Python 3.4: What's New, Release History & Support Lifecycle

What Is New in Python 3.4

Python 3.4 was released on March 16, 2014. The biggest additions are the asyncio module (the event loop and coroutine infrastructure that later became central to Python's async story), pathlib for object-oriented filesystem paths, enum for enumeration types, and statistics for basic descriptive statistics. CPython also gained a pip bundled by default and improved importlib.

asyncio -- Python's Async Foundation (PEP 3156)

The asyncio module landed in 3.4 as a provisional module. It brought an event loop, coroutine support (via yield from -- the await syntax came in 3.5), Future objects, and a transport/protocol framework for network code. This was the module that made Python a credible async platform.

import asyncio

@asyncio.coroutine
def greet_after(delay, name):
    yield from asyncio.sleep(delay)
    print(f"Hello, {name}")

loop = asyncio.get_event_loop()
loop.run_until_complete(greet_after(1.0, "World"))
loop.close()

In practice, 3.4's asyncio was provisional and somewhat verbose. The real productivity gains came with the async/await keywords in 3.5 and the API cleanup in 3.7 (asyncio.run()).

pathlib -- Object-Oriented Paths (PEP 428)

pathlib.Path represents filesystem paths as objects with methods, replacing the fragmented os.path string functions. You can join paths with /, read and write files with methods, and traverse directories with a clean API.

from pathlib import Path

p = Path("/etc") / "hosts"
print(p.exists())          # True
print(p.read_text())       # file contents

data_dir = Path("data")
for csv in data_dir.glob("*.csv"):
    print(csv.stem)        # filename without extension

For any new code that does filesystem work, pathlib is the right starting point. The os.path family still works and is often faster for simple operations, but pathlib reads more clearly and handles edge cases more consistently.

enum -- Enumeration Types (PEP 435)

The enum module provides a way to define named constants with identity, comparison, and iteration support. It replaces the pattern of defining constants as bare integers or strings scattered at module level.

from enum import Enum, IntEnum, auto

class Status(Enum):
    PENDING   = auto()
    RUNNING   = auto()
    DONE      = auto()
    FAILED    = auto()

print(Status.RUNNING.name)   # "RUNNING"
print(Status.RUNNING.value)  # 2
print(Status["DONE"])        # Status.DONE

class Permission(IntEnum):
    READ    = 4
    WRITE   = 2
    EXECUTE = 1

Other Notable Additions in Python 3.4

• statistics module: mean(), median(), mode(), stdev(), variance() -- basic stats without numpy.
• tracemalloc: Memory allocation tracing to find leaks and hot allocators. tracemalloc.start() begins tracing; tracemalloc.take_snapshot() captures a point-in-time view.
• pip bundled: CPython installer now ships pip by default (ensurepip module). No more bootstrapping pip separately after installation.
• Pickle protocol 4: Supports very large objects (over 4 GiB), strings longer than 2³² bytes, and uses more efficient opcodes for common types.
• SSL improvements: Hostname checking enabled by default; SSLv2 and SSLv3 disabled; RC4 ciphers removed.
• codecs.open() deprecation path: io.open() is now the preferred way to open text files with specific encoding.