Personal Software Is Becoming a Trend | Hacker News Reader

Personal Software Is Becoming a Trend

xuanwo.io

42 pointsxuanwo28 comments

I wrote a notes app in react native just for myself a few years ago (out of frustration after changing notes apps every year or so) that does /exactly/ what I want and it has been the most used app on my phone for a few years now.

It's offline-first but syncs reliably, uses the exact interpretation and display of markdown I like, searches and sorts the way I like, integrates with AI only in the ways I want it to (specific search capabilities, summarizing), uses on-device dictation via whisper so speech-to-text works when I'm away from a data connection, tracks location... I could go on and on since I add a new feature every month or so from a note (inside the app!) where I keep track of little things I wish it had.

But most importantly nothing ever gets added to it that I don't want... ever. No one else ever updates the terms of service, the UI layout, the retention period, the formatting, the shortcuts... there isn't some subscription I have to keep track of, or "pro" modes, or popups telling me about new features I should check out.

And since I have access to the backing db I can do all kinds of fun stuff with the data-- I've been playing way more with local LLMs lately with it.

I think it's healthy to remember that not everything has to be a startup or a public github repo.

And this is why people like Emacs or go for no desktop environment on Linux. Customize it to fit what you need and no further. Then you can just use it without learning where everything is again.

This is why I "rice" my desktop. I keep seeing a lot of videos and posts saying "just use the defaults" and "ricing is a tar pit". But its a great way to understand your tools and build it just for you.

Oh, that seems like a really good software.

It's like your describing me, except that I wrote my notes app in Flutter. I also added encryption.

It's also likely my most used app both on my phone and desktop. I absolutely love it.

Personal apps are the best!

I've built a very similar contraption, but without any libraries because I got annoyed at having to relearn a new kind of React after react class components got out of fashion and people started using hooks. I built it out of raw JavaScript and a raw python socket server.

How do you manage sync? Mine is a WebApp so I use indexedDB, and that's nice so that it works on my computer browser, but I don't want multiple tab synchronization on my phone and it's tricky to handle. I've built something that works, but I'd love to explore other designs.

You said React Native, so there might be a regular app sort of data store to disk.

I use watermelonDB for offline-first sync. It’s pretty wonderful.

I guess I'm of a certain vintage in programming these days. This is a common practice for me.

Often the IDE tooling for certain languages doesn't work for me. It's too bloated or slow. I write my own tools instead. They're fast and tailored to my workflow. Custom debuggers, analyzers, etc.

I often need to perform certain tasks... sometimes they don't warrant any kind of automation. However I'll often write snippets of code in a literate `org-mode` file as I work and write down what I was doing/why, etc. Occasionally these will get tangled into a proper script that becomes the automation.

I grew up in an era when computers only came with a manual and a bare bones operating system. Writing your own tools was part of the experience from that time which stuck with me.

Working in software professionally, the internal struggle I seem to have is separating "personal software" from "resume fodder." The line of thinking is generally: "Why create this one-off thing for myself when I can publicize it, get some contributors, and add it to my list of successful OSS projects." It's a terrible loop, to be honest.

a possible hack for that: write personal software to help you write the OSS project

I've been struggling with some sort of variant of this.

I'm a prolific FOSS contributor, but a few years ago I switched to a very time-intensive day job and we had our first kid, and my FOSS contributions have waned a fair bit as a result. I still need to do tech stuff outside work to have an extra creative hobby outlet, so I've been doing HW/SW DIY projects like [1] and [2] that don't put me into the critical path of anyone. They're fun, and I open source them, but I'm permanently stressed and a bit guilt-ridden that compared to my previous spare-time output they're not really of much use for anyone else. My "personal software" dies alone at home, whereas previously it was used by millions.

I'm working on a DIY toy robot now[3], and I'm trying to find a middle ground there by launching a new project with a library, tools and tester GUI to control serial servo motors by various vendors that the robot is the dogfood-test for. I'm hoping this is the solution: Build "personal software", but clearly pull out the parts that are usable for everyone into a module that you put in a little extra effort to make it more available for general consumption.

E.g. with the e-ink newspaper, I should look into making the display driver code I wrote into a lib as well I guess ...

1 = https://github.com/eikehein/hyelicht

2 = https://imgur.com/a/diy-automatic-e-ink-newspaper-using-rust...

3 = https://mero.ng/i/MHgQziFC.jpg

You're not alone in that. I also have a very demanding day job along with a few fairly serious community organizations that I help lead. Oh, and also three kids :) I suspect people in situations like this are less represented in the hacker community simply because of time bias. I have to constantly remind myself that some of the (genuinely amazing) things I see people doing are often in a vastly different context than myself. When I don't, I fall prey to the loop again.

Do you write scripts for you own use? I've usually been able to do that without thinking "I really should publish this".

Yes, I tend not to feel the need to apply the loop to sufficiently small tasks. To give a recent example, though, I use a text-based bookkeeping system for my personal finances. I've recently started work on an API that can sufficiently increase the possibility of automating the more mundane parts. Of course, out of habit, as I'm designing the API, I'm immediately making tweaks to make it more accessible outside of my own personal needs.

It's a culture I've subconsciously adopted, and I appreciate the author's intent to shrug it off.

Like a creative person you have to develop overly experimental bad things before you retire and blame others.

I can relate to this. I've never been able to code anything more advanced than, say, a simple html website, but over the last year I've used LLM's to develop a range of little software tools (and games) that are specific to my needs. Some of these fill a niche that either doesn't exist and some are just me slapping a GUI onto command line tools.

If we had simply followed the Unix philosophy to its full potential, we would have a lot more “personal” software that people could create for whatever niche purpose they need.

Or the emacs philosophy, or the lisp/smalltalk philosophy. Which ultimately means core utilities, tied together by the user on top of an hackable distribution.

in the eighties this was just called learning how to use BASIC

> This idea gained popularity with the rise of AI. With AI's assistance, creating software that meets individual needs is now easier than ever.

For any decent programmer, writing reusable small applications has always been a thing. Nothing new.

Yes, but pawning off the work on an infuriatingly forgetful and hurried junior who doesn't really listen to you is.

Before you needed someone else to have thought of your need, created a solution, made it available, and still be in the sweet spot where they have yet to monetize it. Even then, what telemetry is it sending?

Right now you can just ask for what you need and bring it into existence with a prompt. What a time to be alive.

Let’s hope software complexity doesn’t catch up to this trend.

I wrote a budgeting web app for myself. I love it when I realize there's a feature missing or implemented not quite right and I can immediately add/fix it. But my favorite part is that I never have to log in, because no user accounts (it's safely behind wireguard).

I can see a future where a lot of SaaS gets pushed down to APIs or low level “pipes” and we’re all just speaking / typing requests into a machine that creates what we need just in time. Software kind of unfurling in front of our feet as we walk in our desired direction.

Devs become like steam locomotive enthusiasts, gathering in abandoned software studios to work on their heritage projects and drink unfashionable lattes.

Currently making an invite only social network for myself and my friends.

I am loving the new Framework called Python HttpSimpleServer. It’s totally Bitcoin.

This blog post consist almost entirely of quotes from other people.

Reply on news.ycombinator.com