A small lab, a longer attention span

A useful division of work

The Mac is the editing surface. The Ubuntu laptop is where the code runs.

That separation makes the environment concrete. Files live on the Linux machine, builds happen there, and a CUDA program runs against the GPU that is physically in that laptop.

SSH makes the distance small. Tailscale provides a route to the machine when I’m away from the local network. A long-running job can live in a persistent terminal session while I move on to something else.

Editör önümde, deney evde. Bağlantı kopsa da iş devam edebilsin.

Give the machine a role

The interesting question is what to use regularly. For this setup, the answer is a development and systems lab:

1. Write and compile small C++ programs.
2. Work through CUDA exercises.
3. Run a service with a database and a cache.
4. Keep the inputs, commands, and observations together.

The API sandbox is deliberately small: Python, PostgreSQL, and Redis. Docker Compose makes the group easy to start and stop. The first checks establish that a record can be written, read from the database, and then served from the cache.

A session should leave a trace

An experiment is easier to return to when it includes a little context:

question:
  What am I trying to understand?

environment:
  What was running, and on which machine?

procedure:
  Which command produced the observation?

next question:
  What should I change or investigate next?

Even a short note is useful. It turns a half-remembered terminal session into a starting point for the next one.

Small enough to understand

A small setup makes it easier to follow a request, notice a hidden dependency, and connect a graph to a process. I want the lab to stay understandable as it grows.

This notebook is one part of that arrangement: a place to leave the question, the setup, and eventually the result.

Next: the working lab.