Developer Coding Practices

Philosophy

Functional-lite. Fail fast. Defer defense. Ship the meaningful change, abstract the rest.

Functional Style

• •Declarative over imperative — say what, not how
• •Immutable by default — mutate only for perf (1000s of objects)
• •Pure functions — separate data from functions that operate on it
• •Segregate state — isolate side-effects, avoid shared state
• •Composition over inheritance — curry, compose, pipe
• •Unary functions — general config first, specific data last
• •Point-free when clean — reduce temp variables, reduce surface area
• •Recursion over loops — but mind the tail-call stack

Code Style

• •Descriptive names, readable syntax — no abbreviations unless universal
• •Variables: follow language convention (camelCase JS/TS, snake_case Python)
• •Files/folders: kebab-case
• •No comments or explanations in code
• •No defensive coding in first iterations
• •Fix data issues upstream, not with defensive checks
• •Defer validation, error handling, logging, metrics until asked
• •Mention critical edge cases in chat; don't code them unsolicited

JS/TS Specifics

• •undefined only, never null

Architecture

Orchestrator → Pipeline → Helpers

1. •Main function: receives inputs, coordinates
2. •Data pipeline: transforms through helper chain
3. •Helpers: one logical chunk each, easily testable

Main can be long. Helpers stay focused. No arbitrary length rules.

Organization

• •Helpers live with orchestrator until 2-3+ accumulate
• •Extract testable logic: transforms, algorithms, pure business logic
• •Isolate side-effects

App vs Library Code

Config

• •Constants → constants file
• •Config → config file
• •Validate config before runtime

Problem-Solving

1. •Decompose: iterator, transform, state
2. •Patterns repeated 2-3x → make generic
3. •Custom parts reveal themselves
4. •10x = show the change, hide the noise