0. emacs -Q
1. C-h f ert-font-lock-deftest RET

*Help* now reports the macro's calling convention as:

  (ert-font-lock-deftest NAME () [DOCSTRING] ...)

but the second () argument is not actually supported by the macro, whose
arglist reads:

  (name &rest docstring-keys-mode-and-str)

Here, ert-font-lock--parse-macro-args expects
docstring-keys-mode-and-str to start with an (optional) string,
not the empty arglist '().

The symptom of this mismatch is that calls to ert-font-lock-deftest are
misindented (malindented? :) and misfontified.  For example, in
ert-font-lock-tests.el:

  (ert-font-lock-deftest test-macro-test--correct-highlighting
      emacs-lisp-mode
    "
  (defun fun ())
  ;; ^ font-lock-keyword-face
  ;;      ^ font-lock-function-name-face")

Here, the string literal has font-lock-doc-face instead of
font-lock-string-face.

  (ert-font-lock-deftest test-macro-test--docstring
      "A test with a docstring."
    emacs-lisp-mode
    "
  (defun fun ())
  ;; ^ font-lock-keyword-face"
    )

Here, both strings have font-lock-string-face, whereas the first should
have font-lock-doc-face.

Finally, if a test actually followed ert-font-lock-deftest's docstring
in including an empty arglist (), then ert-describe-test would claim
that it is not documented.

The attached patch tries to address these issues.
WDYT?

Thanks,
-- 
Basil