Request: Infer generics

The request is to let infer support generics.

Background:

I recently published a recursive Pipe and Compose. It offers some key advantages, including the preservation of variable names.

https://github.com/babakness/pipe-and-compose-types

Its shortcoming is in the case of generics. This is because infer does not work well with generics. Example

/**
 * Extracts function arguments
 */
export type ExtractFunctionArguments < Fn > = Fn extends  ( ...args: infer P ) => any  ? P : never

/**
 * Extracts function return values
 */
export type ExtractFunctionReturnValue<Fn> = Fn extends  ( ...args: any[] ) => infer P  ? P : never

type Foo = ExtractFunctionArguments< <A>(a:A, b:number) => A >
// Foo has type [ {} , number ]

TypeScript 3.4 improved how pipe and compose functions work with generics with higher order type inference from generic functions.

Common implementations of pipe and compose rely on parameter overloading. However, even here, there is an issue with variadic functions. See example in this issue:

#30727

The broader issue with the parameter overloading version of pipe and compose is that parameter names are lost, decreasing code clarity. The recursive version, relying on infer does maintain parameter names.

Providing support generics with infer will present a major productivity bump and will nearly complete the community's need for the common place pipe and compose in FP patterns without compromise.

The only outlier is a small situation with parameter names on variadic functions of a minimum arity of zero as opposed to at least one. Can provide examples if needed since it is a separate issue.

Comienza reproduciendo los ejemplos de funciones genéricas del issue, centrándote en cómo infer maneja los argumentos y los valores de retorno de las funciones. Lee el contexto de inferencia genérica de orden superior de TypeScript 3.4 y compáralo con la discusión sobre funciones variádicas en el issue #30727. Se considera terminado cuando los tipos genéricos se conservan a través de infer sin perder la información prevista de los parámetros.