Add more ctype extension data structures #23
Conversation
This is to make subctypep work for negations of member types and non-member types
Also export defexclusives, and defexistential
list-of, array-of and hash-table-of
| @@ -57,3 +57,15 @@ | |||
| "cfunction" "packages")) | |||
| (:file "parse" | |||
| :depends-on ("generic-functions" "create" "classes" "config" "packages")))) | |||
|
|
|||
| (asdf:defsystem :ctype/ext | |||
There was a problem hiding this comment.
Perhaps in the future this system could consume the other bundled extensions. Alternatively perhaps you want them to be more granular than this.
| (:documentation "Homogeneous array ctype.")) | ||
|
|
||
| (defun carray-of (element-ctype &optional (dims '*) (upgraded-element-type '*) simplicity) | ||
| (if simplicity |
There was a problem hiding this comment.
I noticed something about not all implementations supporting complex vs simple arrays, but apparently they all do? Should I worry about this here?
There was a problem hiding this comment.
SICL will have only simple arrays, and in general I don't want to impose any more requirements than the standard does. The standard does allow having only simple arrays.
There was a problem hiding this comment.
carray has disjoin/2 defined so that (disjoin (array integer) (array float)) becomes (array (or integer float)). I expect it to become just (or (array integer) (array float)).
There was a problem hiding this comment.
As with list-of, I think you're right and the code's wrong.
| (deftype ,name ,lambda-list | ||
| ,documentation | ||
| ,@simple) | ||
| (setf (get ',name 'extended-type-parser) |
There was a problem hiding this comment.
I use the symbol property list. I hope that's okay.
There was a problem hiding this comment.
It means the environment is ignored, but I have some thoughts on that I'll put elewhere
There was a problem hiding this comment.
I have some concerns about the extended specifiers, since it's essentially a language extension, and I think it needs a bit more work. But as an experimental extension to this library I guess it's fine, provided that it doesn't interfere with using this library as an unextended typep/subtypep. I'm going to make a issue (#24) separate from this PR for how I think parsing ought to work.
cmember.lisp
Outdated
| (lambda (single-member) | ||
| (ctypep single-member ctype)) | ||
| (cmember-members cmember)) | ||
| (not (some (lambda (single-member) |
There was a problem hiding this comment.
Please use notany instead of not some
There was a problem hiding this comment.
I half blame git for this one.
| (:documentation "Homogeneous array ctype.")) | ||
|
|
||
| (defun carray-of (element-ctype &optional (dims '*) (upgraded-element-type '*) simplicity) | ||
| (if simplicity |
There was a problem hiding this comment.
SICL will have only simple arrays, and in general I don't want to impose any more requirements than the standard does. The standard does allow having only simple arrays.
parse.lisp
Outdated
| EXTENDED is a list of forms that return a ctype that completely represents the | ||
| custom type. | ||
|
|
||
| Both the SIMPILE and the EXTENDED forms share the parameters of LAMBDA-LIST. |
|
|
||
| Both the SIMPILE and the EXTENDED forms share the parameters of LAMBDA-LIST. | ||
|
|
||
| LAMBDA-LIST is an ordinary lambda list that also allows &environment." |
There was a problem hiding this comment.
Why is it not a deftype lambda list? More broadly, I think the right thing to do here is to have these work like macroexpanders - the expansion function takes the whole type specifier and the environment as arguments, and then parses the type specifier according to the non-env parts of the lambda list. But that will take work.
There was a problem hiding this comment.
To make it a deftype lambda list I will need to add &whole; I can use the same trick that I used for &environment. I will also need to make '* the default for &optional and &key. which will be more cumbersome.
There was a problem hiding this comment.
I think there is some code for this in https://github.com/s-expressionists/ecclesia and/or SICL
There was a problem hiding this comment.
To be clear, I know that this would be work and I'll accept the PR without it.
There was a problem hiding this comment.
Deftype lambda lists also allow for destructuring parameters. I'm not sure how to do that cleanly.
| (deftype ,name ,lambda-list | ||
| ,documentation | ||
| ,@simple) | ||
| (setf (get ',name 'extended-type-parser) |
There was a problem hiding this comment.
It means the environment is ignored, but I have some thoughts on that I'll put elewhere
generic-functions.lisp
Outdated
| @@ -1,5 +1,10 @@ | |||
| (in-package #:ctype) | |||
|
|
|||
| (defmacro define-commutative-method (name arg1 arg2 &body body) | |||
There was a problem hiding this comment.
define-commutative-method should look like defmethod - so the lambda list here should be (name (arg1 arg2) &body body). This confused me a couple times looking through the code.
There was a problem hiding this comment.
Do you agree that it should still only be for methods taking 2 arguments?
Add define-commutative-method and refactor existing methods to use it.
Add define-extended-type and extended-specifier-ctype for extending custom type specifiers to parse into ctypes.
Add extended types for list-of, array-of, and hash-table-of.
Add ctype/ext ASDF system.