Strings and bytes

Python 2’s basic original strings are called 8-bit strings, and they play a dual role in Python 2 as both ASCII text and as byte sequences. While Python 2 also has a unicode string type, the fundamental ambiguity of the core string type, coupled with Python 2’s default behavior of supporting automatic coercion from 8-bit strings to unicode objects when the two are combined, often leads to UnicodeErrors. Python 3’s standard string type is Unicode based, and Python 3 adds a dedicated bytes type, but critically, no automatic coercion between bytes and unicode strings is provided. The closest the language gets to implicit coercion are a few text-based APIs that assume a default encoding (usually UTF-8) if no encoding is explicitly stated. Thus, the core interpreter, its I/O libraries, module names, etc. are clear in their distinction between unicode strings and bytes. Python 3’s unicode support even extends to the filesystem, so that non-ASCII file names are natively supported.

This string/bytes clarity is often a source of difficulty in transitioning existing code to Python 3, because many third party libraries and applications are themselves ambiguous in this distinction. Once migrated though, most UnicodeErrors can be eliminated.

Numbers

Python 2 has two basic integer types, a native machine-sized int type, and an arbitrary length long type. These have been merged in Python 3 into a single int type analogous to Python 2’s long type.

In addition, integer division now produces floating point numbers for non-integer results.

Classes

Python 2 has two core class hierarchies, often called classic classes and new-style classes. The latter allow for such things as inheriting from the builtin basic types, support descriptor based tools like the property builtin and provide a generally more sane and coherent system for dealing with multiple inheritance. Python 3 provided the opportunity to completely drop support for classic classes, so all classes in Python 3 automatically use the new-style semantics (although that’s a misnomer now). There is no need to explicitly inherit from object or set the default metatype to enable them (in fact, setting a default metatype at the module level is no longer supported - the default metatype is always object).

The mechanism for explicitly specifying a metaclass has also changed to use a metaclass keyword argument in the class header line rather than a __metaclass__ magic attribute in the class body.

Multiple spellings

There are many cases in Python 2 where multiple spellings of some constructs exist, such as repr() and backticks, or the two inequality operators != and <>. In all cases, Python 3 has chosen exactly one spelling and removed the other (e.g. repr() and != were kept).

Imports

In Python 3, implicit relative imports within packages are no longer available - only absolute imports and explicit relative imports are supported. In addition, star imports (e.g. from x import *) are only permitted in module level code.

Also, some areas of the standard library have been reorganized to make the naming scheme more intuitive. Some rarely used builtins have been relocated to standard library modules.

Iterators and views

Many APIs, which in Python 2 returned concrete lists, in Python 3 now return iterators or lightweight views.