bug#68956: M-q → Invalid predicate, see ‘treesit-thing-settings’ for valid forms for a predicate: nil

[Eli Zaretskii <eliz@gnu.org>]

Ping!  Dominique, could you please answer Yuan's question below?

> Cc: 68956@debbugs.gnu.org
> Date: Wed, 10 Apr 2024 13:51:33 +0300
> From: Eli Zaretskii <eliz@gnu.org>
> 
> > From: Yuan Fu <casouri@gmail.com>
> > Date: Tue, 9 Apr 2024 23:03:21 -0700
> > Cc: 68956@debbugs.gnu.org
> > 
> > [Adding the bug tracker back, forgot to replay-all]
> 
> And it might be a good idea to add Dominique as well ;-)
> 
> > 
> > > On Apr 8, 2024, at 8:42 PM, Yuan Fu <casouri@gmail.com> wrote:
> > > 
> > > 
> > > Eli Zaretskii <eliz@gnu.org> writes:
> > > 
> > >>> Cc: 68956@debbugs.gnu.org
> > >>> Date: Sat, 10 Feb 2024 10:44:01 +0200
> > >>> From: Eli Zaretskii <eliz@gnu.org>
> > >>> 
> > >>>> From: Dominique Quatravaux <dominique@quatravaux.org>
> > >>>> Date: Tue, 6 Feb 2024 18:32:27 +0100
> > >>>> 
> > >>>> When invoking `prog-fill-reindent-defun' from inside a bash-ts-mode buffer, an exception (regardless of where
> > >>>> point is), an error is raised that reads
> > >>>> 
> > >>>> Invalid predicate, see ‘treesit-thing-settings’ for valid forms for a predicate: nil
> > >>>> 
> > >>>> Investigation reveals that `prog-fill-reindent-defun' calls `treesit-node-match-p' with 'text as the second
> > >>>> argument, which can only succeed if `(assq 'text (assq 'bash treesit-thing-settings))' is non-nil; whereas the
> > >>>> corresponding setting for `treesit-thing-settings' prepared by `bash-ts-mode', doesn't provide a value for 'text
> > >>>> (only for 'sentence).
> > >>>> 
> > >>>> I had success applying the following workaround:
> > >>>> 
> > >>>> (advice-add 'bash-ts-mode
> > >>>>            :after (lambda (&rest ignored)
> > >>>>                     (when-let ((bash-things (assq 'bash treesit-thing-settings)))
> > >>>>                       (setf (alist-get 'text (cdr bash-things)) '(sentence)))))
> > >>>> 
> > >>>> Now M-q does the right thing for me everywhere in bash-ts-mode.
> > >>> 
> > >>> Yuan, any comments or ideas?
> > >>> 
> > >>> Thanks.
> > > 
> > > Sorry for the delay. Dominique, what version are you using? I can see
> > > that this is fixed on master[1], but not on emacs-29, perhaps due to
> > > the size of the change.
> > > 
> > > [1] b392d05089014ac6d41e35a016d46f496964f879
> > > 
> > > Yuan
> > 
> > 
> 
> 
> 
>