A PyFrame is the context for the execution of code. Different concrete sub-classes of PyFrame exist to execute different compiled representations of Python code. For example, there is one for CPython 3.11 byte code and (we expect) another for Java byte code. The type of code object supported is the parameter C to the class.

In order that argument processing may be uinform irrespective of concrete type, a PyFrame presents an abstraction that has arguments laid out in an array. For example, the function definition:

def func(a, b, c=3, d=4, /, e=5, f=6, *aa, g=7, h, i=9, **kk):
    v, w, x = b, c, d, e
    return u

the layout of the local variables in a frame would be as below

A Python frame

frame

a

b

c

d

e

f

g

h

i

aa

kk

u

v

w

x

code

argcount

kwonlyargcount

*

**

posonlyargcount

function

defaults

kwdefaults

In the last row of the table, the properties are supplied by the function object during each call. defaults apply in the position show, in order, while kwdefaults (in a map) apply to keywords wherever the name matches. The names in the frame are those in the PyCode#varnames field of the associated code object

The frame presents an abstraction of an array of named local variables, and two more of cell and free variables, while concrete subclasses are free to implement these in whatever manner they choose.