I’ve been writing monthly roadmaps for Tanagram development in private for a few months. For 2022 I’ve decided to publish them, in part because I believe doing so will increase accountability for progress, and also because writing these for an audience will force me to be more precise with what I have in mind and enable clearer did-I-or-didn’t-I comparisons.

Tanagram remains a nights-and-weekends project. In 2021, my progress pace was roughly 0.5–1 workdays per week (i.e. I got as much done each week as I would’ve gotten done in 0.5–1 regular workdays if I was working full-time). Some weeks came in around 0 when my day job, urgent errands, or the occasional fun mini-project1 occupied all my brain space; some other weeks I had very productive Sundays and clocked closer to 1.5 workdays per week.

For 2022, I have some ideas for how to rearrange my weekday routine to carve out more high-energy time for Tanagram. I’ll see how well those ideas work over the next few weeks.

On to the roadmap update:

Alpha 1

Alpha 1 will be the first release of Tanagram (although it’ll probably be limited to a few close friends to start). It’ll have the following functionality:

You’ll be able to deploy an instance of the Tanagram kernel server on any server2 or connect to an existing server. Each server is called a “run instance”3.

Each run instance will have its own set of commands4. You can create or edit commands or run them with a form-input UI that is automatically generated from the command’s implementation.

Edits to commands sync immediately to the server once you press save and the code is live-reloaded for the next time you run the command; there’s no waiting for a deploy5.

You can view the run history of each command and search by input or output values6.

You’ll be able to assemble commands into HTTP servers with basic URL routing support.

All these pieces technically exist and are powering this blog, but much of it is very manual (e.g. I deploy code changes by SFTPing the source code onto my server and compiling a build) and, in short, Alpha 1 is all about building a user-friendly interface around all that.

Tanagram is designed to be self-hosted. “Owning your system” is important for developers, especially when that system is also a runtime for your code. It’s also straightforward to offer a cloud-hosted version of a self-hosted product if I decide that I want to do so as part of the business model. The kernel server runs on anything Elixir supports and has an API for all of the functionality listed above. I’m also building a client interface that is a native Mac app (i.e. AppKit) because I like native Mac apps, but the kernel API exists so that other people can build their own UIs if they don’t like mine.

Roadmap: January 2022

These roadmap updates will be a listing of functionality that I know I’ll need to build. I don’t yet have enough predictability in my working schedule to make a timeboxed “sprint commitment”, i.e. it’s likely that not all these will be done in January 2022. They are, however, listed in the order I plan to build them.

0. UI to create and edit HTTP servers and routes (✅).
1. Input UI to run a command (⏳).
2. UI and server support for editing the implementation of a command.
3. UI and server support for creating new commands.
4. UI and server support for showing the run history of each command.
5. Server support for de-retaining command run history.
6. UI and server support for creating/deploying a new run instance.

(✅ means “done”; ⏳ means “currently in-progress”).