bara-lang/docs/en/05-user-guide.md

← Back to Index

---

Bara Lang User Guide

Installation

Prerequisites

• Nim >= 2.0
• GCC or Clang
• make

Build from Source

git clone https://gitlab.com/balvatar/lisp-nim.git
cd lisp-nim
make build

CLI Commands

compile — Compile to Nim

./cljnim compile input.clj output.nim

Generates a .nim file from Clojure source.

run — Compile and Execute

./cljnim run examples/hello.clj

Compiles to Nim, then to C, then to a binary, and runs it.

read — Parse and Print AST

./cljnim read examples/hello.clj

Shows the Clojure AST as S-expressions.

repl — Interactive REPL

# Human-friendly mode
./cljnim repl

# AI mode (structured JSON)
./cljnim repl --json

Writing Bara Lang Programs

Basic Syntax

; Comments start with semicolon

; Define a variable
(def x 42)

; Define a function
(defn greet [name]
  (println "Hello, " name))

; Call a function
(greet "World")

; Arithmetic
(+ 1 2 3)      ; => 6
(* 10 20)      ; => 200
(/ 100 4)      ; => 25

; Conditionals
(if (> x 0)
  "positive"
  "non-positive")

; Local bindings
(let [a 10
      b 20]
  (+ a b))     ; => 30

Working with Data

; Vectors (use Nim seq internally)
(def nums [1 2 3 4 5])

; Keywords
(def person {:name "Alice" :age 30})

; Maps and sets use persistent HAMT data structures
; with structural sharing and O(log₃₂ n) operations.

Recursion

(defn factorial [n]
  (if (= n 0)
    1
    (* n (factorial (- n 1)))))

(println (factorial 5))  ; => 120

AI REPL Guide

The JSON REPL is designed for programmatic interaction.

Start AI REPL

./cljnim repl --json

Evaluate a Form

{"op": "eval", "form": "(+ 1 2 3)"}

Response:

{
  "status": "ok",
  "result": {"printed": "6"},
  "meta": {"ns": "user", "ms": 861, "form": "(+ 1 2 3)"}
}

Batch Evaluation

{"op": "eval-batch", "forms": ["(defn f [x] x)", "(f 42)"]}

List Definitions

{"op": "get-defs"}

Response:

{"status": "ok", "defs": ["f"], "ns": "user"}

Clear Session

{"op": "clear"}

Exit

{"op": "quit"}

Tips

• Use :help in human REPL for available commands.
• Definitions in REPL persist across evaluations in the same session.
• The --json flag makes the REPL fully machine-readable.