Giving faces compliant names

Question

The description of defface in (info "(elisp) Defining Faces") says that the name of a face should not end in -face. But most faces declared in font-lock.el do use the suffix -face. How does this fit together?

Answer 1

There are exceptions. ;-)

Some may be vestiges. In the Olden Days it was common to have both a variable for the face and the face itself, and the variable, or both, had a name that ended in -face.

Some are not just vestiges, at least for the variables. One reason to have a variable is to be able to bind it, to temporarily change which face is used. This is not common.

In the case of font-lock faces, there is this comment (from @Stefan Monnier) in the code (font-lock.el), which explains things:

;; Originally these variable values were face names such as `bold' etc.
;; Now we create our own faces, but we keep these variables for compatibility
;; and they give users another mechanism for changing face appearance.
;; We now allow a FACENAME in `font-lock-keywords' to be any expression that
;; returns a face.  So the easiest thing is to continue using these variables,
;; rather than sometimes evalling FACENAME and sometimes not.  sm.

;; Note that in new code, in the vast majority of cases there is no
;; need to create variables that specify face names.  Simply using
;; faces directly is enough.  Font-lock is not a template to be
;; followed in this area.
(defvar font-lock-comment-face      'font-lock-comment-face
  "Face name to use for comments.")