bug#72511: 30.0.50; prefix-completions is always nil in help--symbol-completion-table

[Chris Roberts via "Bug reports for GNU Emacs, the Swiss army knife of text editors" <bug-gnu-emacs@gnu.org>]

On lines 209-214 inside the function `help--symbol-complete-table' in
help-fns.el there is a following block of code:

(when help-enable-completion-autoload
  (let ((prefixes (radix-tree-prefixes (help-definition-prefixes) string)))
    (help--load-prefixes prefixes)))
(let ((prefix-completions
       (and help-enable-completion-autoload
            (mapcar #'intern (all-completions string definition-prefixes)))))

By default, `help-enable-completion-autoload' is t, so this code should
normally run. The perceived problem here is that the definition of
`help-definition-prefixes' function is as follows:

(defun help-definition-prefixes ()
  "Return the up-to-date radix-tree form of `definition-prefixes'."
  (when (> (hash-table-count definition-prefixes) 0)
    (maphash (lambda (prefix files)
               (let ((old (radix-tree-lookup help-definition-prefixes prefix)))
                 (setq help-definition-prefixes
                       (radix-tree-insert help-definition-prefixes
                                          prefix (append old files)))))
             definition-prefixes)
    (clrhash definition-prefixes))
  help-definition-prefixes)

Because of the `clrhash', `definition-prefixes' will always be empty
after the function call to `help-definition-prefixes'. So in practice,
`definition-prefixes' will always have zero elements when we're trying
to calculate `prefix-completions', which makes the calls to
`all-completions' and `mapcar' basically no-ops. The end result is that
`prefix-completions' will always be NIL.

This can be verified by evaluating the following code in emacs -Q:

(require 'radix-tree)
(require 'help-fns)

(let ((string "string-remove-prefix"))
  (format "Definition prefixes count before: %d\n%sDefinition prefixes count after: %d\nPrefix completions: %s\n"
          (hash-table-count definition-prefixes)
          (or
           (when help-enable-completion-autoload
             (let ((prefixes (radix-tree-prefixes (help-definition-prefixes) string)))
               (help--load-prefixes prefixes)))
           "")
          (hash-table-count definition-prefixes)
          (let ((prefix-completions
                 (and help-enable-completion-autoload
                      (mapcar #'intern (all-completions string definition-prefixes)))))
            prefix-completions)))

This has been first introduced in commit fd8084a (Automatically find
vars and functions via definition-prefixes)

I'm not sure if this was the intended behavior or not, but it seemed
suspicious to me, so I decided it's better to report it.

It might also be useful to note that `prefix-completion' is used in the
following way:

(complete-with-action action obarray string
                      (if pred (lambda (sym)
                                 (or (funcall pred sym)
                                     (memq sym prefix-completions)))))

I'm not sure what `pred' will be in this context, or what impact
`prefix-completions' being always NIL will have, if any. But I believe
that at the very least the code pertaining to `prefix-completions' could
concievably be removed.

Also, was clearing out the entirety of `definition-prefixes' on completion
really the intended behavior? As it stands, just C-h f s <TAB> is enough
to completely truncate it.

In GNU Emacs 30.0.50 (build 1, x86_64-pc-linux-gnu, GTK+ Version
 3.24.33, cairo version 1.16.0) of 2024-01-09 built on cdr
Repository revision: 774c8ec74c98d69d56b2511a613145f2b69fb2eb
Repository branch: master
Windowing system distributor 'The X.Org Foundation', version
11.0.12101004 System Description: Linux Mint 21.2

Configured features:
CAIRO DBUS FREETYPE GIF GLIB GMP GNUTLS GSETTINGS HARFBUZZ JPEG
LIBSELINUX LIBXML2 MODULES NOTIFY INOTIFY PDUMPER PNG SECCOMP SOUND
SQLITE3 THREADS TIFF TOOLKIT_SCROLL_BARS X11 XDBE XIM XINPUT2 XPM GTK3
ZLIB