Re: An anonymous IRC user's opinion

[Dmitry Gutov]

On 06/11/2024 15:25, Eli Zaretskii wrote:
> > Date: Wed, 06 Nov 2024 14:46:48 +0200
> > From: "Dmitry Gutov" <dmitry@gutov.dev>
> > Cc: johan.myreen@gmail.com, emacs-devel@gnu.org
> >
> >
> > On Wed, Nov 6, 2024, at 2:15 PM, Eli Zaretskii wrote:
> >
> >   > Date: Tue, 5 Nov 2024 22:59:09 +0200
> >   > Cc: johan.myreen@gmail.com, emacs-devel@gnu.org
> >   > From: Dmitry Gutov <dmitry@gutov.dev>
> >   >
> >   > On 05/11/2024 21:53, Eli Zaretskii wrote:
> >   > >> Suppose we have this in css-mode.el:
> >   > >>
> >   > >>     ;;;###autoload
> >   > >>     (if (and (treesit-available-p)
> >   > >>              treesit-modes-enabled)
> >   > >>         (add-to-list 'auto-mode-alist '("\\.css\\'" . css-ts-mode)))
> >   > >>
> >   > >> And treesit-mode-enabled is originally nil.
> >   > >>
> >   > >> Then the user customizes it to t.
> >   > >>
> >   > >> What does its setter do? For css-ts-mode and other modes.
> >   > >
> >   > > It could turn the mode on in every buffer that visits a .css file.  Or
> >   > > not.
> >   >
> >   > Based on which information?
> >
> >   buffer-file-name and auto-mode-alist, I guess?
> >
> > But auto-mode-alist won't have any relevant entries because when its form 
> > (above) was evaluated,
> > treesit-modes-enabled was nil.
>
> Are you saying that it is impossible to turn on the mode in relevant
> buffers?  I find that hard to believe, but if you are right, then we
> could refrain from doing this retroactively.
>
> (And I still don't understand where this discussion is going.  What
> point are you trying to make?)

I'm not trying to make a point, but to find the right tradeoffs.

You wanted a user option - sure, no problem. But if taken the most 
straightforward approach, the option would only have effect after 
restart, and not on the current session. Otherwise it would need more 
information available somehow. You asked which data - the description 
was in the previous messages.