🚀 go-pugleaf

GNU Emacs 18.53.11 on NeXT 1.0.1

I don't understand why file-of-tag scans ALL the file headers in the
buffer up to the current one.  Isn't \n\f\n guaranteed to occur only
between tag subfiles?  (If not, why is it there at all?)  In which
case, the following code is clearer (also smaller and faster).

It would be nice, by the way, if data structures (e.g. TAGS table,
byte-compile string) were systematically documented somewhere
independently of the code that manipulates them.

(defun file-of-tag ()
  "Return the file name of the file whose tags point is within.
Assumes the tag table is the current buffer.
File name returned is relative to tag table file's directory."
    (save-excursion
      (search-backward "\n\f\n")
      (forward-char 3)
      (buffer-substring (point)
			(progn (skip-chars-forward "^,\n")
			       (point)))))

If we can assume that \f appears only between tag subfiles, then:

(defun file-of-tag ()
  "Return the file name of the file whose tags point is within.
Assumes the tag table is the current buffer.
File name returned is relative to tag table file's directory."
    (save-excursion
      (skip-chars-backward "^\f")
      (forward-char 1)
      (buffer-substring (point)
			(progn (skip-chars-forward "^,\n")
			       (point)))))