TezzNative combines the clean clarity of Python with bare-metal native performance, built-in SIMD, native tensors, Erlang-style actor supervision, and zero garbage collection pauses.
TezzNative is statically typed with compile-time verification and zero runtime overhead. Built-in primitive types map directly to CPU registers and native SIMD vectors.
// Primitive scalar types
let a:int = 42 // 64-bit signed integer
let b:float = 3.14159 // 64-bit double precision float
let c:char = 'Z' as char // 8-bit character
let s:str = "TezzNative" // UTF-8 string slice
// Built-in SIMD Vector Types
v4:Vec4f // 128-bit SIMD vector (4 x f32)
v8:Vec8f // 256-bit AVX vector (8 x f32)
Variables are declared with optional let or direct type annotation name:type = value. The standard io module provides high-throughput buffered and streaming file operations.
import "io"
fn process_file(path:str) -> int:
h:*File = io.file_open(path, "w")
if h == 0:
ret 0 - 1
io.file_write_line(h, "TezzNative High-Performance I/O")
io.file_close(h)
say "File written successfully to:", path
ret 0
Clean indentation-based blocks with if / else, while loops, and arithmetic modulo operators.
let count:int = 0
while count < 10:
if count % 2 == 0:
say count, "is even"
else:
say count, "is odd"
count = count + 1
Functions compile to standard native calling conventions with register-passed arguments. Tail-recursive functions are automatically optimized into zero-overhead loop jumps without stack growth.
// Guaranteed Tail-Call Optimized Fibonacci
fn fib_tail(n:int, a:int, b:int) -> int:
if n == 0: ret a
if n == 1: ret b
ret fib_tail(n - 1, b, a + b) // Lowered to zero-overhead JMP
fn main() -> int:
say "fib(10) =", fib_tail(10, 0, 1)
ret 0
Structs lay out fields continuously in memory without hidden padding. Dot syntax works seamlessly on values (val.field) and pointers (ptr.field).
struct TensorShape:
ndim:int
rows:int
cols:int
total_elements:int
fn create_matrix(rows:int, cols:int) -> TensorShape:
shape:TensorShape
shape.ndim = 2
shape.rows = rows
shape.cols = cols
shape.total_elements = rows * cols
ret shape
TezzNative features robust Erlang-style process supervision with isolated mailboxes, message passing, restart policies, and fault tolerance built into the core runtime.
import "actor"
fn main() -> int:
sys:*ActorSystem = actor.system_new(4, 2, 2)
if sys == 0:
ret 1
worker:int = actor.spawn(sys, "worker")
actor.send(sys, worker, 10, 1, 0, 0, 0)
say "Actor spawned with ID:", worker
say "Mailbox length:", actor.mailbox_len(sys, worker)
actor.system_free(sys)
ret 0
First-class tensor primitives with SIMD AVX-512 and SSE accelerated kernels. Supports multidimensional representations, stride manipulation, and hardware-accelerated computation.
import "tensor"
fn main() -> int:
// Allocate SIMD-aligned 2D tensor descriptor [128 x 512]
desc:TensorDesc = tensor.tensor_desc(128, 512, tensor.tensor_dtype_f32(), tensor.tensor_device_cpu(), 64)
say "Tensor shape rows:", desc.rows, "cols:", desc.cols
say "Total elements:", desc.size, "allocated bytes:", desc.bytes
ret 0
Integrated native text-to-speech synthesis with zero external dependencies. Supports customizable voice profiles, speech rate, pitch tuning, and direct speaker output.
import "tts"
fn main() -> int:
// Initialize native neural/rule-based TTS engine
eng:*TtsEngine = tts.tts_new()
tts.tts_speak(eng, "Welcome to TezzNative voice assistant.")
tts.tts_free(eng)
ret 0
Native 60 FPS immediate-mode desktop GUI framework with rounded surfaces, anti-aliased font rendering, theme palettes, and input event dispatching.
import "tezzui"
fn main() -> int:
ui:TzUI
ui_create(&ui, "TezzNative Desktop App", 800, 600)
while ui_alive(&ui) != 0:
ui_begin(&ui)
ui_rect_r(&ui, 50, 50, 700, 100, TZ_SURFACE, 16)
ui_label(&ui, 80, 85, "Hello TezzNative Native GUI!", TZ_ACCENT, 2, 1)
ui_end(&ui)
ret 0
TezzNative guarantees safety for all regular code. When low-level hardware memory, pointer arithmetic, or OS driver interaction is required, explicit unsafe: blocks isolate raw pointers.
fn peek_raw_byte(addr:int) -> int:
byte_val:int = 0
unsafe:
ptr:*char = addr as *char
byte_val = ptr[0] as int
ret byte_val