Saucer Engine

Saucer engine is a Open Source 2D Game engine I made during 2 months with C++/OpenGL/GLFW/etc. There is a builtin editor and support for Lua scripting. The name *Saucer* came simply from my name ("Pires" literally means Saucer) , but it came together with the proposal to "be lightweight" hum? *ba dum tsss*

Github Page

Development

Ohhh the naive old me...

The plan was to make a "simple retro-like Chess game" (motivated by the popular Netflix show with the pretty girl you know about). But I also always had wanted to learn low-level graphics stuff with C++... so it was settled: I would create this chess game with a game engine I made myself.

Fast forward: I made it!!! Here is the post I wrote about it.

Even though my first expectations was to complete everything in 1 month (naive old me!!!), now I don't think 2 months was too much. My objective wasn't to complete only the simple retro-like Chess game. This was postponed until the engine was complete, and this one was the task that took 95% of the time. It was a little further than my first concepts of what was a complete engine, and still, even then I see that it is not quite complete yet.

Feature-wise, I'm pretty satisfied with the result:

Texture(PNG) and custom font(TTF) rendering, with support to write own GLSL shaders.
Audio sample(WAV) & stream(OGG) player.
Embedded Lua scripting & runtime profiling.
Physics engine (Box2D) integrated
Assets packaging for release build (Using zlib)
`Scene` objects are saved as git-friendly YAML files.
Very lightweight (release export is around 4MB, aside assets).
Builtin editor for scene editing (There is no way I'd have decided to make it without ImGui)

If I ever want to create another 2D simple game or maybe some general application with graphics, I'd even reuse it again. But, to be honest, it was TOO²³²³³ much work for something you could achieve with other tools.

Now, I'm not saying I regret doing it. I learned a lot about C++ development and stuffs I was targeting to learn. These C++ metaprogramming templates Hell ? Scare me no more. Macros can be useful with the right amount of care taken. And in the end of the day I can think of many ways to optimize this code but hey, I don't even need it because C++ is lighting speedy boy.

Benjamin Franklin: 'Those who gave up performance for easiness deserve neither'

Documentation

Simply put: there isn't any.

Sorry mom. It was already too much of unpaid work to I even think about documentating it for other users. Tried Doxygen but ehh... I decided I should expose the Lua API instead the C++. That's why I'm saying the engine is not quite complete yet.

The most important thing I have to say is that you should run it with a debug environment (like GDB). There is too many ways the engine can crash that I haven't made a proper runtime guard yet.

Mr. Turner: 'This is where I'd put my documentation if I had one

Building

Figure it yourself, good luck.

Just kidding. This was probably one of the most painful tasks. Until I had to move to Windows from Linux, it was all a customized Makefile. It was nothing fantastic but did the job. Unfortunately, Windows doesn't simply builds from a Makefile. So I took some time to learn CMake.

There is a CMake file. I pray that you can simply build it. As I'm writing this page, it works on Windows VSCode + MSVC 2019 with the famous CMake plugin out there.

I very appreciate the effort of the CMake team. To be honest, I don't even know if it could be easier. But there isn't a single one easy `Getting started` tutorial. All the good efforts out there from outside the official documentation doesn't cover everything you will need to know. And the official docs is just TOO much, when all you want is some simple steps. My conclusion is that if I ever had to hire a team of at least 3 C++ programmers, I'd hire someone entirely dedicated to compile & link stuffs.

Conclusions

These are my personal overall conclusions and suggestions on creating a custom general-purpose game engine.

C++ is just the right amount of hard and unlimited amount of power.
You can do bollocks and C++ will still be faster than other solutions. It may even be worth to actually create your critical-performance game in C++ than try some hacks using a general-purpose game engine. That's why I really appreciate engines (like Godot!) that stands in the sweet spot (helps you embed low-level code while having high-level features).
Spend some reasonably time on a proper building & debugging environment. These aren't things you can show off but are crucial.
Making games with C++ is a totally different task from making a general-purpose game engine. To further optimize a custom game is way easier than try to optimize it in a generic way.
Embedding Lua was way easier than expected. I just don't like it's syntax!! But it still is a very useful tool.
I wasn't expecting to learn things like the bare minimum of audio programming (it isn't just "play me this file please" you know... still, kinda satisfying knowledge)
Accept that some features will not be perfect. The assets packaging, the last feature I made, was not in any way hack-proof, but I convinced myself that barely no game out there is?
In the end, I wasn't any motivated to play a single game of Chess (!!!). I wrote more about it here.

Acknowledgments

LearnOpenGL
Aside from OpenGL, every other integrated library has a proper documentation when necessary. Check them on the Github Page