Rely

Things are coming together now. I've spent the last few days doing refactoring of the frontend library that connects to my API.

Library + API

I've become a big fan of separating frontend and backend into two applications. One of the big draws is the separation of concerns, it also allows me to use Python or Elixir on the backend and common nodejs frameworks on the frontend (Svelte for this app).

One of the challenges though is the need to wrap the API functionality in the nodejs app. Cursor has been a big help here. I've managed to create a neat wrapper library around my API to allow me access to the different backend capabilities.

There were a few more updates, like moving svelte to v5 but it's now operational and working fine. The only challenge was that my AppRunner deployment wasn't giving me the best output logs so I've switched the frontend to Heroku for now.

Next Steps

The core database entry stuff is working now. My next thing to work on is hooking up the database entries to the scheduling mechanism to schedule the worker tasks that will complete the healthchecks.

This part has been a little daunting to me as I'm not sure exactly the mechanism to make this part of the system resilient to failures.

I'll get it up and working over the next few days and then test.

One of the things that's been gnawing at me a bit recently is this project. While I'm ambitious an optimistic about Tempstore, I've also thought about how close to finish this project is.

I initially stopped working on Rely because I got scared of some of the technical challenges of building a highly available system on my own. There were a few things in both scheduling and multi-node parts of the app that made me think: "it might be too hard".

Devs vs Normies

The other thing is that Rely is focused more on developers and technical audience and so I guess I'm worried about my reputation if I get it wrong or build something that's not up to my own high standards.

Rely is also a bit more boring when compared to the flurry of AI hype that could propel Tempstore. Tempstore's audience is more regular people, and so I thought that the TAM might be higher, and therefore easier to get to scale.

Having time to mull over this has lead me to realise that there is in-fact ways to build in AI into Rely too. Something that wouldn't be too gimmicky but actually useful.

A New Hope

Spawning from a recent podcast talk, I'm not realising that if rely is so close to being functional, I could get it to a point that it's operating and then start to market. Even if it provides basic functionality, it could be enough to get some users while I build tempstore.

So I've allowed myself 1 week, maybe 2 max to bring the final parts together on Rely and get it out to the world.

Frankenstein

My marketing plan is now to get it operational, create a "Free Uptime Badge" type feature and then point paid ads at the thing and see if it jolts to life.

Let's see, I'll update folks on here with the progress.

I've been building Rely for a few months now but have not started marketing heavily yet.

What held me back you say?

I guess I thought maybe I should have a few more core features. I've been busy implementing HTTP / HTTPS checks on top of TCP checks (already there). It's taken longer than I though due to lack of focus.

Nevertheless this week I've decided to push harder on marketing, knowing that it might take some time to build a following and some users.

Nevertheless, I've built Rely, it should work if you need to monitor a tcp app somewhere on the internet.

Appreciate any feedback. I'll have the HTTP checks available end of the week.