Re: macOS metal rendering engine in mac port

[Eli Zaretskii]

> From: Aaron Jensen <aaronjensen@gmail.com>
> Date: Sun, 30 May 2021 10:00:00 -0700
> Cc: Alan Third <alan@idiocy.org>, emacs-devel@gnu.org
> 
> >   (gdb) break lface_from_face_name_no_resolve
> >   (gdb) commands
> >   > pp face_name
> >   > continue
> >   > end
> 
> This is a sampling of some of the faces and their counts. It's only
> about 300 of the 1300, I'm having tmux copy/paste trouble:
> 
>   41 default
>    6 font-lock-builtin-face
>    4 font-lock-comment-delimiter-face
>    7 font-lock-comment-face
>    1 font-lock-function-name-face
>    6 font-lock-preprocessor-face
>   12 font-lock-string-face
>    1 font-lock-variable-name-face

These are expected: they are faces used by buffer text.

>    1 header-line

Also expected.

>   41 line-number
>   20 line-number-current-line

Expected when line numbers are displayed.

>   48 nano-face-faded
>   21 nano-face-header-ancillary
>   13 nano-face-header-default
>    5 nano-face-header-faded
>   12 nano-face-header-strong
>   12 nano-face-popout
>    6 nano-face-salient
>   14 nano-face-strong

These are most probably due to massive inheritance in the nano theme.

>   41 shadow

This is because line-number and line-number-current-line inherit from
shadow.

> When emacs is idle, it seems to spam face lookups continuously. Here's
> a sample of that:

Is this "emacs -Q"?  If so, does it help to disable blink-cursor-mode
and global-eldoc-mode?

FWIW, I don't see any "spamming" like you do.  If the above 2 minor
modes don't explain that, I would be interested to see backtraces from
some of these lookups.