Matus Goljer

(rx (or (and (? "#") "'") "," "`" ",@")) should not be analyzed in the traditional sense

3

This presents a lot of "unreachable expression" false warnings.

Implement handlers for all special forms

3

https://www.gnu.org/software/emacs/manual/html_node/elisp/Special-Forms.html - [x] and - [ ] catch - [x] cond - [x] condition-case - [x] defconst - [x] defvar - [x] function - [x] if - [x] interactive -...

Something like `(:foo 1 :foo 2)` should raise a warning since this is probably an error.

There is not really much to implement except the unification mechanism and then some simple substitution magic on `elsa-get-type` which would take the types from the surroundings and fill the...

Add some generic mechanism for normalizing composite types

1

Types such as `(or (cons int int) (cons int symbol))` should be normalized to `(cons int (or int symbol))`. This looks like a homomorphism... maybe we can construct some sort...

This function is invalid ```elisp (defun x (a) (1+ a) (string-to-number a)) ``` even when we have no idea what `a` is because the overlap of the domains for argument...

Introduce ternary logic

2

Make `elsa-type-accept` work with ternary logic. When we have a predicate like `(stringp x)` the results can be: - `x` is a `String`, in which case always return `true` -...

Add support for `cl-defmethod`

1

Blocked by #36. - [x] parse the `cl-defmethod` as if it were a `defun` - [x] handle `cl-defgeneric` - [ ] add type information for arguments which have the native...

See ```elisp (cl-defmethod elsa-check-check ((_ elsa-check-elsa-prefer-elsa-get-name) form scope state) (let* ((prop (elsa-get-name (elsa-nth 2 form)))) (when (eq prop 'name) (elsa-state-add-message state (elsa-make-notice (elsa-car form) "Prefer (elsa-get-name %s) to (oref %s...

This is really more generic than just form sequences I think. We should be able to work with forms or lists seamlessly. (this can also be implemented in terms of...