process_experiments: A Rust repository from shelbyd

Purpose

I've been thinking a lot lately about the approaches we use for developing software (and systems in general). Many practices have clear benefits to their practitioners, TDD, refactoring, functional programming. But, I'm curious about optimizing the higher level algorithm of software development.

My goal is to find the best "system algorithm" that: given a software system, change its behavior, such that existing (desired) behavior is preserved, and subjective experience is maximized.

Approach

I'll implement the same set of behavior changes, in the same order, with different algorithms and compare the approaches afterwards. I'll be focusing on my subjective experience of these approaches, since I'll have far too little data to draw any scientific conclusions.

I'll implement the systems completely in one algorithm before moving on to the next. Sequences of commits in this repository will show the evolution of the system inside each algorithm.

Project

To draw any useful conclusions about software projects in general, we must implement something more complex than a single module.

Spec

I'll be making a command line twitter client modeled after Gary Bernhardt's Toot, reconstructed here. This will be enlightening because it:

includes networking
integrates with an external api
requires user interaction
can be command based or interactive
can require posting to an external api
can become realtime
can use a local mock api for QA testing

Constants

To minimize explorable surface area, all implementations will:

Be written in rust
Target the same behavior
Target behavior in the same order
Only consider existing and next desired behavior, not known future modifications
Attempt to not reuse knowledge learned from previous examples with regards to architecture

Approaches TODO