Dungeon Quest Devlog

brett · March 5, 2025, 8:59pm

This thread is a devlog for a minimalist dungeon crawling RPG where the events are procedurally generated with each run.

It's actively in development, but you can check out the latest build here:

Original Description for Posterity's Sake

I originally set out to take on a challenge of sorts where I made a bunch of small games in a short time window, but I quickly realized it wasn't realistic. But I've saved the description for context with the first few posts.

This devlog is something a little different than usual! It's not for one game but rather for three games. Three games that I'm going to make over the next six months. Since they'll be relatively small, I thought why not make one thread instead of many?

An idea I've been chewing on as a hobbyist game developer is how do I get good at making games? And how do I do that with my limited free time? And how do I not start and abandon so many projects?

Inspired by Bennet Foddy (Season 1 Zipper developer) and Douglas Wilson's talk at GDC in 2018 about the courses they teach where students make one game a week, I wanted to design something similar for myself that's realistic yet holds me accountable. I've got two young kids and a full-time job. So my time is limited. I don't think I could sustain doing one game a week without burning myself out.

I also love the One Game a Month project, which has people making one game a month. Game jams also have a similar spirit but are often during a much shorter time window.

So I've come up with what I'm calling Game Dev Jump-Start: a malleable structure to help someone dive a bit deeper into their game dev practice.

Approach

For my first GDJS, I'm going to do this pattern of increasing size of each game:

March 2025: 1 game in 31 days
April, May 2025: 1 game in 60 days
June, July, & August 2025: 1 game in 90 days

So 3 games in 6 months! One could modify this accordingly, doing more games per week or less. But the intent is to finish games and not work on them forever and ever. It also forces learning scope and cutting things.

The third and final game, with 3 months of time to build it, should hopefully be my largest, most complex game I've ever made.

Rules

I must publish a game on brettchalupa.itch.io at the end of each cycle, even if it's buggy or broken; prototypes are okay! But ideally they'll be polished enough to be playable
I will not be using any AI for any part of the game development process, including no AI coding assistance
I will make all of the games open source to share what I'm learning
Sequels are okay if it is different enough from the original to warrant it and not just a direct continuation, gaming the system!

Goals

Part of this exercise, which I can imagine repeating throughout my life, is to have specific goals or outcomes I want from these 6 months.

Get really comfortable with Lua
Get really comfortable with the Playdate SDK
Improve as a pixel artist making 1-bit art
Make multiple original, non-trivial games that aren't just clones for learning purposes
Break down RPGs and roguelikes into smaller chunks and learn how to make them, as they're the games I'm primarily interested in making; although I have other ideas I'm interested in exploring (but maybe not for this GDJS)

Games

I'll share updates as new posts in this thread and then keep this game list up to date as I finish them, as a sort of index.

Game 1: Dungeon Quest (31 days)

~Made from March 1 to March 31,~ Dungeon Quest is an RPG for busy people. It distills down the essence of JRPGs and roguelikes into a procedurally generated quest with no fluff.

Actively working on this game!

brett · March 5, 2025, 9:09pm

I'm 5 days into Dungeon Quest, my attempt at making a minimalist JRPG with a little bit of procedural generation thrown in. I want to explore if it's possible to make a game like Dragon Quest or Shiren the Wanderer in 15 days, distilled, without any map exploration. This might not work out, it might not be fun to play. But I'll learn a lot.

There are no sprites yet, but the basic flow of the game is working:

You "Descend" into the dungeon, starting the game. Your goal is to beat the boss on the eighth floor. The enemies and events on each floor are procedurally generated, trying to mimic a roguelike feel but without map exploration. You proceed from event to event, needing to be strategic about item usage and combat.

I've got attacking and continuing working. The other actions will come next, they'll be a fair amount simpler than combat.

I've coded up about 4 enemies so far, a mid-boss, and the final boss. It'll need to be expanded upon, but I've got a good sense of the scope of this thing.

It doesn't feel like a game yet, but it's close. I keep wondering if it'll be fun for people or if it just won't click together. I'm hoping between item usage, a Scavenger who sells you things, various enemies, and the procgen that it'll be engaging.

If it isn't, well, then, I guess walking around a map and grinding and hanging out can't be discounted.

brett · March 6, 2025, 7:03pm

I added leveling up and experience points today. When you level up, your HP is restored but your fullness is not. Your strength also increases.

dungeon-quest-early-battle

The game is inching closer to being fun/interesting. I'm going to add a bunch of new enemies now and have fullness or hp running out mean game over.

Something a little bit interesting is that since all of the events and enemies are just Lua tables, adding new ones is very fast and easy. Not having sprites (yet) lets me mess around with the systems and combat and numbers without spending a bunch of time on the art.

Spellthorn · March 6, 2025, 8:51pm

First, I think your game dev challenge is a great idea and excited to see your progress. I'm curious why you decided to not to use AI code, and how has that caused you to be worse/better.

With your updates so far your game is coming along nicely, really liking the simplified layout, but also the little touches to the little cubes moving just adds a special touch.

It would also be cool if you updated us on your coding process, like code examples, your thoughts process etc, as I feel that would help a lot of other developers with improving as well.

I'm excited to see where this game goes and your future updates, keep up the great work

brett · March 14, 2025, 2:16am

Thanks for your support! Will definitely share learnings and takeaways.

I didn't use AI to assist with coding for years, I resisted it and felt conflicted about it. At a new job I started recently, my colleagues use it (and rely upon it), so I thought I'd give it a try. I was impressed with how quickly and helpful it seemed. They use the Cursor editor, which I guess helps learn your codebase and such and be even more useful than just GitHub Copilot.

But after about 4 months I decided to turn it off and stop using the AI in my editor or to help with coding. I noticed it was causing a few things to happen:

It would interrupt my thinking with its suggestions, which is the biggest problem. I'd lose my train of thought processing the suggestion AI made.
It was more often than not just wrong enough to actually make it take longer than it would have otherwise for me to do it.
AI is good at writing tests to cover code, which makes sense, but I noticed it was actually creating bad habits where I actually prefer to use tests to design my code.
I like coding, it's fun for me. Using AI was taking away some of that critical thought and learning that I cherish.

I also have some conflicting thoughts/concerns about the ethics of it. The code its trained on and outputs often don't respect or share the license of that code. This is all a gray area right now for sure.

No judgement at all for people who use the AI for coding, I think it's a personal choice. But it wasn't craft-changing for me, at least at this stage.

Here's a super simple juice/polish tip for games. To get that smooth animation of the active cursor, you just pass it into math.sin or math.cos with a value that's changing regularly, like playdate.getCurrentTimeMilliseconds()`. It'll cause that value to loop, with some modifiers, and look quite nice.

For example, here's the code I use for the menu cursor:

gfx.fillRect(math.sin(playdate.getCurrentTimeMilliseconds() / 140) + 10, menuYPos(currentOption) + 6, 8, 8, 0)

Dungeon Quest Update

I've two 2 days left for the first game! I've continued to work on Dungeon Quest daily. I've been primarily focused on expanding the combat, enemies, and menu options. I've got a handful of enemies, proc gen for the floor events, critical hits and misses, rest events, a mid-boss, a final boss, and player levels.

I don't think I'll be able to get art added in time. So I think this prototype will ultimately just be a text-based RPG. With my two remaining days, I want to try to make it as much of a fun experience as possible. I definitely want to revisit the idea with some graphics and new systems. I think there's some potential here I didn't quiet get to.

Spellthorn · March 14, 2025, 11:44am

Very valid points I used AI a lot to make my game and you're right I found myself a lot of times instead of taking the code it gave me and then editing it to my liking I would just tell it to edit it to my liking instead, and then copy paste it in. It seemed like I became more of the idealistic roll in the making process for certain things like. I want this and it's like yea sure here you go and alright well I want this now too.

I will also say having the knowledge is definitely important when using AI and if you haven't been coding it long enough to learn that, AI can most certainly make it harder. There were a few parts where it would tell me stuff that didn't even exist and luckily I knew it didn't exist just because I knew how to do it but on things I didn't know yep I went down that

It's cool to hear other people's perspective on it as I personally feel it's a tool/guide to make you better but also see how it can quickly become it did most of it for you and did you even make this or did AI becomes very gray area quickly.

As for your game I think text based is completely fine already looking nicely and great tip on the cursor. Those little details can make even just text based feel unique

brett · March 16, 2025, 3:48am

Today, in theory, is supposed to be the end of and release of Game 1. But I've decided to change the timeline since I bit off more than I could chew in 15 days, I want to keep working on the game, and my life with two young ones and a full-time job is a bit too chaotic to consistently finish small games. Some weeks, like this past week, just left me with no energy and time for games aside from some small tweaks/improvements.

So for Dungeon Quest I'm going to plan to release the prototype publicly on March 31. Then I'll either do a two month game or two one month games after. Not sure!

But I don't want to not release anything today, so I've prepared and released version 0.2 of the game, which anyone following can download here (it's password protected with slime in case that's lost in the URL embed:

I've added talking, inspecting, and expanded the amount of enemies. There's leveling, but it's not balanced at all.

Some recent screenshots:

And finally, here's a little code snippet. It's a Ruby script (I know Lua would make more sense but I know Ruby much better ) that compiles and uploads builds to itch.io. It uses some additional metadata I've added to source/pdxinfo (hopefully that isn't a bad thing to do!):

release script:

#!/usr/bin/env ruby

# Builds the game for release and uploads to itch.io

pdxinfo = File.read("source/pdxinfo").split("\n").map { |l| l.split("=") }.to_h

fileName = pdxinfo["fileName"]
version = pdxinfo["version"]
itchUser = pdxinfo["itchUser"]
itchPath = pdxinfo["itchPath"]

system("rm -rf #{fileName}")
system("pdc -s source #{fileName}")
system("zip -r #{fileName}.zip #{fileName}")

puts "Pushing #{version} to itch:"
system("butler push --userversion=#{version} #{fileName}.zip #{itchUser}/#{itchPath}:playdate")

I put that in release, and then chmod +x release once, and then run it after I bump the build and version. I'll expand it to do the bumping interactively or with CLI args, but for now, this makes my life a little bit easier! And it'll come in handy for future games too.

Spellthorn · March 17, 2025, 12:41am

Great job on the game so far. It's a lot of fun. I died on my very first enemy apparently talking to a Wom-bat is a bad idea

Not sure if there is a way to get back fullness? Because I died on floor 4 because ran out of it.

I would also say everything triggers when I first click the encounter like level up health gain etc, but I feel it shouldn't until I read it other than that its a fun game and I'm glad you're taking more time on it as it is a fun game and will give you time to add more to it

brett · April 1, 2025, 12:43am

Adding food is my next major feature to implement. Thanks for checking it out and the feedback!

I haven't been able to work on the game at all over the last two weeks. My whole household was sick, and I've been swamped at work. It's been a very humbling experience to have this ambitious multi-project effort and drop the ball out of the gate. It's helped me realize that my capacity is not consistent and is quite limited at this time in my life. Even if I can dream up projects and ideas, if it's not realistic, then it's just setting myself up for failure.

So after spending some time away from the Game-Dev Jumpstart and even Dungeon Quest, I've decided I'm going to just abandon the whole timeline effort and instead just work on Dungeon Quest and the Playdate book as time allows and not stress out about it much. I very much want to revisit the Jumpstart idea in the future though.

But for now, just going to sustain myself!

Given that, I'm going to transition this thread into just being specifically a devlog for Dungeon Quest! Hoping to be able to dig back into the project soon.