The difference between a side project and software you use

I fix phones. For a while I tracked every repair in a notebook — customer name, model, fault, what I replaced. It worked fine until it didn't. I'd lose track of which jobs were still waiting on parts, or what I'd charged for the same repair two months ago.

So I built FixLog. It's a React Native app backed by Supabase, and it does four things: records a device, logs the fault, tracks status from intake to collected, and remembers what each repair cost.

I've shipped side projects I was proud of and never opened again. FixLog is different, and the difference taught me more about building software than anything else I've made.

You can't avoid the rough edges

With a side project, you use it the way you built it. You take the happy path, because you're the one who designed it.

FixLog is different. It gets used when I'm standing at a counter with a customer waiting, or when I'm tired at the end of the day and just want to close out a job. Every extra tap is a real cost. Every screen that takes a second too long is a second I notice.

The first version had a form with eleven fields. I filled it in twice before I cut it to five. That decision didn't come from a design principle — it came from being annoyed.

What I cut

Things I removed, in order:

• Categories for repairs. I thought I needed to classify every job — screen, battery, charging port, water damage. In practice, typing what's wrong is faster than picking from a list, and I never once used the categories to search.
• A dashboard. Looked good in a screenshot. Told me nothing I didn't already know.
• Photos of each device. Useful in theory. In practice, I'm not going to photograph every phone that comes in, and a feature you don't use is just a screen you have to skip past.
• Notes formatting. Just a text box now. It was a rich text editor for about a week.

None of this felt like a feature cut at the time. It felt like removing a speed bump I kept hitting.

What that changed about my other work

CircleNet has far more features than FixLog. Some of them I'd defend. But building something I depend on changed how I think about the rest.

I ask a different question now: not "would this be useful?" but "would I actually use this on a bad day, in a hurry, when I'm not in the mood?" If the answer is no, it probably doesn't need to exist.

That's a much harder test, and it kills a lot more ideas.

The honest part

FixLog isn't polished. It has a bug where the status list doesn't refresh if you background the app and come back. I know about it. I haven't fixed it because it's a minor annoyance and I've been working on other things.

That's also the difference. A side project with a known bug feels unfinished. A tool you use with a known bug just feels like your tool.