Implot integration - introduction & acknowledgements

[pthom]

ImPlot is a fantastic graphing library. I worked on it in 2020, by providing an interactive emscripten demo for it.

Python bindings

Usage of implot with python

Its usage in ImGui Bundle, especially in Python is a joy to use; and the auto-completion works like a charm.

Usage with jupyter notebook

It can also be used inside jupyter notebook

Demo video (40 seconds)

Note: the notebook integration cannot rival the level of integration ipywidgets, but it can be useful in the case of applications for which a Gui is needed inside and outside of jupyter, or in the case of python application for which a later port to C++ is planned.

Technical notes about the python bindings

The generated python bindings follow closely the ImPlot API. See the python bindings stubs (i.e. declaration) inside implot.pyi, and compare it with the implot.h from which they were generated.

The bindings transcription adheres to the following principles:

• CamelCase to snake_case
• Enum names and member names are shortened (since all of ImPlot lies in the imgui_bundle.implot namespace): see for example AxisFlags_ in Python vs ImPlotAxistFlags_ in C++
• ImPlot functions which use a C-like data buffer are ported to python functions which use a numpy array. See for example PlotLine in C++ vs plot_line in python, and the autogenerated binding that does the template transcription for this function

ImPlot is compiled with IMPLOT_CUSTOM_NUMERIC_TYPES defined for all types, so that almost all numpy matrices can be handled.

Integration inside ImmApp:

ImmApp is a thin extension of HelloImGui that enables to easily initialize the ImGuiBundle addons that require additional setup at startup.

Note: using ImmApp is perfectly optional. It is possible to instantiate a standard ImGui app manually.

Automatic context creation

When using ImmApp, a context for ImPlot can be created/destroyed automatically, by filling an AddOnsParams
structure.

The code below shows the bare minimum to run an app with implot and immapp:

In C++:

#include "immapp/immapp.h"
#include "implot/implot.h"
#include <vector>
#include <cmath>


int main(int, char**)
{
    constexpr double pi = 3.1415926535897932384626433;
    std::vector<double> x, y1, y2;
    for (double _x = 0; _x < 4 * pi; _x += 0.01)
    {
        x.push_back(_x);
        y1.push_back(std::cos(_x));
        y2.push_back(std::sin(_x));
    }

    auto gui = [x,y1,y2]()
    {
        if (ImPlot::BeginPlot("Plot"))
        {
            ImPlot::PlotLine("y1", x.data(), y1.data(), x.size());
            ImPlot::PlotLine("y2", x.data(), y2.data(), x.size());
            ImPlot::EndPlot();
        }
    };

    HelloImGui::SimpleRunnerParams runnerParams { .guiFunction = gui, .windowSize = {600, 400} };
    ImmApp::AddOnsParams addons { .withImplot = true };
    ImmApp::Run(runnerParams, addons);

    return 0;
}

Run a C++ ImPlot app with minimal boilerplate code

The previous C++ example can be compiled standalone by simply adding this CMakeLists.txt from the _example_integration folder

(alternatively, HelloImGui provides a similar quickstart example, which can be used to compile an application where implot's code is provided manually)