bug#61726: [PATCH] Eglot: Support positionEncoding capability

[João Távora]

On Sun, Feb 26, 2023 at 5:31 AM Eli Zaretskii <eliz@gnu.org> wrote:
>
> > From: João Távora <joaotavora@gmail.com>
> > Cc: Augusto Stoffel <arstoffel@gmail.com>,  61726@debbugs.gnu.org
> > Date: Sat, 25 Feb 2023 22:13:29 +0000
> >
> > Eli Zaretskii <eliz@gnu.org> writes:
> >
> > >> From: Augusto Stoffel <arstoffel@gmail.com>
> > >> Cc: joaotavora@gmail.com,  61726@debbugs.gnu.org
> > >> Date: Sat, 25 Feb 2023 15:14:06 +0100
> > >>
> > >> On Sat, 25 Feb 2023 at 15:47, Eli Zaretskii wrote:
> > >>
> > >> >> > Can you please humor me and implement eglot-bytewise-column like 
> > >> >> > that?
> > >> >>
> > >> >> I would be glad to do that, but unfortunately I'd have to ask your
> > >> >> advice as to how to make the corresponding adaptation of
> > >> >> eglot-move-to-bytewise-column.
> > >>          ^^^^^^^
> > >
> > > Sorry.  Here:
> > >
> > >   (defun eglot-move-to-bytewise-column (column)
> > >     "Move to COLUMN as computed using the LSP `utf-8' criterion."
> > >     (let* ((bol (line-beginning-position))
> > >        (goal-byte (+ (position-bytes bol) column))
> > >        (eol (line-end-position)))
> > >       (goto-char bol)
> > >       (while (and (< (position-bytes (point)) goal-byte)
> > >               (< (point) eol))
> > >     (if (>= (char-after) #x3fff80)  ; raw bytes take 2 bytes in the buffer
> > >         (setq goal-byte (1+ goal-byte)))
> > >     (forward-char 1))))
> >
> > In eglot-move-to-lsp-abiding-column (the utf-16 sibling of this
> > function) we use a binary search instead of a linear search.  I remember
> > measuring a visible improvement.  I'm not sure the conditions are
> > exactly the same with this one.  Could/should we do the same here?
>
> Fine by me, but optimizing a method that is not yet used sounds a bit
> premature, no?  I won't object, though.

There's nothing to optimize there, there's no benefit to binary search.
I think this type of function, which has some utf-8/16 knowledge
directly in them, is the fastest option, much faster than using
encode-coding-region like I was doing before.

João