Ask HN: What are you working on? (April 2025)

410 points by david927 ↗ HN
What are you working on? Any new ideas that you're thinking about?

1,245 comments

[ 3.1 ms ] story [ 599 ms ] thread
Currently working on HN Alerts — a simple free site I made to alert me (via email) to trending stories on Hacker News.

It sends me an email once a story hits a certain number of upvotes per minute, so it's useful for keeping track of breaking news.

https://hnalerts.com

I have a similar domain - https://hackernewsalerts.com - but it's for tracking replies to comments and posts you've made. It's on maintenance mode at the moment, couldn't gather as much interest as I'd hoped. Have open sourced it.
Is there any difference to the existing one that made you built another one?
Yep, it notifies you when you get comments on your HN posts. The existing one only tracks replies to comments.
I’m still working on these.

SaaS - I'm working on this mostly marketing that tech.. harder than it looks am I right? https://prfrmhq.com - see https://news.ycombinator.com/item?id=43538744 [Show HN: My SaaS for performance reviews setting goals and driving success]

- Shows I can use AI and I've integrated into AWS Bedrock

- Shows I can integrate with Stripe for payments

Consulting (Architecture, Strategy, Tech) - I'm working on getting my consultancy started. If anyone wants the kind of skills I offer here let’s talk https://architectfwd.com

Next SaaS - Starting a SaaS for managing core strategy and tech concepts. I created goals for it but I’m failing to kick the tyres

Last night I actually also started playing with firebase studio, though the app I prompted isn’t even doing save of the document properly. I figure can’t be me but will try again and work through the errors.

And playing drums, must get better

A reactive notebook with managed side effects for building backend/AI-engineering pipelines.

Reactivity can update the state of the notebook automatically, so you don't have to keep track of which cells to execute again. Side effects are managed to make it easier to reason about while maintaining reactivity and ability to interact with the outside world.

(comment deleted)
I just discovered a new meta-discipline, which most likely will become a new science.

I know, it sounds crazy.

In a month or so, I’ll be sharing some news.

I've recently added autobatching to my SFML fork (https://github.com/vittorioromeo/VRSFML/tree/bubble_idle). Drawing multiple objects that use the same RenderStates will now be automatically coalesced into a single draw call, for example:

for (int i = 0; i < 10000; ++i) renderWindow.draw(sf::Sprite{/* ... */});

Upstream SFML: - 10000 draw calls (!) - My fork: 1 draw call

This (opinionated) fork of SFML also supports many other changes:

- Modern OpenGL and first-class support for Emscripten - Batching system to render 500k+ objects in one draw call - New audio API supporting multiple simultaneous devices - Enhanced API safety at compile-time - Flexible design approach over strict OOP principles - Built-in SFML::ImGui module - Lightning fast compilation time - Minimal run-time debug mode overhead - Uses SDL3 instead of bespoke platform-dependent code

It is temporarily named VRSFML (https://github.com/vittorioromeo/VRSFML) until I officially release it.

You can read about the library and its design principles in this article: https://www.vittorioromeo.com/index/blog/vrsfml.html

You can read about the batching system in this article: https://www.vittorioromeo.com/index/blog/vrsfml2.html

You can find the source code here: https://github.com/vittorioromeo/VRSFML

You can try out the interactive demos online in your browser here: https://vittorioromeo.github.io/VRSFML_HTML5_Examples/

The target audience is mostly developers familiar with SFML who are looking for a library very similar in style but offering more power and flexibility. Upstream SFML remains more suitable for complete beginners.

I have used this fork to create and release my second commercial game, BubbleByte. It's open-source (https://github.com/vittorioromeo/VRSFML/tree/bubble_idle) and available now on Steam: https://store.steampowered.com/app/3499760/BubbleByte/

BubbleByte is a laid-back incremental game that mixes clicker, idle, automation, and a hint of tower defense, all inspired by my cat Byte’s fascination with soap bubbles.

A trailer is available here: https://www.youtube.com/watch?v=Db_zp66OHIU

The first ever SQL debugger – runs & visualizes your query step-by-step, every clause, condition, expression, incl. GROUP BY, aggregates / windows, DISTINCT (ON), subqueries (even correlated ones!), CTEs, you name it.

You can search for full or partial rows and see the whole query lineage – which intermediate rows from which CTEs/subqueries contributed to the result you're searching for.

Entirely offline & no usage of AI. Free in-browser version (using PGLite WASM), paid desktop version.

No website yet, here's a 5 minute showcase (skip to middle): https://www.loom.com/share/c03b57fa61fc4c509b1e2134e53b70dd

Is this postgres only? What an interesting idea!
For now, yes, but I'll start working on adding support for all other DBs (especially OLAP) as soon as possible. The geberal approach is the same, I just have to handle all the edge cases of the SQL dialects
Was thinking today... not a debugger but even a SQL progess bar, so I know that my add column will take say 7 hours in advance.
This seems like it could be extremely useful.
Thanks! Would you mind sharing what would be your use cases?

At my job, all of our business logic (4 KLOC of network topology algorithms) is written in a niche query language, which we have been migrating to PostgreSQL. When an inconsistency/error is found, tracking it can take days, manually commenting out parts of query and looking at the results.

Am not the person you asked, but feel that it could have good value for education and learning as well, besides debugging.
Cool! We're dealing with many complex CTEs and costly queries. Would be useful to have those visualized one by one.
What database are you using? I'd be happy to hear about your usecases and hopefully help you, shoot me an email (in profile)
This is awesome! I’m work with a team of analysts and data engineers who own a pretty big snowflake data warehouse. We write a ton of dbt models and have a range of sql skill levels on the team. This would be the perfect way to allow more junior devs to build their skills quickly and support more complex models.

I would recommend you target data warehouses like snowflake and bigquery where the query complexity and thus value prop for a tool like this is potentially much higher.

Thank you, nice to get some idea validation from folks in the industry. For sure data warehouses are the top priority on my TODO list, I picked PG first because that's what I'm familiar with.

I can ping you via email when the debugger is ready, if you're interested. My email is in my profile

This would be incredible to understand why some queries execute slow; most of the time it's one of the steps in between that takes 99% of the execution time at our company. Do you record the time each step takes?
You're onto the original idea I started out with! Unfortunately it's very difficult to correlate input SQL to an output query plan – but possible. It's definitely in future plans
Can you not use EXPLAIN ANALYZE to identify steps that had the highest compute time? I think most databases have some form of this.
This is a great command everyone should know. We once had a long running database query that was blocking a pipeline (code was written in a week and of course became integral to operations). Ran it, 15 minutes of thinking, added a new index on an now important column, and cut the run time down from almost 30 minutes to 5 seconds.
MSSQL has the execution plan thing that will tell you which steps are involved and how long they take.
this is very cool! Where can I follow you to see updates?
Finish it, shut up, take my money! This looks really good - make a website just to make it possible to sign up for updates.
Thanks for the motivation to finish this as soon as possible :) I'm working on a basic landing page with screenshots/videos and a "get notified" button right now – shoot me an (empty, if you want) email (in profile) and I'll ping you as soon as it's ready
https://dequery.io :) Added a little signup form (with possibility of providing additional feedback)
That was quick. I wanted to be quick too, but learned that I was too eager and missed that the are optional fields to pick from. I then re-submitted with the same email (will it go through or you have uniqueness validation?) providing optional stuff this time. Maybe play with positioning of the optional fields so they would be more apparent before submitting email address.
A TigerBeetle client for Haskell.

The smallest (in terms of system calls and code) event sourcing database I can make.

Being more present.

I'm working on a workflow automation tool that lets devs write workflows in simple yaml files, and then deploy them to the cloud _or_ on premise. Each workflow is a set of actions and a trigger that can transform data, make api calls, run AI models, or really anything (via docker!). Each step relies on the output of the last step, and the workflow framework is engineering to be declarative, testable, and versioned. Similar to GitHub actions, but for *anything*. Think webhook to slack, email to support ticket, nightly aws backup & restore, mirror a file each night, etc.
premises, not premise :-)
Thank you! Learned something new.
I've been working on mock: https://dhuan.github.io/mock/

the process of creating APIs for testing and automation should be as easy possible. the tools that exist nowadays for this purpose aren't good enough IMHO, which led me to build it.

No Pizza On Luna, a graphic novel about a future run by AIs who have discovered that the best way to get humans to do what they want is to present as patronizing, unctuous clowns. Http://egypt.urnash.com/npol/
Still working on https://theretowhere.com since I announced it to HN in February.

It's an website who's goal is to make it easier to find apartments/hotels/etc that fit your housing preferences (starting with places that are close to the people and things you care about). It's flagship feature is the ability to make heatmaps of cities based on your preferences.

Since February I've slowed down on feature development temporarily as I try and find a way to sustainably increase it's popularity and learn what's the most important thing to focus on next.

A language learning app for couples (https://couplingcafe.com). I wanted to learn my wife's native language, so I've been building this on my own for a long time and testing solutions! Just a few paying happy users. Cooking up a lot of ideas
Looks nice. I really tried but couldn't find it: Pricing please! :)
I’m working on extending Postgres to run on top of FoundationDB. The goal would be turning Postgres into a distributed, horizontally scalable database with automatic sharding and replication.

Hoping to share a first version of it soon. It’s been absolutely fascinating digging into Postgres internals!

Is this something like what TiDb does with MySQL compatibility? Sounds fascinating!
Yes, pretty similar although I don't think TiDB is actually built on top of MySQL, instead it just reimplements the protocol for compatibility. My project is actually an extension of Postgres, which hopefully means much better Postgres compatibility in the long run!

Planning on publishing to the repo here if you want to keep an eye on it: https://github.com/fabianlindfors/pgfdb

I am working on the https://moviemovie.club/about, it's a tiny website about film review.

It works like a run club, where you have to make a review first to see other people's reviews.

I am currently implementing watchlists, comments and a mural to make it feel a bit less lonely. Right now I like the UI but it feels to lonely.

This seems like it would only work if “reviews” would be something rare to come by. Like some forums where you have to contribute to be able to download attachments, or see higher level subforums.

But reviews are everywhere, good ones too so it will be a hard chicken egg problem to solve.

Just finished porting Boot.dev's backend learning path to TypeScript (used to only be available in Python/Go, now also Python/TS)

Official release is Cinco de Mayo, I'm very excited!

https://www.boot.dev

I'm working on a WASM clone of the original Castle Wolfenstein. I'm going to call it Castle WASMstein.
Please keep posting updates about this because if I could instantly fire up a game in my browser, I would definitely pay for that and play with it all day!
I'm working on a correlation matrix with Svelte 5.

It has hierarchical clustering, rolling correlation charts, a minimap, time series data detrending, and 2D matrix virtualization (to render only visible cells to the DOM).

It has up to 130K matrix cells and correlates up to 23.5M time series data points.

https://covary.xyz

Schematic and PCB design relating to Lighting and Control Systems for my main job. Schematics and PCB Design after hours as a contractor too, because I have a daughter now, my wife can't work, and life has become /very/ expensive in Sydney.

What I'd love to be working on: Try to initiate a high voltage arc through the air to a target device, and modulate it to send "Data over Lightning", like Alyx does in Half-Life 2. It won't work the way it does in the game, but I'd it's an idea I've had for a long time and I'd love to prototype it some day.

Iterating on my geography site: https://geolede.com

Next feature is search.

This is a dope idea, nice job!

I would love to see some UI/UX improvements like split view where the map is on the left and the news reading/scrolling happens on the right reading pane instead of on the bottom while horizontally scrolling.

You could even use AI/LLM's to summarize the most important news from each country etc.

A TUI for categorizing financial transactions into valid plain text accounting records.
I'm working on a distributed object storage system to be the backing store behind my website (https://scmscx.com). It currently uses back blaze b2 which is good and cheap but I thought it would be fun to roll my own.