2023 ends and now we are going again for a whole other year.
But before that, it is important to reflect upon what happened during the whole year and to check on what has been done, discovered, and fucked up.
What has been done
2023 has been quite an important year for me, as this is the moment cubic-server started to gain some traction and got some actual work done on it.
This is also the very first year where I had to live by myself in another country for the very first time, and that experience has been quite the important one (The Netherlands has been one of the best country I could have hoped for to live in and the whole time there might actually need its own blog post all by itself).
I also got a new job as a teacher assistant (Or at least an equivalent of it) at my school which was for me one of the best work experience I ever had (This might also need its own blog post as I am still working there and might be still working there until August 2024, and so many things have already happened that it might warrant its own retrospective).
I also participated a bit in AdventOfCode2023 before time and work got after me and I couldn’t participate much, it was overall very interesting as I took the time to compete with collegues, friends, and students alike on who could beat the new days as they came out with either the worst looking code, or just the fastest/most clever solution of all.
What I have discovered
Zig
Zig is a low level language that aims to be a simple language with no hidden control flow, no hidden memory allocations, and no no preprocessor nor macros (All of this was taken directly from within the main page of the project).
This project pretty much wants to be a successor to C, which in my opinion rust failed to be and continues to fail, by just being as simple as possible while also providing really advanced metaprogramming capabilities, especially with its comptime features which are freaking great.
Hare
Hare is also something that I took a bit an interest in this year as it was also a simple system programming language with some extra features from C that made quite alluring. I also had the occasion to meet Drew Devault, the main maintainer of hare during a test talk about hare and its use in making a simple microkernel based OS he presented at Technologia Incognita to prepare for FOSDEM.
The Dutch
From the start of September 2022 up until August 2023 I lived in the Netherlands in the not so small city of Eindhoven.
It is not really a secret but I am French, and I was a little scared at the idea of living a whole year in a country that I did not know the language of (Even though I am currently working on getting better at it), it actually was completely fine as everyone speaks English there, even older people (I encountered only one person in the Netherlands that couldn’t speak English, and he also couldn’t speak Dutch).
That is a very sharp contrast from France where people will literally curse you for not speaking French to the perfection. In the Netherlands even if you start speaking Dutch people will realize you are not Dutch and will instantly start speaking to you in English (Which is really frustrating if you are trying to learn the language).
For now I will stop here about the Dutch and the Netherlands, because I could go into much more details which would need its own post by itself (I might do that one day ;) ).
What I fucked up
cubic-server
Let’s start by saying that I am really happy to have started cubic-server, it was a wonderful idea even though the Minecraft protocol is an absolute pain, because it made me work constantly on one single idea/project and made me step up my programming and management skills.
Though now a lot of problems arose from that project and the first big one is the architecture.
The architecture
We wanted to do more, to think better, to think ahead, that was pretty much an error, and now we pay for it by very long compilation times and chats that can be accessed from a player by getting its dimension, then the world, then the worldgroup, and then you could get the chat object (Which is also kind of overingineered quite a bit). Needless to say a lot of that architecture could be much more simplified/
Out of Scope
When we started that project it was all for just having something for our end of study project, and for that we needed to prove that there was a lot of work to be done, and so we did, maybe a bit too much. So now we have a whole generation system and survival aspects working while the server could have just been a simple modular base to add more complex parts on top of it.
All those features added on top of each other mixed with the needlessly complicated architecture make for one hell of a time when you want to dev anything for it.
And the main question remains, which users are we targeting?
Survival users? We will never have parity with vanilla, so that’s a big no-no. Creative players? Well it is possible, but the sheer amount of blocks and behaviour of them to handle on newer versions of the game make the task very complicated. Redstone players? We don’t even have a working redstone engine right now, and we have no idea if it will be hard, easy, long, or anything on it. It’s just being worked on as we go and we hope it will work at some point in time. Minigame players? Depending on which minigame it is it might just be our best bet, as there is little interactivity with the map and most of what is happening is in general already heavily scripted to fit the minigame’s rules.
So the best bet as of right now for the target audience would be to replace small Minecraft server’s instances and replace them with a highly modular cubic-server which can then be modified to fit that specific’s minigame’s need.
The language
C++ is great… until it isn’t. Let me explain more on this:
C++ is basically just a superset of C, and the quantity of features is not really a problem, the problem is which features are better than others, and when to use them (This has been a long lasting problem in our team as not everyone has the same skill level with the language and many PRs have been delayed again and again because it was not exactly what we wanted at the time).
C++ also has a compilation time problem, it will take forever to parse files or to generate the bytecode the moment you use some templated library (Especially boost omg), and if one character changes in one header file then you will have to recompile all the cpp files that depended on that hpp (Try to change one thing in Server.hpp and be ready to recompile the whole project).
C++ headers make it really hard to do proper incremental compilation and are just going to be in your way 90% of the time (Headers are also incredibly frustrating as if you make a single mistake in one of them, not only are you going to have to rebuild a good chunk of the project but you will also have to understand whatever cryptic error the compiler through at you on the terminal).
The solution
Honestly… A rewrite. Not in C++ for sure, but this project will need a rewrite. When? Well… you’ll see :p (It is already planned in time and we are starting to have a roadmap for it which should be finished soon:tm:).
Afterwords
2023 was clearly one of the most important year of my life and I hope things get better from now on.
I have most likely forgotten a lot of things in this blog post, but that will only make me more eager to come back with more posts this time (Promise I will actually try to write on here more than before :D).