Description
When filling in a docstring, there is no proper adaptive filling and it always seems to fill to the beginning column of the docstring. This problem has annoyed me for more than 5 years, so I stepped a bit through the debugger, but I don't understand Emacs Lisp well enough to provide a proper fix.
An arg - asdfasdfasd sadf as dfas df asdf asdf as df as df asd
fas df asdf asdf asdf as df asd fa sd fa sdf as df asdf as dfa sdf
Put the cursor somewhere into the indented section and press M-q. In Clojure mode, it will ruin the indentation. Proper adaptive filling will cut as dfa sdf
and put in on the next line at col 9.
My best solution so far is to set adaptive-fill-function
to ignore
and fill-paragraph-function
to nil
. So far, I haven't observed any adverse side effects but sure it breaks something.
In general, there seems to be a bunch of logic implemented around docstrings in Clojure mode, but it is not very well documented and probably not working properly. Probably one could draw inspiration for how docstring indentation is solved in other LISP major modes.
I noticed that in Emacs Lisp for instance, docstrings have no special indentation behavior, and in Clojure there are two spaces prepended to each line. Its a weird convention but it seems to be somehow rooted in Clojure core? Is that what all the logic is about? Maybe getting rid of that is too much to ask for, but it would be nice if you can make adaptive filling work.
;;; Workaround, use at your own risk
(add-hook 'clojure-mode-hook
(lambda ()
(interactive)
(setq-local fill-paragraph-function nil)
(setq-local adaptive-fill-function 'ignore)))