We have plenty of posts on DEV announcing new tools, libraries, agents, and frameworks.
I want to hear about the opposite.
What did you build tha...
For further actions, you may consider blocking this person and/or reporting abuse
Don't know if this qualifies as useless, but I use it daily to track Egg collection and Chicken flock from my Indigenous chicken, I also record observations like weather, strange bird behaviour, bird poop, etc etc.
I would say this is technically impressive, but definitely not useless :)
you identified a real problem, built around your own workflow, and use it every day. That is more validation than many products ever get!
Recording weather, behavior, and health observations alongside egg collection could reveal some fascinating patterns over time. I am interested, have you discovered any correlations?
Thank You Ravi. As farmers, we do NOT keep records, we rely on our memory, hence the maYAI project. The average age of a farmer is 56yrs, relying on memory is a bit tricky, so I built this for me, also because the chicken, though I raise them, are for the entire family so, I need to keep the records.
Yes, there are very many correlations, a single one that keeps repeating is, PS I am NOT recording the feeds yet, but I have noticed that if I give them kitchen food scraps, some weeds, sweet potato leaves, etc etc, majority of the hens lay more eggs, i.e I get more eggs if I give the chicken "natural" foods vs the commercial feeds we give them daily.
I have a whole series of them - take your pick! (I think my favourite is the pure HTML animation - if it still works!) -> dev.to/grahamthedev/series/12287
Wow, that looks like one impressive series to follow up :) I will try to have a look when i get some time. I think i might find some motivation there. Thanks for sharing Graham.
no offence Graham, I went through your series. Useless (in terms of tech stack) I agree, but also it does showcase some real application and use-cases, so there is use to it as well. What made you discontinue that series ?
Got bored ? or just moved on from HTML ?
Literally just added one today (the snake with browser windows one, no 15), not discontinued just that I have been busy and not had much chance to write!
sorry for jumping to conclusion like that. I did not see that. Just had a look. I will be looking forward to more :)
Thissss github.com/sylwia-lask/text-goes-boom 😁
It also has live demo in docs.
I opened 'Text Goes Boom', and this is a perfect answer to the post: completely unnecessary, technically impressive, and strangely satisfying 😄
I especially like that the particles do not merely scatter; I tried my very best but they return to their original positions, ready to be destroyed again.
Thanks for sharing Sylwia
Hi Sylwia, well done! I checked your project, and the UI looks very nice.
I actually love it. It's so meditative.
I got a new 1 for ya. For no reason whatsoever, I decided to make Dwarven Stronghold, basically Dwarf Fortress, but entirely built in Rust... 200k LOC and counting, with strict zero-allocation and over 2600 unit tests. Not particularly useless per say, but given I havent booted it since it was just 10k LOC, pray for me, I hope it actually runs 😭
200k lines of Rust, zero allocation, and 2,600 unit tests : is either extraordinary engineering or the most elaborate suspense story ever written 😭
Please boot it. I’m emotionally invested now. Even if it summons a segmentation fault demon.
My thought was, DF was built in C++, it runs bad... But it works... So if I write it in Rust and interconnect the systems a little smarter, so it's modularized (think merkle root mixed with hub and spokes design), I should be able to get the game to run better than DF and with the Wikis Generator, it'll actually be well documented. Then instead of spending 20 years developing it feature at a time, I set up a page where people drop their requests and a background process runs the AI task to implement the feature and patch it to the canary branch (Beta Channel), otherwise offer a paid system, where you can request custom features and have it bypass the clearance stage and get patched straight to yours, if the system detects it's worth it, it implements it to canary too, but if you want dwarves with green beards then by golly you shall have dwarves with green beards within the hour.
300k LOC, 8192 Unit Tests. Still not started, but it's now at the linking stage
I'm excited too, though, I wanna leave it a surprise for myself too, I recently discovered Qoder, so I've been using it and letting it just go ham at it with Ultimate and with Qwen 3.8 Max, to see if AI is finally at a point where it can build in-depth simulators. So Imma let it keep building and expanding, steering it where necessary, so it gets to a point where when I boot it up, it's better than DF. Cuz I added 3 mandates to it that DF doesnt have. FPS mode, where you can take control of a dwarf in first-person. 2, ElevenLabs and a LLM for taking the dwarves' states and create dialog for interactions, so you click a dwarf, you start hearing their thoughts in realtime, you want their history, go to their lore tab and click play and listen away to the 300 year long life-story of a dwarf with half an ear and a bent nose. (atleast that's the goal)
BrowserTest. I built it yesterday because I wanted to test a feature and it was annoying filling in the fields everytime, so it records my actions based on the AOM and executes them. Also added a stripped down MCP-Lite to it, so the agent can record a session too. It stores the page's AOM and the chain, so afterwards I can wire it as a UI test, or automation script where you can loop it, add variables, etc.
Thank you for sharing :)
seems like you are building something in line to playwright. So it started off from filling the fields automatically, to now adding MCP to it and making it agent friendly.
Do you see this as something others would need ? or not at all ?
Are you there yet, I mean the phase :“the world needs this product.”, or you think that will never happen?
Was more just to solve a problem I had. Namely lazy testing UI functionality. Solutions already exist that rely on the DOM, not really for the AOM, so it's a bit less breakable than those. The MCP addition is also, given that MCP-Lite saves on token drain, it acts as a shim into your current browser, so it can interact directly.
This 1 I imagine wont ever see the light of day, it's just meant as an internal tool to speed up my day-to-day. For anything outside my job, I'd sooner use V.A.L.I.D. and not need to record anything, just have the fuzzer handle it.
Fair enough. Not every tool needs to “see the light of day” or become a product. If it removes a repetitive annoyance from your own workflow, it has already done its job. So I would not consider this to be useless ;) sounds more useful to me.
Pardon my ignorance, I’m curious about V.A.L.I.D., though. Is that your own tool ?
V.A.L.I.D. Have a look. I used it to build Doccit (a 500k LOC Autonomous Accounting Suite), it generated 82% of the codebase for me, including MCP, Unit Tests, DALs BOs, I literally just had to define the constraints in the DTO (ValidObjects) and the UI markdown. Reason why I'd rather use it, is because of the V.A.V.I.D. HUD and Fuzzer testing, so it can hammer every entry, every button, every dropdown and search rapidly to see if anything breaks. At 3600 mutations per second (with multi-parameter rule-checking and cross-contamination). It also has a simple script creation mode, where you just create a workflow based on the ValidObject and it runs it, so no need for playwright. But go give it a try, once you get the hang of using it, it makes life much easier
So back when I was just starting to learn web development, I only know JavaScript, HTML, CSS and I was inspired by a laboratory challenge from freeCodeCamp.
What if I built a past, present, future tarot reading web app?
At first, this wasn't my thinking. I was just curious so I searched up the web and I found some sites that does it but it doesn't capture what I wanted (I'm too picky). So I decided to make it myself, Veil and Visions
Oh boy, it was hell to maintain. Comments everywhere and all logic lives in one JavaScript file. Multiple bugs found after deploying it.
Was it worth it? Absolutely!!!
It taught me things on a much deeper level fundamentally and I have grown much more since then. I still open the site from time to time whenever I feel bored. It's also a good reminder that I considered this one of my milestones in my software engineering journey. I will sure as hell laugh it off whenever I go back to this site after 5 years (hopefully by then, I'll be happier and also become a reliable senior software engineer or manager, fingers crossed)
I actually opened Veil and Visions after reading your comment. The card artwork and the whole shuffle draw sequence are genuinely charming. It feels more like a complete little experience than an early JS exercise, so I’m not sure it even qualifies as useless!
Unfortunately, the cards did not reveal whether you will become a senior engineer or a manager, but I’m rooting for you either way!
Thanks for sharing Elmar :)
That makes a lot of sense! Playing around without client constraints or deadlines really gives you room to over-engineer and break things on purpose.
Also, checked out your Persona project—that’s a super interesting stack! Moving from a basic assistant to a full roleplaying assistant sounds like a fun rabbit hole to go down. Thanks for sharing!
I have so many of them. I have created lot of projects (pre-AI era), painstakingly a lot, which in the current time would take AI perhaps 10minutes to make. When i was young I did a lot of "the world needs this product" because if my ignorance. I really did not know what the world needed, I only knew i want the world to need what i make. And what i made was not something the world wanted then and now. Over time, it made me realize, anything that you make out of necessity has served its purpose when it made your life easier. After that it has no purpose, you can just put it out there, and hopefully few years down the line someone will find it useful. Thats all there is to it.
Anyway my most impressive project was building my own ray-tracing library in C, it was amazing at that time, in terms of performance and had its own openGL optimizations. But sadly i was the only user to ever use it. That project will always be bitter-sweet for me.
Thanks for sharing Ryan. I feel like i am still in my "the world needs this product" phase. The whole reason for me joining DEV was to showcase that. But I am grasping the reality slowly, that, possibly not the right approach about it. I will consider your words for sure.
Ray-tracing sounds like a heavy project (pre-AI), sometimes the hobby project we often are proud of, is often the one no one cares about. I hope with time the bitter-sweet just becomes sweet.
exactly... thanks for this good blog
Love this topic! Do you find that building these 'useless' side projects actually helps you learn architecture concepts faster than working on real-world production code?
Definitely, but in a different way. Side projects let me experiment quickly, over-engineer freely (with AI), and explore architectural ideas without deadlines or constraints or some obscure product requirements from a client. Production code teaches the harder lessons: changing requirements, collaboration, maintenance, and tradeoffs.
One such side project I did was this : github.com/ravipurohit1991/Persona...
created to learn the stt, tts, and an llm to be the bridge between. But eventually, over engineering, use of AI, ideas after ideas exploration, made this into a poor mans roleplaying assistant from just assistant. But I still like the tech stack, and choices made in this project. Learnt a lot for sure.
Interesting
thanks for share
🤔