the obvious correct thing: xform the slide bounds completely, then limit the translation that will be applied after that. Slides are now correctly pinned to screen edge regardless of how much zoom is present in the default matrix. Note: There are still several bugs related to all of this code, but given the web of xform state, it's hard to unravel. The touch gesture still doesn't know about viewer's zoom, so that's ignored when doing the pinning. Beyond that, it doesn't even know about window resize - it only configures the translation limit when setting up a slide. I had a fix for all of this (doing the translation limiting in computeMatrix), but then the touch gesture doesn't know about it, and can accumulate drag motion that needs to be un-dragged to get back on-screen, even though the slide is never really translated that far. SkTouchGesture is in include. No one uses it except viewer: TBR=bsalomon@google.com Bug: skia: Change-Id: I460cc07c3de6d36e63826f57d359faf1facf5ab3 Reviewed-on: https://skia-review.googlesource.com/18524 Reviewed-by: Brian Osman <brianosman@google.com> Reviewed-by: Yuqian Li <liyuqian@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-06-05 12:46:04 +00:00
const SkISize slideSize = fSlides[fCurrentSlide]->getDimensions();
const SkRect slideBounds = SkRect::MakeIWH(slideSize.width(), slideSize.height());
const SkRect windowRect = SkRect::MakeIWH(fWindow->width(), fWindow->height());
// Start with a matrix that scales the slide to the available screen space
if (fWindow->scaleContentToFit()) {
if (windowRect.width() > 0 && windowRect.height() > 0) {
fDefaultMatrix.setRectToRect(slideBounds, windowRect, SkMatrix::kStart_ScaleToFit);
}
}
Simplify some Viewer code, and fix a few bugs The content rect was always identical to the window rect, so most of the related code did nothing. The translation limit code is always useful (to avoid dragging the slide way off-screen with the mouse), so always include it. The auto-scaling to fit the screen is also still useful, but just base it on the window rect. The zoom code has four state variables, only used two of them, and one was a trivially derived computation. Fold most of that work into computeMatrix. (The translation was always zero -- we never changed the zoom center.) Include fDefaultMatrix in the matrix from computeMatrix, rather than needing to apply it specially to the canvas. Don't apply the inverse default matrix to touch or mouse points. The absolute positions of those touch points is not important, but because that matrix includes scale (and sometimes very large or very small scale), it just had the effect of greatly amplifying or damping the drag speed. Without it, the slide always pans at the speed of the touch/mouse drag -- which seems more desirable. The use of the inverse default matrix was a clever trick, but it caused the translation (applied to the global mtx) to be scaled, so the slide was always pinned incorrectly. Instead, supply the unmodified window rect and the default matrix, so the trans limit code can do the obvious correct thing: xform the slide bounds completely, then limit the translation that will be applied after that. Slides are now correctly pinned to screen edge regardless of how much zoom is present in the default matrix. Note: There are still several bugs related to all of this code, but given the web of xform state, it's hard to unravel. The touch gesture still doesn't know about viewer's zoom, so that's ignored when doing the pinning. Beyond that, it doesn't even know about window resize - it only configures the translation limit when setting up a slide. I had a fix for all of this (doing the translation limiting in computeMatrix), but then the touch gesture doesn't know about it, and can accumulate drag motion that needs to be un-dragged to get back on-screen, even though the slide is never really translated that far. SkTouchGesture is in include. No one uses it except viewer: TBR=bsalomon@google.com Bug: skia: Change-Id: I460cc07c3de6d36e63826f57d359faf1facf5ab3 Reviewed-on: https://skia-review.googlesource.com/18524 Reviewed-by: Brian Osman <brianosman@google.com> Reviewed-by: Yuqian Li <liyuqian@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-06-05 12:46:04 +00:00
// Prevent the user from dragging content so far outside the window they can't find it again
fGesture.setTransLimit(slideBounds, windowRect, fDefaultMatrix);
this->updateTitle();
this->updateUIState();
if (previousSlide >= 0) {
fSlides[previousSlide]->unload();
}
this->resetMeasurements();
fWindow->inval();
}
#define MAX_ZOOM_LEVEL 8
#define MIN_ZOOM_LEVEL -8
void Viewer::changeZoomLevel(float delta) {
fZoomLevel += delta;
Simplify some Viewer code, and fix a few bugs The content rect was always identical to the window rect, so most of the related code did nothing. The translation limit code is always useful (to avoid dragging the slide way off-screen with the mouse), so always include it. The auto-scaling to fit the screen is also still useful, but just base it on the window rect. The zoom code has four state variables, only used two of them, and one was a trivially derived computation. Fold most of that work into computeMatrix. (The translation was always zero -- we never changed the zoom center.) Include fDefaultMatrix in the matrix from computeMatrix, rather than needing to apply it specially to the canvas. Don't apply the inverse default matrix to touch or mouse points. The absolute positions of those touch points is not important, but because that matrix includes scale (and sometimes very large or very small scale), it just had the effect of greatly amplifying or damping the drag speed. Without it, the slide always pans at the speed of the touch/mouse drag -- which seems more desirable. The use of the inverse default matrix was a clever trick, but it caused the translation (applied to the global mtx) to be scaled, so the slide was always pinned incorrectly. Instead, supply the unmodified window rect and the default matrix, so the trans limit code can do the obvious correct thing: xform the slide bounds completely, then limit the translation that will be applied after that. Slides are now correctly pinned to screen edge regardless of how much zoom is present in the default matrix. Note: There are still several bugs related to all of this code, but given the web of xform state, it's hard to unravel. The touch gesture still doesn't know about viewer's zoom, so that's ignored when doing the pinning. Beyond that, it doesn't even know about window resize - it only configures the translation limit when setting up a slide. I had a fix for all of this (doing the translation limiting in computeMatrix), but then the touch gesture doesn't know about it, and can accumulate drag motion that needs to be un-dragged to get back on-screen, even though the slide is never really translated that far. SkTouchGesture is in include. No one uses it except viewer: TBR=bsalomon@google.com Bug: skia: Change-Id: I460cc07c3de6d36e63826f57d359faf1facf5ab3 Reviewed-on: https://skia-review.googlesource.com/18524 Reviewed-by: Brian Osman <brianosman@google.com> Reviewed-by: Yuqian Li <liyuqian@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-06-05 12:46:04 +00:00
fZoomLevel = SkScalarPin(fZoomLevel, MIN_ZOOM_LEVEL, MAX_ZOOM_LEVEL);
}
SkMatrix Viewer::computeMatrix() {
SkMatrix m;
Simplify some Viewer code, and fix a few bugs The content rect was always identical to the window rect, so most of the related code did nothing. The translation limit code is always useful (to avoid dragging the slide way off-screen with the mouse), so always include it. The auto-scaling to fit the screen is also still useful, but just base it on the window rect. The zoom code has four state variables, only used two of them, and one was a trivially derived computation. Fold most of that work into computeMatrix. (The translation was always zero -- we never changed the zoom center.) Include fDefaultMatrix in the matrix from computeMatrix, rather than needing to apply it specially to the canvas. Don't apply the inverse default matrix to touch or mouse points. The absolute positions of those touch points is not important, but because that matrix includes scale (and sometimes very large or very small scale), it just had the effect of greatly amplifying or damping the drag speed. Without it, the slide always pans at the speed of the touch/mouse drag -- which seems more desirable. The use of the inverse default matrix was a clever trick, but it caused the translation (applied to the global mtx) to be scaled, so the slide was always pinned incorrectly. Instead, supply the unmodified window rect and the default matrix, so the trans limit code can do the obvious correct thing: xform the slide bounds completely, then limit the translation that will be applied after that. Slides are now correctly pinned to screen edge regardless of how much zoom is present in the default matrix. Note: There are still several bugs related to all of this code, but given the web of xform state, it's hard to unravel. The touch gesture still doesn't know about viewer's zoom, so that's ignored when doing the pinning. Beyond that, it doesn't even know about window resize - it only configures the translation limit when setting up a slide. I had a fix for all of this (doing the translation limiting in computeMatrix), but then the touch gesture doesn't know about it, and can accumulate drag motion that needs to be un-dragged to get back on-screen, even though the slide is never really translated that far. SkTouchGesture is in include. No one uses it except viewer: TBR=bsalomon@google.com Bug: skia: Change-Id: I460cc07c3de6d36e63826f57d359faf1facf5ab3 Reviewed-on: https://skia-review.googlesource.com/18524 Reviewed-by: Brian Osman <brianosman@google.com> Reviewed-by: Yuqian Li <liyuqian@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-06-05 12:46:04 +00:00
SkScalar zoomScale = (fZoomLevel < 0) ? SK_Scalar1 / (SK_Scalar1 - fZoomLevel)
: SK_Scalar1 + fZoomLevel;
m = fGesture.localM();
m.preConcat(fGesture.globalM());
Simplify some Viewer code, and fix a few bugs The content rect was always identical to the window rect, so most of the related code did nothing. The translation limit code is always useful (to avoid dragging the slide way off-screen with the mouse), so always include it. The auto-scaling to fit the screen is also still useful, but just base it on the window rect. The zoom code has four state variables, only used two of them, and one was a trivially derived computation. Fold most of that work into computeMatrix. (The translation was always zero -- we never changed the zoom center.) Include fDefaultMatrix in the matrix from computeMatrix, rather than needing to apply it specially to the canvas. Don't apply the inverse default matrix to touch or mouse points. The absolute positions of those touch points is not important, but because that matrix includes scale (and sometimes very large or very small scale), it just had the effect of greatly amplifying or damping the drag speed. Without it, the slide always pans at the speed of the touch/mouse drag -- which seems more desirable. The use of the inverse default matrix was a clever trick, but it caused the translation (applied to the global mtx) to be scaled, so the slide was always pinned incorrectly. Instead, supply the unmodified window rect and the default matrix, so the trans limit code can do the obvious correct thing: xform the slide bounds completely, then limit the translation that will be applied after that. Slides are now correctly pinned to screen edge regardless of how much zoom is present in the default matrix. Note: There are still several bugs related to all of this code, but given the web of xform state, it's hard to unravel. The touch gesture still doesn't know about viewer's zoom, so that's ignored when doing the pinning. Beyond that, it doesn't even know about window resize - it only configures the translation limit when setting up a slide. I had a fix for all of this (doing the translation limiting in computeMatrix), but then the touch gesture doesn't know about it, and can accumulate drag motion that needs to be un-dragged to get back on-screen, even though the slide is never really translated that far. SkTouchGesture is in include. No one uses it except viewer: TBR=bsalomon@google.com Bug: skia: Change-Id: I460cc07c3de6d36e63826f57d359faf1facf5ab3 Reviewed-on: https://skia-review.googlesource.com/18524 Reviewed-by: Brian Osman <brianosman@google.com> Reviewed-by: Yuqian Li <liyuqian@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-06-05 12:46:04 +00:00
m.preConcat(fDefaultMatrix);
m.preScale(zoomScale, zoomScale);
return m;
}
void Viewer::setBackend(sk_app::Window::BackendType backendType) {
fBackendType = backendType;
fWindow->detach();
#if defined(SK_BUILD_FOR_WIN)
// Switching between OpenGL, Vulkan, and ANGLE in the same window is problematic at this point
// on Windows, so we just delete the window and recreate it.
DisplayParams params = fWindow->getRequestedDisplayParams();
delete fWindow;
fWindow = Window::CreateNativeWindow(nullptr);
// re-register callbacks
fCommands.attach(fWindow);
fWindow->registerBackendCreatedFunc(on_backend_created_func, this);
fWindow->registerPaintFunc(on_paint_handler, this);
fWindow->registerTouchFunc(on_touch_handler, this);
fWindow->registerUIStateChangedFunc(on_ui_state_changed_handler, this);
fWindow->registerMouseFunc(on_mouse_handler, this);
fWindow->registerMouseWheelFunc(on_mouse_wheel_handler, this);
fWindow->registerKeyFunc(on_key_handler, this);
fWindow->registerCharFunc(on_char_handler, this);
// Don't allow the window to re-attach. If we're in MSAA mode, the params we grabbed above
// will still include our correct sample count. But the re-created fWindow will lose that
// information. On Windows, we need to re-create the window when changing sample count,
// so we'll incorrectly detect that situation, then re-initialize the window in GL mode,
// rendering this tear-down step pointless (and causing the Vulkan window context to fail
// as if we had never changed windows at all).
fWindow->setRequestedDisplayParams(params, false);
#endif
fWindow->attach(backend_type_for_window(fBackendType));
}
void Viewer::setColorMode(ColorMode colorMode) {
fColorMode = colorMode;
// When we're in color managed mode, we tag our window surface as sRGB. If we've switched into
// or out of legacy/nonlinear mode, we need to update our window configuration.
DisplayParams params = fWindow->getRequestedDisplayParams();
bool wasInLegacy = !SkToBool(params.fColorSpace);
bool wantLegacy = (ColorMode::kLegacy == fColorMode) ||
(ColorMode::kColorManagedSRGB8888_NonLinearBlending == fColorMode);
if (wasInLegacy != wantLegacy) {
params.fColorSpace = wantLegacy ? nullptr : SkColorSpace::MakeSRGB();
fWindow->setRequestedDisplayParams(params);
}
this->updateTitle();
fWindow->inval();
}
void Viewer::drawSlide(SkCanvas* canvas) {
SkAutoCanvasRestore autorestore(canvas, false);
// By default, we render directly into the window's surface/canvas
SkCanvas* slideCanvas = canvas;
fLastImage.reset();
// If we're in any of the color managed modes, construct the color space we're going to use
sk_sp<SkColorSpace> cs = nullptr;
if (ColorMode::kLegacy != fColorMode) {
auto transferFn = (ColorMode::kColorManagedLinearF16 == fColorMode)
? SkColorSpace::kLinear_RenderTargetGamma : SkColorSpace::kSRGB_RenderTargetGamma;
SkMatrix44 toXYZ(SkMatrix44::kIdentity_Constructor);
SkAssertResult(fColorSpacePrimaries.toXYZD50(&toXYZ));
if (ColorMode::kColorManagedSRGB8888_NonLinearBlending == fColorMode) {
cs = SkColorSpace::MakeRGB(fColorSpaceTransferFn, toXYZ);
} else {
cs = SkColorSpace::MakeRGB(transferFn, toXYZ);
}
}
// If we're in F16, or we're zooming, or we're in color correct 8888 and the gamut isn't sRGB,
// we need to render offscreen. We also need to render offscreen if we're in any raster mode,
// because the window surface is actually GL.
sk_sp<SkSurface> offscreenSurface = nullptr;
if (Window::kRaster_BackendType == fBackendType ||
ColorMode::kColorManagedLinearF16 == fColorMode ||
fShowZoomWindow ||
(ColorMode::kColorManagedSRGB8888 == fColorMode &&
!primaries_equal(fColorSpacePrimaries, gSrgbPrimaries))) {
SkColorType colorType = (ColorMode::kColorManagedLinearF16 == fColorMode)
? kRGBA_F16_SkColorType : kN32_SkColorType;
// In nonlinear blending mode, we actually use a legacy off-screen canvas, and wrap it
// with a special canvas (below) that has the color space attached
sk_sp<SkColorSpace> offscreenColorSpace =
(ColorMode::kColorManagedSRGB8888_NonLinearBlending == fColorMode) ? nullptr : cs;
SkImageInfo info = SkImageInfo::Make(fWindow->width(), fWindow->height(), colorType,
kPremul_SkAlphaType, std::move(offscreenColorSpace));
offscreenSurface = Window::kRaster_BackendType == fBackendType ? SkSurface::MakeRaster(info)
: canvas->makeSurface(info);
slideCanvas = offscreenSurface->getCanvas();
}
std::unique_ptr<SkCanvas> xformCanvas = nullptr;
if (ColorMode::kColorManagedSRGB8888_NonLinearBlending == fColorMode) {
xformCanvas = SkCreateColorSpaceXformCanvas(slideCanvas, cs);
slideCanvas = xformCanvas.get();
}
int count = slideCanvas->save();
slideCanvas->clear(SK_ColorWHITE);
slideCanvas->concat(computeMatrix());
// Time the painting logic of the slide
double startTime = SkTime::GetMSecs();
fSlides[fCurrentSlide]->draw(slideCanvas);
fPaintTimes[fCurrentMeasurement] = SkTime::GetMSecs() - startTime;
slideCanvas->restoreToCount(count);
// Force a flush so we can time that, too
startTime = SkTime::GetMSecs();
slideCanvas->flush();
fFlushTimes[fCurrentMeasurement] = SkTime::GetMSecs() - startTime;
// If we rendered offscreen, snap an image and push the results to the window's canvas
if (offscreenSurface) {
fLastImage = offscreenSurface->makeImageSnapshot();
// Tag the image with the sRGB gamut, so no further color space conversion happens
sk_sp<SkColorSpace> srgb = (ColorMode::kColorManagedLinearF16 == fColorMode)
? SkColorSpace::MakeSRGBLinear() : SkColorSpace::MakeSRGB();
auto retaggedImage = SkImageMakeRasterCopyAndAssignColorSpace(fLastImage.get(), srgb.get());
SkPaint paint;
paint.setBlendMode(SkBlendMode::kSrc);
canvas->drawImage(retaggedImage, 0, 0, &paint);
}
}
void Viewer::onBackendCreated() {
this->updateTitle();
this->updateUIState();
this->setupCurrentSlide(-1);
this->resetMeasurements();
fWindow->show();
fWindow->inval();
}
void Viewer::onPaint(SkCanvas* canvas) {
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
// Update ImGui input
ImGuiIO& io = ImGui::GetIO();
io.DeltaTime = 1.0f / 60.0f;
io.DisplaySize.x = static_cast<float>(fWindow->width());
io.DisplaySize.y = static_cast<float>(fWindow->height());
io.KeyAlt = io.KeysDown[static_cast<int>(Window::Key::kOption)];
io.KeyCtrl = io.KeysDown[static_cast<int>(Window::Key::kCtrl)];
io.KeyShift = io.KeysDown[static_cast<int>(Window::Key::kShift)];
ImGui::NewFrame();
this->drawSlide(canvas);
// Advance our timing bookkeeping
fCumulativeMeasurementTime += fAnimateTimes[fCurrentMeasurement] +
fPaintTimes[fCurrentMeasurement] +
fFlushTimes[fCurrentMeasurement];
fCumulativeMeasurementCount++;
fCurrentMeasurement = (fCurrentMeasurement + 1) & (kMeasurementCount - 1);
SkASSERT(fCurrentMeasurement < kMeasurementCount);
// Draw any overlays or UI that we don't want timed
if (fDisplayStats) {
drawStats(canvas);
}
fCommands.drawHelp(canvas);
this->drawImGui(canvas);
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
// Update the FPS
this->updateUIState();
}
bool Viewer::onTouch(intptr_t owner, Window::InputState state, float x, float y) {
if (GestureDevice::kMouse == fGestureDevice) {
return false;
}
void* castedOwner = reinterpret_cast<void*>(owner);
switch (state) {
case Window::kUp_InputState: {
fGesture.touchEnd(castedOwner);
break;
}
case Window::kDown_InputState: {
Simplify some Viewer code, and fix a few bugs The content rect was always identical to the window rect, so most of the related code did nothing. The translation limit code is always useful (to avoid dragging the slide way off-screen with the mouse), so always include it. The auto-scaling to fit the screen is also still useful, but just base it on the window rect. The zoom code has four state variables, only used two of them, and one was a trivially derived computation. Fold most of that work into computeMatrix. (The translation was always zero -- we never changed the zoom center.) Include fDefaultMatrix in the matrix from computeMatrix, rather than needing to apply it specially to the canvas. Don't apply the inverse default matrix to touch or mouse points. The absolute positions of those touch points is not important, but because that matrix includes scale (and sometimes very large or very small scale), it just had the effect of greatly amplifying or damping the drag speed. Without it, the slide always pans at the speed of the touch/mouse drag -- which seems more desirable. The use of the inverse default matrix was a clever trick, but it caused the translation (applied to the global mtx) to be scaled, so the slide was always pinned incorrectly. Instead, supply the unmodified window rect and the default matrix, so the trans limit code can do the obvious correct thing: xform the slide bounds completely, then limit the translation that will be applied after that. Slides are now correctly pinned to screen edge regardless of how much zoom is present in the default matrix. Note: There are still several bugs related to all of this code, but given the web of xform state, it's hard to unravel. The touch gesture still doesn't know about viewer's zoom, so that's ignored when doing the pinning. Beyond that, it doesn't even know about window resize - it only configures the translation limit when setting up a slide. I had a fix for all of this (doing the translation limiting in computeMatrix), but then the touch gesture doesn't know about it, and can accumulate drag motion that needs to be un-dragged to get back on-screen, even though the slide is never really translated that far. SkTouchGesture is in include. No one uses it except viewer: TBR=bsalomon@google.com Bug: skia: Change-Id: I460cc07c3de6d36e63826f57d359faf1facf5ab3 Reviewed-on: https://skia-review.googlesource.com/18524 Reviewed-by: Brian Osman <brianosman@google.com> Reviewed-by: Yuqian Li <liyuqian@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-06-05 12:46:04 +00:00
fGesture.touchBegin(castedOwner, x, y);
break;
}
case Window::kMove_InputState: {
Simplify some Viewer code, and fix a few bugs The content rect was always identical to the window rect, so most of the related code did nothing. The translation limit code is always useful (to avoid dragging the slide way off-screen with the mouse), so always include it. The auto-scaling to fit the screen is also still useful, but just base it on the window rect. The zoom code has four state variables, only used two of them, and one was a trivially derived computation. Fold most of that work into computeMatrix. (The translation was always zero -- we never changed the zoom center.) Include fDefaultMatrix in the matrix from computeMatrix, rather than needing to apply it specially to the canvas. Don't apply the inverse default matrix to touch or mouse points. The absolute positions of those touch points is not important, but because that matrix includes scale (and sometimes very large or very small scale), it just had the effect of greatly amplifying or damping the drag speed. Without it, the slide always pans at the speed of the touch/mouse drag -- which seems more desirable. The use of the inverse default matrix was a clever trick, but it caused the translation (applied to the global mtx) to be scaled, so the slide was always pinned incorrectly. Instead, supply the unmodified window rect and the default matrix, so the trans limit code can do the obvious correct thing: xform the slide bounds completely, then limit the translation that will be applied after that. Slides are now correctly pinned to screen edge regardless of how much zoom is present in the default matrix. Note: There are still several bugs related to all of this code, but given the web of xform state, it's hard to unravel. The touch gesture still doesn't know about viewer's zoom, so that's ignored when doing the pinning. Beyond that, it doesn't even know about window resize - it only configures the translation limit when setting up a slide. I had a fix for all of this (doing the translation limiting in computeMatrix), but then the touch gesture doesn't know about it, and can accumulate drag motion that needs to be un-dragged to get back on-screen, even though the slide is never really translated that far. SkTouchGesture is in include. No one uses it except viewer: TBR=bsalomon@google.com Bug: skia: Change-Id: I460cc07c3de6d36e63826f57d359faf1facf5ab3 Reviewed-on: https://skia-review.googlesource.com/18524 Reviewed-by: Brian Osman <brianosman@google.com> Reviewed-by: Yuqian Li <liyuqian@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-06-05 12:46:04 +00:00
fGesture.touchMoved(castedOwner, x, y);
break;
}
}
fGestureDevice = fGesture.isBeingTouched() ? GestureDevice::kTouch : GestureDevice::kNone;
fWindow->inval();
return true;
}
bool Viewer::onMouse(float x, float y, Window::InputState state, uint32_t modifiers) {
if (!fSlides[fCurrentSlide]->onMouse(x, y, state, modifiers)) {
if (GestureDevice::kTouch == fGestureDevice) {
return false;
}
switch (state) {
case Window::kUp_InputState: {
fGesture.touchEnd(nullptr);
break;
}
case Window::kDown_InputState: {
fGesture.touchBegin(nullptr, x, y);
break;
}
case Window::kMove_InputState: {
fGesture.touchMoved(nullptr, x, y);
break;
}
}
fGestureDevice = fGesture.isBeingTouched() ? GestureDevice::kMouse : GestureDevice::kNone;
}
fWindow->inval();
return true;
}
void Viewer::drawStats(SkCanvas* canvas) {
static const float kPixelPerMS = 2.0f;
static const int kDisplayWidth = 192;
static const int kGraphHeight = 100;
static const int kTextHeight = 60;
static const int kDisplayHeight = kGraphHeight + kTextHeight;
static const int kDisplayPadding = 10;
static const int kGraphPadding = 3;
static const SkScalar kBaseMS = 1000.f / 60.f; // ms/frame to hit 60 fps
SkISize canvasSize = canvas->getBaseLayerSize();
SkRect rect = SkRect::MakeXYWH(SkIntToScalar(canvasSize.fWidth-kDisplayWidth-kDisplayPadding),
SkIntToScalar(kDisplayPadding),
SkIntToScalar(kDisplayWidth), SkIntToScalar(kDisplayHeight));
SkPaint paint;
canvas->save();
paint.setColor(SK_ColorBLACK);
canvas->drawRect(rect, paint);
// draw the 16ms line
paint.setColor(SK_ColorLTGRAY);
canvas->drawLine(rect.fLeft, rect.fBottom - kBaseMS*kPixelPerMS,
rect.fRight, rect.fBottom - kBaseMS*kPixelPerMS, paint);
paint.setColor(SK_ColorRED);
paint.setStyle(SkPaint::kStroke_Style);
canvas->drawRect(rect, paint);
paint.setStyle(SkPaint::kFill_Style);
int x = SkScalarTruncToInt(rect.fLeft) + kGraphPadding;
const int xStep = 3;
int i = fCurrentMeasurement;
double ms = 0;
double animateMS = 0;
double paintMS = 0;
double flushMS = 0;
int count = 0;
do {
// Round to nearest values
int animateHeight = (int)(fAnimateTimes[i] * kPixelPerMS + 0.5);
int paintHeight = (int)(fPaintTimes[i] * kPixelPerMS + 0.5);
int flushHeight = (int)(fFlushTimes[i] * kPixelPerMS + 0.5);
int startY = SkScalarTruncToInt(rect.fBottom);
int endY = SkTMax(startY - flushHeight, kDisplayPadding + kTextHeight);
paint.setColor(SK_ColorRED);
canvas->drawLine(SkIntToScalar(x), SkIntToScalar(startY),
SkIntToScalar(x), SkIntToScalar(endY), paint);
startY = endY;
endY = SkTMax(startY - paintHeight, kDisplayPadding + kTextHeight);
paint.setColor(SK_ColorGREEN);
canvas->drawLine(SkIntToScalar(x), SkIntToScalar(startY),
SkIntToScalar(x), SkIntToScalar(endY), paint);
startY = endY;
endY = SkTMax(startY - animateHeight, kDisplayPadding + kTextHeight);
paint.setColor(SK_ColorMAGENTA);
canvas->drawLine(SkIntToScalar(x), SkIntToScalar(startY),
SkIntToScalar(x), SkIntToScalar(endY), paint);
double inc = fAnimateTimes[i] + fPaintTimes[i] + fFlushTimes[i];
if (inc > 0) {
ms += inc;
animateMS += fAnimateTimes[i];
paintMS += fPaintTimes[i];
flushMS += fFlushTimes[i];
++count;
}
i++;
i &= (kMeasurementCount - 1); // fast mod
x += xStep;
} while (i != fCurrentMeasurement);
paint.setTextSize(16);
SkString mainString;
mainString.appendf("%4.3f ms -> %4.3f ms", ms / SkTMax(1, count),
fCumulativeMeasurementTime / SkTMax(1, fCumulativeMeasurementCount));
paint.setColor(SK_ColorWHITE);
canvas->drawString(mainString.c_str(), rect.fLeft+3, rect.fTop + 14, paint);
SkString animateString;
animateString.appendf("Animate: %4.3f ms", animateMS / SkTMax(1, count));
paint.setColor(0xffff66ff); // pure magenta is hard to read
canvas->drawString(animateString.c_str(), rect.fLeft+3, rect.fTop + 28, paint);
SkString paintString;
paintString.appendf("Paint: %4.3f ms", paintMS / SkTMax(1, count));
paint.setColor(SK_ColorGREEN);
canvas->drawString(paintString.c_str(), rect.fLeft+3, rect.fTop + 42, paint);
SkString flushString;
flushString.appendf("Flush: %4.3f ms", flushMS / SkTMax(1, count));
paint.setColor(0xffff6666); // pure red is hard to read
canvas->drawString(flushString.c_str(), rect.fLeft+3, rect.fTop + 56, paint);
canvas->restore();
}
static ImVec2 ImGui_DragPrimary(const char* label, float* x, float* y,
const ImVec2& pos, const ImVec2& size) {
// Transform primaries ([0, 0] - [0.8, 0.9]) to screen coords (including Y-flip)
ImVec2 center(pos.x + (*x / 0.8f) * size.x, pos.y + (1.0f - (*y / 0.9f)) * size.y);
// Invisible 10x10 button
ImGui::SetCursorScreenPos(ImVec2(center.x - 5, center.y - 5));
ImGui::InvisibleButton(label, ImVec2(10, 10));
if (ImGui::IsItemActive() && ImGui::IsMouseDragging()) {
ImGuiIO& io = ImGui::GetIO();
// Normalized mouse position, relative to our gamut box
ImVec2 mousePosXY((io.MousePos.x - pos.x) / size.x, (io.MousePos.y - pos.y) / size.y);
// Clamp to edge of box, convert back to primary scale
*x = SkTPin(mousePosXY.x, 0.0f, 1.0f) * 0.8f;
*y = SkTPin(1 - mousePosXY.y, 0.0f, 1.0f) * 0.9f;
}
if (ImGui::IsItemHovered()) {
ImGui::SetTooltip("x: %.3f\ny: %.3f", *x, *y);
}
// Return screen coordinates for the caller. We could just return center here, but we'd have
// one frame of lag during drag.
return ImVec2(pos.x + (*x / 0.8f) * size.x, pos.y + (1.0f - (*y / 0.9f)) * size.y);
}
static void ImGui_Primaries(SkColorSpacePrimaries* primaries, SkPaint* gamutPaint) {
ImDrawList* drawList = ImGui::GetWindowDrawList();
// The gamut image covers a (0.8 x 0.9) shaped region, so fit our image/canvas to the available
// width, and scale the height to maintain aspect ratio.
float canvasWidth = SkTMax(ImGui::GetContentRegionAvailWidth(), 50.0f);
ImVec2 size = ImVec2(canvasWidth, canvasWidth * (0.9f / 0.8f));
ImVec2 pos = ImGui::GetCursorScreenPos();
// Background image. Only draw a subset of the image, to avoid the regions less than zero.
// Simplifes re-mapping math, clipping behavior, and increases resolution in the useful area.
// Magic numbers are pixel locations of the origin and upper-right corner.
drawList->AddImage(gamutPaint, pos, ImVec2(pos.x + size.x, pos.y + size.y),
ImVec2(242, 61), ImVec2(1897, 1922));
ImVec2 endPos = ImGui::GetCursorPos();
// Primary markers
ImVec2 r = ImGui_DragPrimary("R", &primaries->fRX, &primaries->fRY, pos, size);
ImVec2 g = ImGui_DragPrimary("G", &primaries->fGX, &primaries->fGY, pos, size);
ImVec2 b = ImGui_DragPrimary("B", &primaries->fBX, &primaries->fBY, pos, size);
ImVec2 w = ImGui_DragPrimary("W", &primaries->fWX, &primaries->fWY, pos, size);
// Gamut triangle
drawList->AddCircle(r, 5.0f, 0xFF000040);
drawList->AddCircle(g, 5.0f, 0xFF004000);
drawList->AddCircle(b, 5.0f, 0xFF400000);
drawList->AddCircle(w, 5.0f, 0xFFFFFFFF);
drawList->AddTriangle(r, g, b, 0xFFFFFFFF);
// Re-position cursor immediate after the diagram for subsequent controls
ImGui::SetCursorPos(endPos);
}
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
void Viewer::drawImGui(SkCanvas* canvas) {
// Support drawing the ImGui demo window. Superfluous, but gives a good idea of what's possible
if (fShowImGuiTestWindow) {
ImGui::ShowTestWindow(&fShowImGuiTestWindow);
}
if (fShowImGuiDebugWindow) {
// We have some dynamic content that sizes to fill available size. If the scroll bar isn't
// always visible, we can end up in a layout feedback loop.
ImGui::SetNextWindowSize(ImVec2(400, 400), ImGuiSetCond_FirstUseEver);
DisplayParams params = fWindow->getRequestedDisplayParams();
bool paramsChanged = false;
if (ImGui::Begin("Tools", &fShowImGuiDebugWindow,
ImGuiWindowFlags_AlwaysVerticalScrollbar)) {
if (ImGui::CollapsingHeader("Backend")) {
int newBackend = static_cast<int>(fBackendType);
ImGui::RadioButton("Raster", &newBackend, sk_app::Window::kRaster_BackendType);
ImGui::SameLine();
ImGui::RadioButton("OpenGL", &newBackend, sk_app::Window::kNativeGL_BackendType);
#if SK_ANGLE && defined(SK_BUILD_FOR_WIN)
ImGui::SameLine();
ImGui::RadioButton("ANGLE", &newBackend, sk_app::Window::kANGLE_BackendType);
#endif
#if defined(SK_VULKAN)
ImGui::SameLine();
ImGui::RadioButton("Vulkan", &newBackend, sk_app::Window::kVulkan_BackendType);
#endif
if (newBackend != fBackendType) {
fDeferredActions.push_back([=]() {
this->setBackend(static_cast<sk_app::Window::BackendType>(newBackend));
});
}
const GrContext* ctx = fWindow->getGrContext();
bool* inst = &params.fGrContextOptions.fEnableInstancedRendering;
if (ctx && ImGui::Checkbox("Instanced Rendering", inst)) {
paramsChanged = true;
}
bool* wire = &params.fGrContextOptions.fWireframeMode;
if (ctx && ImGui::Checkbox("Wireframe Mode", wire)) {
paramsChanged = true;
}
if (ctx) {
int sampleCount = fWindow->sampleCount();
ImGui::Text("MSAA: "); ImGui::SameLine();
ImGui::RadioButton("0", &sampleCount, 0); ImGui::SameLine();
ImGui::RadioButton("4", &sampleCount, 4); ImGui::SameLine();
ImGui::RadioButton("8", &sampleCount, 8); ImGui::SameLine();
ImGui::RadioButton("16", &sampleCount, 16);
if (sampleCount != params.fMSAASampleCount) {
params.fMSAASampleCount = sampleCount;
paramsChanged = true;
}
}
if (ImGui::TreeNode("Path Renderers")) {
GpuPathRenderers prevPr = params.fGrContextOptions.fGpuPathRenderers;
auto prButton = [&](GpuPathRenderers x) {
if (ImGui::RadioButton(gPathRendererNames[x].c_str(), prevPr == x)) {
if (x != params.fGrContextOptions.fGpuPathRenderers) {
params.fGrContextOptions.fGpuPathRenderers = x;
paramsChanged = true;
}
}
};
if (!ctx) {
ImGui::RadioButton("Software", true);
} else if (fWindow->sampleCount()) {
prButton(GpuPathRenderers::kDefault);
prButton(GpuPathRenderers::kAll);
if (ctx->caps()->shaderCaps()->pathRenderingSupport()) {
prButton(GpuPathRenderers::kStencilAndCover);
}
if (ctx->caps()->sampleShadingSupport()) {
prButton(GpuPathRenderers::kMSAA);
}
prButton(GpuPathRenderers::kTessellating);
prButton(GpuPathRenderers::kNone);
} else {
prButton(GpuPathRenderers::kDefault);
prButton(GpuPathRenderers::kAll);
if (GrCoverageCountingPathRenderer::IsSupported(*ctx->caps())) {
prButton(GpuPathRenderers::kCoverageCounting);
}
prButton(GpuPathRenderers::kSmall);
prButton(GpuPathRenderers::kTessellating);
prButton(GpuPathRenderers::kNone);
}
ImGui::TreePop();
}
}
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
if (ImGui::CollapsingHeader("Slide")) {
static ImGuiTextFilter filter;
static ImVector<const char*> filteredSlideNames;
static ImVector<int> filteredSlideIndices;
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
filter.Draw();
filteredSlideNames.clear();
filteredSlideIndices.clear();
int filteredIndex = 0;
for (int i = 0; i < fSlides.count(); ++i) {
const char* slideName = fSlides[i]->getName().c_str();
if (filter.PassFilter(slideName) || i == fCurrentSlide) {
if (i == fCurrentSlide) {
filteredIndex = filteredSlideIndices.size();
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
}
filteredSlideNames.push_back(slideName);
filteredSlideIndices.push_back(i);
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
}
}
int previousSlide = fCurrentSlide;
if (ImGui::ListBox("", &filteredIndex, filteredSlideNames.begin(),
filteredSlideNames.size(), 20)) {
fCurrentSlide = filteredSlideIndices[filteredIndex];
setupCurrentSlide(previousSlide);
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
}
}
if (ImGui::CollapsingHeader("Color Mode")) {
ColorMode newMode = fColorMode;
auto cmButton = [&](ColorMode mode, const char* label) {
if (ImGui::RadioButton(label, mode == fColorMode)) {
newMode = mode;
}
};
cmButton(ColorMode::kLegacy, "Legacy 8888");
cmButton(ColorMode::kColorManagedSRGB8888_NonLinearBlending,
"Color Managed 8888 (Nonlinear blending)");
cmButton(ColorMode::kColorManagedSRGB8888, "Color Managed 8888");
cmButton(ColorMode::kColorManagedLinearF16, "Color Managed F16");
if (newMode != fColorMode) {
// It isn't safe to switch color mode now (in the middle of painting). We might
// tear down the back-end, etc... Defer this change until the next onIdle.
fDeferredActions.push_back([=]() {
this->setColorMode(newMode);
});
}
// Pick from common gamuts:
int primariesIdx = 4; // Default: Custom
for (size_t i = 0; i < SK_ARRAY_COUNT(gNamedPrimaries); ++i) {
if (primaries_equal(*gNamedPrimaries[i].fPrimaries, fColorSpacePrimaries)) {
primariesIdx = i;
break;
}
}
// When we're in xform canvas mode, we can alter the transfer function, too
if (ColorMode::kColorManagedSRGB8888_NonLinearBlending == fColorMode) {
ImGui::SliderFloat("Gamma", &fColorSpaceTransferFn.fG, 0.5f, 3.5f);
}
if (ImGui::Combo("Primaries", &primariesIdx,
"sRGB\0AdobeRGB\0P3\0Rec. 2020\0Custom\0\0")) {
if (primariesIdx >= 0 && primariesIdx <= 3) {
fColorSpacePrimaries = *gNamedPrimaries[primariesIdx].fPrimaries;
}
}
// Allow direct editing of gamut
ImGui_Primaries(&fColorSpacePrimaries, &fImGuiGamutPaint);
}
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
}
if (paramsChanged) {
fDeferredActions.push_back([=]() {
fWindow->setRequestedDisplayParams(params);
fWindow->inval();
this->updateTitle();
});
}
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
ImGui::End();
}
SkPaint zoomImagePaint;
if (fShowZoomWindow && fLastImage) {
if (ImGui::Begin("Zoom", &fShowZoomWindow, ImVec2(200, 200))) {
static int zoomFactor = 4;
ImGui::SliderInt("Scale", &zoomFactor, 1, 16);
zoomImagePaint.setShader(fLastImage->makeShader());
zoomImagePaint.setColor(SK_ColorWHITE);
// Zoom by shrinking the corner UVs towards the mouse cursor
ImVec2 mousePos = ImGui::GetMousePos();
ImVec2 avail = ImGui::GetContentRegionAvail();
ImVec2 zoomHalfExtents = ImVec2((avail.x * 0.5f) / zoomFactor,
(avail.y * 0.5f) / zoomFactor);
ImGui::Image(&zoomImagePaint, avail,
ImVec2(mousePos.x - zoomHalfExtents.x, mousePos.y - zoomHalfExtents.y),
ImVec2(mousePos.x + zoomHalfExtents.x, mousePos.y + zoomHalfExtents.y));
}
ImGui::End();
}
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
// This causes ImGui to rebuild vertex/index data based on all immediate-mode commands
// (widgets, etc...) that have been issued
ImGui::Render();
// Then we fetch the most recent data, and convert it so we can render with Skia
const ImDrawData* drawData = ImGui::GetDrawData();
SkTDArray<SkPoint> pos;
SkTDArray<SkPoint> uv;
SkTDArray<SkColor> color;
for (int i = 0; i < drawData->CmdListsCount; ++i) {
const ImDrawList* drawList = drawData->CmdLists[i];
// De-interleave all vertex data (sigh), convert to Skia types
pos.rewind(); uv.rewind(); color.rewind();
for (int i = 0; i < drawList->VtxBuffer.size(); ++i) {
const ImDrawVert& vert = drawList->VtxBuffer[i];
pos.push(SkPoint::Make(vert.pos.x, vert.pos.y));
uv.push(SkPoint::Make(vert.uv.x, vert.uv.y));
color.push(vert.col);
}
// ImGui colors are RGBA
SkSwapRB(color.begin(), color.begin(), color.count());
int indexOffset = 0;
// Draw everything with canvas.drawVertices...
for (int j = 0; j < drawList->CmdBuffer.size(); ++j) {
const ImDrawCmd* drawCmd = &drawList->CmdBuffer[j];
// TODO: Find min/max index for each draw, so we know how many vertices (sigh)
if (drawCmd->UserCallback) {
drawCmd->UserCallback(drawList, drawCmd);
} else {
SkPaint* paint = static_cast<SkPaint*>(drawCmd->TextureId);
SkASSERT(paint);
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
canvas->save();
canvas->clipRect(SkRect::MakeLTRB(drawCmd->ClipRect.x, drawCmd->ClipRect.y,
drawCmd->ClipRect.z, drawCmd->ClipRect.w));
canvas->drawVertices(SkVertices::MakeCopy(SkVertices::kTriangles_VertexMode,
drawList->VtxBuffer.size(), pos.begin(),
uv.begin(), color.begin(),
drawCmd->ElemCount,
drawList->IdxBuffer.begin() + indexOffset),
SkBlendMode::kModulate, *paint);
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
indexOffset += drawCmd->ElemCount;
canvas->restore();
}
}
}
}
void Viewer::onIdle() {
for (int i = 0; i < fDeferredActions.count(); ++i) {
fDeferredActions[i]();
}
fDeferredActions.reset();
double startTime = SkTime::GetMSecs();
fAnimTimer.updateTime();
bool animateWantsInval = fSlides[fCurrentSlide]->animate(fAnimTimer);
fAnimateTimes[fCurrentMeasurement] = SkTime::GetMSecs() - startTime;
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
ImGuiIO& io = ImGui::GetIO();
if (animateWantsInval || fDisplayStats || fRefresh || io.MetricsActiveWindows) {
fWindow->inval();
}
}
void Viewer::updateUIState() {
if (!fWindow) {
return;
}
if (fWindow->sampleCount() < 0) {
return; // Surface hasn't been created yet.
}
// Slide state
Json::Value slideState(Json::objectValue);
slideState[kName] = kSlideStateName;
slideState[kValue] = fSlides[fCurrentSlide]->getName().c_str();
if (fAllSlideNames.size() == 0) {
for(auto slide : fSlides) {
fAllSlideNames.append(Json::Value(slide->getName().c_str()));
}
}
slideState[kOptions] = fAllSlideNames;
// Backend state
Json::Value backendState(Json::objectValue);
backendState[kName] = kBackendStateName;
backendState[kValue] = kBackendTypeStrings[fBackendType];
backendState[kOptions] = Json::Value(Json::arrayValue);
for (auto str : kBackendTypeStrings) {
backendState[kOptions].append(Json::Value(str));
}
// MSAA state
Json::Value msaaState(Json::objectValue);
msaaState[kName] = kMSAAStateName;
msaaState[kValue] = fWindow->sampleCount();
msaaState[kOptions] = Json::Value(Json::arrayValue);
if (sk_app::Window::kRaster_BackendType == fBackendType) {
msaaState[kOptions].append(Json::Value(0));
} else {
for (int msaa : {0, 4, 8, 16}) {
msaaState[kOptions].append(Json::Value(msaa));
}
}
// Path renderer state
GpuPathRenderers pr = fWindow->getRequestedDisplayParams().fGrContextOptions.fGpuPathRenderers;
Json::Value prState(Json::objectValue);
prState[kName] = kPathRendererStateName;
prState[kValue] = gPathRendererNames[pr];
prState[kOptions] = Json::Value(Json::arrayValue);
const GrContext* ctx = fWindow->getGrContext();
if (!ctx) {
prState[kOptions].append("Software");
} else if (fWindow->sampleCount()) {
prState[kOptions].append(gPathRendererNames[GpuPathRenderers::kDefault]);
prState[kOptions].append(gPathRendererNames[GpuPathRenderers::kAll]);
if (ctx->caps()->shaderCaps()->pathRenderingSupport()) {
prState[kOptions].append(gPathRendererNames[GpuPathRenderers::kStencilAndCover]);
}
if (ctx->caps()->sampleShadingSupport()) {
prState[kOptions].append(gPathRendererNames[GpuPathRenderers::kMSAA]);
}
prState[kOptions].append(gPathRendererNames[GpuPathRenderers::kTessellating]);
prState[kOptions].append(gPathRendererNames[GpuPathRenderers::kNone]);
} else {
prState[kOptions].append(gPathRendererNames[GpuPathRenderers::kDefault]);
prState[kOptions].append(gPathRendererNames[GpuPathRenderers::kAll]);
if (GrCoverageCountingPathRenderer::IsSupported(*ctx->caps())) {
prState[kOptions].append(gPathRendererNames[GpuPathRenderers::kCoverageCounting]);
}
prState[kOptions].append(gPathRendererNames[GpuPathRenderers::kSmall]);
prState[kOptions].append(gPathRendererNames[GpuPathRenderers::kTessellating]);
prState[kOptions].append(gPathRendererNames[GpuPathRenderers::kNone]);
}
// Instanced rendering state
Json::Value instState(Json::objectValue);
instState[kName] = kInstancedRenderingStateName;
instState[kValue] = kOFF;
instState[kOptions] = Json::Value(Json::arrayValue);
if (ctx) {
if (fWindow->getRequestedDisplayParams().fGrContextOptions.fEnableInstancedRendering) {
instState[kValue] = kON;
}
instState[kOptions].append(kOFF);
instState[kOptions].append(kON);
}
// Softkey state
Json::Value softkeyState(Json::objectValue);
softkeyState[kName] = kSoftkeyStateName;
softkeyState[kValue] = kSoftkeyHint;
softkeyState[kOptions] = Json::Value(Json::arrayValue);
softkeyState[kOptions].append(kSoftkeyHint);
for (const auto& softkey : fCommands.getCommandsAsSoftkeys()) {
softkeyState[kOptions].append(Json::Value(softkey.c_str()));
}
// FPS state
Json::Value fpsState(Json::objectValue);
fpsState[kName] = kFpsStateName;
int idx = (fCurrentMeasurement + (kMeasurementCount - 1)) & (kMeasurementCount - 1);
fpsState[kValue] = SkStringPrintf("%8.3lf ms\n\nA %8.3lf\nP %8.3lf\nF%8.3lf",
fAnimateTimes[idx] + fPaintTimes[idx] + fFlushTimes[idx],
fAnimateTimes[idx],
fPaintTimes[idx],
fFlushTimes[idx]).c_str();
fpsState[kOptions] = Json::Value(Json::arrayValue);
Json::Value state(Json::arrayValue);
state.append(slideState);
state.append(backendState);
state.append(msaaState);
state.append(prState);
state.append(instState);
state.append(softkeyState);
state.append(fpsState);
fWindow->setUIState(state);
}
void Viewer::onUIStateChanged(const SkString& stateName, const SkString& stateValue) {
// For those who will add more features to handle the state change in this function:
// After the change, please call updateUIState no notify the frontend (e.g., Android app).
// For example, after slide change, updateUIState is called inside setupCurrentSlide;
// after backend change, updateUIState is called in this function.
if (stateName.equals(kSlideStateName)) {
int previousSlide = fCurrentSlide;
fCurrentSlide = 0;
for(auto slide : fSlides) {
if (slide->getName().equals(stateValue)) {
this->setupCurrentSlide(previousSlide);
break;
}
fCurrentSlide++;
}
if (fCurrentSlide >= fSlides.count()) {
fCurrentSlide = previousSlide;
SkDebugf("Slide not found: %s", stateValue.c_str());
}
} else if (stateName.equals(kBackendStateName)) {
for (int i = 0; i < sk_app::Window::kBackendTypeCount; i++) {
if (stateValue.equals(kBackendTypeStrings[i])) {
if (fBackendType != i) {
fBackendType = (sk_app::Window::BackendType)i;
fWindow->detach();
fWindow->attach(backend_type_for_window(fBackendType));
}
break;
}
}
} else if (stateName.equals(kMSAAStateName)) {
DisplayParams params = fWindow->getRequestedDisplayParams();
int sampleCount = atoi(stateValue.c_str());
if (sampleCount != params.fMSAASampleCount) {
params.fMSAASampleCount = sampleCount;
fWindow->setRequestedDisplayParams(params);
fWindow->inval();
this->updateTitle();
this->updateUIState();
}
} else if (stateName.equals(kPathRendererStateName)) {
DisplayParams params = fWindow->getRequestedDisplayParams();
for (const auto& pair : gPathRendererNames) {
if (pair.second == stateValue.c_str()) {
if (params.fGrContextOptions.fGpuPathRenderers != pair.first) {
params.fGrContextOptions.fGpuPathRenderers = pair.first;
fWindow->setRequestedDisplayParams(params);
fWindow->inval();
this->updateTitle();
this->updateUIState();
}
break;
}
}
} else if (stateName.equals(kInstancedRenderingStateName)) {
DisplayParams params = fWindow->getRequestedDisplayParams();
bool value = !strcmp(stateValue.c_str(), kON);
if (params.fGrContextOptions.fEnableInstancedRendering != value) {
params.fGrContextOptions.fEnableInstancedRendering = value;
fWindow->setRequestedDisplayParams(params);
fWindow->inval();
this->updateTitle();
this->updateUIState();
}
} else if (stateName.equals(kSoftkeyStateName)) {
if (!stateValue.equals(kSoftkeyHint)) {
fCommands.onSoftkey(stateValue);
this->updateUIState(); // This is still needed to reset the value to kSoftkeyHint
}
} else if (stateName.equals(kRefreshStateName)) {
// This state is actually NOT in the UI state.
// We use this to allow Android to quickly set bool fRefresh.
fRefresh = stateValue.equals(kON);
} else {
SkDebugf("Unknown stateName: %s", stateName.c_str());
}
}
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
bool Viewer::onKey(sk_app::Window::Key key, sk_app::Window::InputState state, uint32_t modifiers) {
return fCommands.onKey(key, state, modifiers);
}
bool Viewer::onChar(SkUnichar c, uint32_t modifiers) {
if (fSlides[fCurrentSlide]->onChar(c)) {
fWindow->inval();
return true;
}
Integrate the ImGui library with viewer Code and docs are at: https://github.com/ocornut/imgui ImGui is an open source immediate mode GUI library that's lightweight and fairly simply to integrate. Widget functions return their state, and the library emits vertex and index data to render everything. It's got a huge set of built-in widgets and really robust layout control. For the initial integration, I had to fix up event handling in the viewer's app framework (to get mouse wheel and more keys, etc...). The new viewer 'Debug' window is toggled with the space bar. For this change, I've added one feature to that window: the slide picker. It's got a list of all slides, with filtering support, and the ability to click to switch slides. I also included the ImGui 'Demo' window (toggled with 'g'). This is nicely laid out, and includes examples of pretty much everything the library can do. It also serves as good documentation - find something that looks like what you want, and then go look at the corresponding code (all of it is in imgui_demo.cpp). I have other CLs with other features (like directly editing the primaries of the working color space), but I wanted to land this chunk first, then start adding more features. Other than adding new debugging features, there are few more outstanding work items: 1) Raster doesn't render the GUI correctly, due to non- invertible pos -> UV matrices. Florin is working on that. 2) Touch inputs aren't being routed yet, so the GUI isn't usable on Android yet. Might also be tough to work with, given the size. 3) ImGui has clipboard integration (that's why it wants the C, X, and V keys), but we need to wire it up to the OS' clipboard functions. 4) Draw commands can carry a void* payload to support drawing images (using whatever mechanism the engine has). I'd like to set that up (probably using SkImage*), which makes it really easy to add visualization of off-screen images in GMs, etc... BUG=skia: Change-Id: Iac2a63e37228d33141cb55b7e4d60bf11b7e9ae1 Reviewed-on: https://skia-review.googlesource.com/7702 Reviewed-by: Brian Salomon <bsalomon@google.com> Commit-Queue: Brian Osman <brianosman@google.com>
2017-02-10 18:36:16 +00:00
return fCommands.onChar(c, modifiers);
}