Session 4 — REPL & Interactive Prompt¶

What Was Built¶

The REPL (Read, Evaluate, Print, Loop) — an interactive prompt that lets you write Terse code one line at a time and see results instantly. The moment Terse became a language you can have a conversation with.

The REPL¶

File: src/interpreter/main.py

Run with no arguments to start the interactive prompt:

python main.py

Terse 0.1 — Goatface Tech
Type "exit" to quit, "clear" to reset knowledge base.

>>>

How It Works¶

The REPL creates a single Interpreter instance and keeps it alive for the whole session. Every line you type gets run through the full Lexer → Parser → Interpreter pipeline. The knowledge base accumulates across lines — everything you teach Terse stays in memory until you exit or clear.

Terse 0.1 — Goatface Tech
Type "exit" to quit, "clear" to reset knowledge base.

>>> know dog is animal
  learned: dog is animal

>>> know dog has fur
  learned: dog has fur

>>> when has fur then is mammal
  rule learned: when has fur then is mammal

>>> infer dog
  inferring: dog
  inferred: dog is mammal

>>> facts
  dog: {'is': 'mammal', 'has': 'fur'}

>>> exit
Goodbye.

Special Commands¶

The REPL has five built-in commands beyond standard Terse syntax:

Command	What it does
`exit`	Quit the REPL cleanly
`clear`	Reset the knowledge base — fresh interpreter
`facts`	Print all known node facts
`relationships`	Print all known edges between nodes
`sequences`	Print all Markov chain transitions with probabilities

These give you full visibility into what Terse knows at any point in a session.

Running Files vs REPL¶

main.py now serves two purposes:

# Start the REPL
python main.py

# Run a .trs file
python main.py ../../examples/hello.trs

If you pass a filename it runs that file. If you pass nothing it starts the interactive prompt.

Why the REPL Matters¶

Before the REPL, Terse was a file runner. You wrote code, saved it, ran it, read the output. The REPL changes that — you can explore the knowledge base interactively, test inference rules live, and teach Terse things one line at a time.

It also makes Terse approachable. Anyone can type know dog is animal and immediately see learned: dog is animal. No setup, no files, no compilation. Just a conversation.

Phase 1 Progress After Session 4¶

Component	Status
Lexer	✅ Complete
Parser	✅ Complete
Interpreter core	✅ Complete
Facts & relationships	✅ Complete
Inference rules	✅ Complete
Functions with scope	✅ Complete
Markov chain learning	✅ Complete
Each loops	✅ Complete
While loops	✅ Complete
main.py	✅ Complete
First .trs program	✅ Complete
REPL	✅ Complete
Error handling	🔵 Next session
Tensors	⚪ Planned
Compression keywords	⚪ Planned
Ethics rule syntax	⚪ Planned

Interpreter completeness: ~75%