https://github.com/guicho271828/type-i

Type Inference Utility on unary type-checking predicates
https://github.com/guicho271828/type-i

Last synced: 29 days ago
JSON representation

Type Inference Utility on unary type-checking predicates

• Host: GitHub
• URL: https://github.com/guicho271828/type-i
• Owner: guicho271828
• Created: 2015-03-07T03:12:12.000Z (almost 10 years ago)
• Default Branch: master
• Last Pushed: 2023-01-27T15:05:30.000Z (almost 2 years ago)
• Last Synced: 2024-10-15T14:10:36.840Z (3 months ago)
• Language: Common Lisp
• Homepage:
• Size: 19.5 KB
• Stars: 4
• Watchers: 3
• Forks: 3
• Open Issues: 3
• Metadata Files:
  • Readme: README.org

Awesome Lists containing this project

README

        

* Type-I : type-inference from a predicate

This library tries to provide a way to detect what kind of type the given

predicate is trying to check. *This is different from inferring the return

type of a function.* For example,

#+BEGIN_SRC lisp

(eq 'null (test-type '(eql nil ?)))

(eq 'string (test-type '(stringp ?)))

#+END_SRC

The inference is done on form basis, and the equivalence of predicates are

determined by =#'equal=. To simplify the design, the argument to check

should be a symbol =?=, exported in package =type-i=.

Function =test-type= returns the inferred type. In contrast, =type-tests=

returns a list of all possible test predicates that returns true when =?=

is bound to the object of interest.

#+BEGIN_SRC lisp

  (is (subset '((TYPEP ? 'integer)

                (integerp ?))

              (type-tests 'integer)))

  ;; more inference on integers, e.g., (< 0 ? 4), should be added

  (is (subset '((TYPEP ? '(mod 5))

                (TYPEP ? '(integer 0 4)))

              (type-tests '(mod 5))))

#+END_SRC

This library is extensible. with =define-inference-rule= macro, you can add

more inferers to improve this library. Each inference rule is a unary

function that takes a predicate form, then returns a list of more

forms.

The =test-type= searches in the form space, adding the results of each

inference rule, until a form =(typep ? X)= is found (where X is

unknown). If it fails to find such a form even if the maximal set

is obtained, then the =test-type= returns nil. =type-tests= just returns

the maximal set.

I currently implemented the following inference rules:

- typep -- =(typep ? 'array)= -> =(arrayp ?)= etc.

- unary -- =(arrayp ?)= -> =(typep ? 'array) (typep ? '(array)) (typep ?

  '(array *))= ...

- null -- =(null ?) (typep ? null) (eql ? nil) (eql nil ?) (eq ? nil)= ...

- derived -- call =typexpand=

Inference on exhaustive partitions, e.g., =(typep ? 'list)= -> =(typep ?

'(or cons null))= is planned.

** Dependencies

This library is at least tested on implementation listed below:

+ SBCL 1.2.8 on X86-64 Linux  3.13.0-46-generic (author's environment)

Also, it depends on the following libraries:

+ Trivia by Masataro Asai ::

     NON-Optimized Pattern Matching Library

+ alexandria by  ::

    Alexandria is a collection of portable public domain utilities.

+ iterate by  ::

    Jonathan Amsterdam's iterator/gatherer/accumulator facility

+ introspect-environment by Bike  ::

    Small interface to portable but nonstandard introspection of CL environments.

** Author

+ Masataro Asai ([email protected])

* Copyright

Copyright (c) 2015 Masataro Asai ([email protected])

* License

Licensed under the LLGPL License.