Notice: While Javascript is not essential for this website, your interaction with the content will be limited. Please turn Javascript on for the full experience.

PEP 612 -- Parameter Specification Variables

Title:Parameter Specification Variables
Author:Mark Mendoza <mendoza.mark.a at>
Sponsor:Guido van Rossum <guido at>
Discussions-To:Typing-Sig <typing-sig at>
Type:Standards Track

Parameter Specification Variables


There currently are two ways to specify the type of a callable, the Callable[[T1, T2], TReturn] syntax defined in PEP 484, and callback protocols from PEP 544. Neither of these support forwarding the parameter types of one callable over to another callable, making it difficult to annotate function decorators. This PEP proposes typing.ParameterSpecification, a new kind of type variable, to support expressing these kinds of relationships.


The existing standards for annotating higher order functions don’t give us the tools to annotate the following common decorator pattern satisfactorily:

from typing import Awaitable, Callable, TypeVar

TReturn = TypeVar("TReturn")

def add_logging(
    f: Callable[..., TReturn]
) -> Callable[..., Awaitable[TReturn]]:
    async def inner(*args: object, **kwargs: object) -> TReturn:
        await log_to_database()
        return f(*args, **kwargs)
    return inner

def foo(x: int, y: str) -> int:
    return x + 7

await foo(1, "A")
await foo("B", 2) # fails at runtime

add_logging, a decorator which logs before each entry into the decorated function, is an instance of the Python idiom of one function passing all arguments given to it over to another function through the combination of the *args and **kwargs features in both parameters and in arguments. When one defines a function (like inner) that takes (*args, **kwargs) and goes on to call another function with (*args, **kwargs), the wrapping function can only be safely called in all of the ways that the wrapped function could be safely called. To type this decorator, we’d like to be able to place a dependency between the parameters of the callable f and the parameters of the returned function. PEP 484 supports dependencies between single types, as in def append(l: typing.List[T], e: T) -> typing.List[T]: ..., but there is no existing way to do so with a complicated entity like the parameters one could pass to a function.

Due to the limitations of the status quo, the add_logging example will type check but will fail at runtime. inner will pass the string “B” into foo, which will try to add 7 to it, triggering a type error. This was not caught by the type checker because the decorated foo was given the type Callable[..., Awaitable[int]] which is specified to do no validation on its arguments.

Without the ability to define dependencies between the parameters of different callable types, there is no way, at present, to make add_logging compatible with all functions, while still preserving the enforcement of the parameters of the decorated function.

With the addition of the ParameterSpecification variables proposed by this PEP, we can rewrite the previous example in a way that keeps the flexibility of the decorator and the parameter enforcement of the decorated function.

from typing import Awaitable, Callable, ParameterSpecification, TypeVar

Ps = ParameterSpecification("Ps")
R = TypeVar("R")

def add_logging(f: Callable[Ps, R]) -> Callable[Ps, Awaitable[R]]:
    async def inner(*args: Ps.args, **kwargs: Ps.kwargs) -> R:
        await log_to_database()
        return f(*args, **kwargs)

    return inner

def foo(x: int, y: str) -> int:
    return x + 7

await foo(1, "A")
await foo("B", 2) # Incompatible parameter type:
                  # Expected `int` for 1st anonymous parameter to call `foo`
                  # but got `str`



A parameter specification variable is defined in a similar manner to a normal typing.TypeVar.

from typing import ParameterSpecification
TParams = ParameterSpecification("TParams") # Accepted
TParams = ParameterSpecification("WrongName") # Rejected

The runtime should accept bounds and covariant and contravariant arguments in the declaration just as typing.TypeVar does, but for now we will defer the standardization of the semantics of those options to a later PEP.

Valid use locations

A declared ParameterSpecification can only be used in the place of the list of types in the declaration of a Callable type, or a user defined class which is generic in a ParameterSpecification variable (i.e., MyClass in the following example).

def foo(
    x: typing.Callable[TParams, int]
) -> typing.Callable[TParams, str]:  # Accepted
def foo(
    x: MyClass[TParams, int]
) -> typing.Callable[TParams, str]:  # Accepted
def foo(x: TParams) -> TParams: ...  # Rejected
def foo(x: typing.List[TParams]) -> None: ... # Rejected
def foo(x: typing.Callable[[int, str], TParams]) -> None: ... # Rejected


The inference rules for the return type of a function invocation whose signature contains a ParameterSpecification variable are analogous to those around evaluating ones with TypeVars.

def foo(
    x: typing.Callable[TParams, int]
) -> typing.Callable[TParams, str]: ...
def bar(a: str, b: bool) -> int: ...
f = foo(bar) # f should be inferred to have the same signature as bar,
             # but returning str
f("A", True) # Accepted
f(a="A", b=True) # Accepted
f("A", "A") # Rejected

Just as with traditional TypeVars, a user may include the same ParameterSpecification multiple times in the arguments of the same function, to indicate a dependency between multiple arguments. In these cases a type checker may choose to solve to a common behavioral supertype (i.e. a set of parameters for which all of the valid calls are valid in both of the subtypes), but is not obligated to do so.

def foo(
    x: typing.Callable[TParams, int], y: typing.Callable[TParams, int]
) -> typing.Callable[TParams, bool]: ...

def x_int_y_str(x: int, y: str) -> int: ...
def y_int_x_str(y: int, x: str) -> int: ...
foo(x_int_y_str, x_int_y_str) # Must return (x: int, y: str) -> int
foo(x_int_y_str, y_int_x_str) # Could return (__a: int, __b: str) -> int
                              # This works because both callables have types
                              # that are behavioral subtypes of
                              # Callable[[int, str], int]

def keyword_only_x(*, x: int) -> int: ...
def keyword_only_y(*, y: int) -> int: ...
foo(keyword_only_x, keyword_only_y) # Must be rejected

Use in Generic Classes

Just as with normal TypeVars, ParameterSpecifications can be used to make generic classes as well as generic functions. These are able to be mixed with normal TypeVars. This also work with protocols in the same manner.

The components of a ParameterSpecification

A ParameterSpecification captures both positional and keyword accessible parameters, but there unfortunately is no object in the runtime that captures both of these together. Instead, we are forced to separate them into *args and **kwargs, respectively. This means we need to be able to split apart a single ParameterSpecification into these two components, and then bring them back together into a call. To do this, we introduce TParams.args to represent the tuple of positional arguments in a given call and TParams.kwargs to represent the corresponding Mapping of keywords to values. These operators can only be used together, as the annotated types for *args and **kwargs .

class G(Generic[TParams]):
    def foo(
        *args: TParams.args, **kwargs: TParams.kwargs
    ) -> int:                                # Accepted
    def bar(
        *args: TParams.kwargs, **kwargs: TParams.args
    ) -> int:                                # Rejected
    def baz(*args: TParams.args) -> int: ... # Rejected
    stored_arguments: TParams.args           # Rejected
    def bap(x: TParams.args) -> int: ...     # Rejected
    def bop(
        *args: List[TParams.args], **kwargs: TParams.kwargs
    ) -> int:                                # Rejected

Because the default kind of parameter in Python ((x: int)) may be addressed both positionally and through its name, two valid invocations of a (*args: TParams.args, **kwargs: TParams.kwargs) function may give different partitions of the same set of parameters. Therefore we need to make sure that these special types are only brought into the world together, and are used together, so that our usage is valid for all possible partitions.

With those requirements met, we can now take advantage of the unique properties afforded to us by this set up:

  • Inside the function, args has the type TParams.args, not Tuple[TParams.args, ...] as would be with a normal annotation (and likewise with the **kwargs)
  • A function of type Callable[TParams, TReturn] can be called with (*args, **kwargs) if and only if args has the type TParams.args and kwargs has the type TParams.kwargs, and that those types both originated from the same function declaration.
  • A function declared as def inner(*args: TParams.args, **kwargs: TParams.kwargs) -> X has type Callable[TParams, X].

With these three properties, we now have the ability to fully type check parameter preserving decorators.

One additional form that we want to support is functions that pass only a subset of their arguments on to another function. To avoid shadowing a named or keyword only argument in the ParameterSpecification we require that the additional arguments be anonymous arguments that precede the *args and *kwargs

def call_n_times(
    __f: Callable[TParams, None],
    __n: int,
    *args: TParams.args,
    **kwargs: TParams.kwargs,
) -> None:
    for x in range(__n);
        __f(*args, **kwargs)

Backwards Compatibility

The only changes necessary to existing features in typing is allowing these ParameterSpecification objects to be the first parameter to Callable and to be a parameter to Generic. Currently Callable expects a list of types there and Generic expects single types, so they are currently mutually exclusive. Otherwise, existing code that doesn't reference the new interfaces will be unaffected.

Reference Implementation

The Pyre type checker supports ParameterSpecifications, .args and .kwargs in the context of functions. Support for use with Generic is not yet implemented. A reference implementation of the runtime components needed for those uses is provided in the pyre_extensions module.

Rejected Alternatives

Using List Variadics and Map Variadics

We considered just trying to make something like this with a callback protocol which was parameterized on a list-type variadic, and a map-type variadic like so:

Treturn = typing.TypeVar(“Treturn”)
Tpositionals = ....
Tkeywords = ...
class BetterCallable(typing.Protocol[Tpositionals, Tkeywords, Treturn]):
  def __call__(*args: Tpositionals, **kwargs: Tkeywords) -> Treturn: ...

However there are some problems with trying to come up with a consistent solution for those type variables for a given callable. This problem comes up with even the simplest of callables:

def simple(x: int) -> None: ...
simple <: BetterCallable[[int], [], None]
simple <: BetterCallable[[], {“x”: int}, None]
BetterCallable[[int], [], None] </: BetterCallable[[], {“x”: int}, None]

Any time where a type can implement a protocol in more than one way that aren’t mutually compatible, we can run into situations where we lose information. If we were to make a decorator using this protocol, we have to pick one calling convention to prefer.

def decorator(
  f: BetterCallable[[Ts], [Tmap], int],
) -> BetterCallable[[Ts], [Tmap], str]:
    def decorated(*args: Ts, **kwargs: Tmap) -> str:
       x = f(*args, **kwargs)
       return int_to_str(x)
    return decorated
def foo(x: int) -> int:
    return x
reveal_type(foo) # Option A: BetterCallable[[int], {}, str]
                 # Option B: BetterCallable[[], {x: int}, str]
foo(7)   # fails under option B
foo(x=7) # fails under option A

The core problem here is that, by default, parameters in Python can either be passed in positionally or as a keyword parameter. This means we really have three categories (positional-only, positional-or-keyword, keyword-only) we’re trying to jam into two categories. This is the same problem that we briefly mentioned when discussing .args and .kwargs. Fundamentally, in order to capture two categories when there are some things that can be in either category, we need a higher level primitive (ParameterSpecification) to capture all three, and then split them out afterward.

Mutations on ParameterSpecifications

There are still a class of decorators still not supported with these features: those that mutate (add/remove/change) the parameters of the given function. Defining operators that do these mutations becomes very complicated very quickly, as you have to deal with name collision issues much more prominently. We will defer that work until there is significant demand, and then we would be open to revisiting it.

Naming this an ArgSpec

We think that calling this a ParameterSpecification is more correct than referring to it as an Argument Specification, since callables have parameters, which are distinct from the arguments which are passed to them in a given call site. A given binding for a ParameterSpecification is a set of function parameters, not a call-site’s arguments.


Thanks to all of the members of the Pyre team for their comments on early drafts of this PEP, and for their help with the reference implementation.

Thanks are also due to the whole Python typing community for their early feedback on this idea at a Python typing meetup, leading directly to the much more compact .args/.kwargs syntax.