Re: macOS metal rendering engine in mac port

[Eli Zaretskii]

> From: Aaron Jensen <aaronjensen@gmail.com>
> Date: Mon, 24 May 2021 09:31:56 -0700
> Cc: emacs-devel@gnu.org, Alan Third <alan@idiocy.org>, 
>       YAMAMOTO Mitsuharu <mituharu@math.s.chiba-u.ac.jp>
> 
> On Mon, May 24, 2021 at 9:29 AM Eli Zaretskii <eliz@gnu.org> wrote:
> >
> > Which faces are the those for which merge_faces was called 40 times,
> > and which faces are those for which it was called 147,882 times.
> 
> I'll have to get back to you on the 40. The 147k appears to be
> entirely line-number and line-number-current-line. If I had to guess,
> Emacs is calling merge-faces with line-number and
> line-number-current-line for every visible line at least once for
> every scroll.

Of course.  But it also does that for every other face, starting with
font-lock faces.  So something is still amiss here, because 40 is too
small a number.

> > Also, what did Emacs do when you collected these numbers, and in which
> > buffer under what major mode?
> 
> emacs -Q src/xdisp.c with scroll-up-benchmark

Then you should see font-lock-comment-face, font-lock-keyword-face,
font-lock-variable-name-face, etc., and calls to merge_face_ref for
them coming from face_at_buffer_position.