Project Climb was a challenge I've set for myself: make a complete game, from scratch, using a new language & framework, in one month. A complete history of my thoughts during the challenge is available on my log: this post-mortem is basically a summary of it with videos.
This was my first time using Haxe, and my first time using an object-oriented language besides C#/Java/C++, so it was pretty disorienting at first. I wasted the first few days on making an entity-based physics engine when my game didn't need an entity-based physics engine. I removed that feature by the sixth day.
Tip: Don't make features you don't need when under a time constraint.
I spent the rest of the week exploring LDtk and making stupid dependency mistakes. By the end of the first week, I had pretty much nothing besides a simple (copy-paste) LDtk renderer and a controllable cube. The future looked bleak.
After removing most of the physics code from the previous week, I've began working on implementing tilemap-based physics & collisions. It took me most of the week to get that working, and it had a ton of bugs (there's actually a bug with knockback in the final release, but I made sure the enemies can't trigger it anymore. It was very annoying).
Tip: Don't make physics engines unless you hate yourself.
Besides the physics stuff, I've begun working on animations & gameplay that week. I've added the snail enemies (which didn't do anything yet at that point) during the last day.
During this week, I've added most of the mechanics of the final game, like the spellcasting logic and all Fire spells, taking damage and multiplayer.
It was a very productive week, and allowed me to add basically everything else relatively easily, but I didn't think that at the time, and thought that I'll have to really tone down my vision - especially since I hadn't programmed the teleport spell yet.
Tip: Don't panic.
During this week, I tried my best to add as many features as possible in as little time as possible. This was also the first time I started really hard-coding stuff, but it paid off.
While making the UI, I thought of adding a Kirby style, always present boss health indicator to force me to make a boss. It worked. In the last four days, I worked on Project Climb a bit too much, neglecting my university homework, to add things like keybindings, four spells and a boss.
Tip: Don't neglect university homework. It doesn't end well.
It was definitely an interesting challenge, and I'm very happy with the end result - I think Project Climb is one the best games I've ever made, even without content. I've also learned a lot from this. Anyway, now I'm back at working on Frogman Gaiden, so I'll have to leave Project Climb. For now.