thing-at-point 'sexp vs `symbol [WIP]

[jacktasia]

Experiments related to #179

I noticed that "things" sexp and symbol will behave differently with thing-at-point based on the current mode. Some example code below

(with-temp-buffer
  (funcall 'lisp-mode) ;; if removed or changed you can get different results for symbol vs sexp
  (let ((expected "#/something")) 
    (insert (concat "(" expected ")"))
    (goto-char (point-min))
    (forward-char 5)
    (let* ((cur (thing-at-point 'sexp)) ;; vs 'symbol
           (line (thing-at-point 'line))
           (matches (string= expected cur)))
           (message "thing: %s | matched: %s | line: %s" cur matches line))))

because of this I am considering adding two new options dumb-jump-default-thing and dumb-jump-language-thing. Where you can either define a "thing" per language or a default.

For anyone reading this please let me know if you have any feedback either way. Thanks!

[coveralls]

Coverage decreased (-0.5%) to 96.84% when pulling 411ec72 on thing-at-point-sexp into c4b3e4b on master.

[netromdk]

I think that sounds reasonable. So you'll default both to 'symbol?

[jacktasia]

I think that sounds reasonable. So you'll default both to 'symbol?

Yeah, basically I was thinking that dumb-jump-default-thing would default to 'symbol and dumb-jump-language-thing would be an alist (I think?) and if the current language was not in there then it would fallback to the default. Any thoughts/concerns? Thanks!

[netromdk]

Sounds fine to me. Would be interesting with some meaningful tests showing the different cases probably.