Making Emacs More Presentable


I do occasionally hold talks, mostly about Lisp-related topics. My medium of choice is PDF, as generated by Org’s export with the Beamer backend. When it’s demonstration time, Emacs isn’t nearly as simple to adjust for comfortable viewing as a browser or terminal. My first instinct was to look for a function that allows increasing the font size, similar to C-+. It turns out that C-x C-+ is a thing, however it’s not ideal as it only increases the font size of the current buffer. A quick look at the sources reveals why:

(define-minor-mode text-scale-mode
  :lighter (" " text-scale-mode-lighter)
  (when text-scale-mode-remapping
    (face-remap-remove-relative text-scale-mode-remapping))
  (setq text-scale-mode-lighter
        (format (if (>= text-scale-mode-amount 0) "+%d" "%d")
  (setq text-scale-mode-remapping
        (and text-scale-mode
             (face-remap-add-relative 'default
                                          (expt text-scale-mode-step
  (force-window-update (current-buffer)))

text-scale-mode is implemented in terms of face-remap-add-relative, a function describing itself as “Add a face remapping entry of FACE to SPECS in the current buffer.”. Funnily enough, both live in face-remap.el, probably because scaling text is merely a demonstration of the buffer-local face remapping capabilities of Emacs. While it’s a cute demo, it’s clearly not what I’d want from a C-+ replacement, so I wrote an alternative solution operating on the frame:

(defun my-alter-frame-font-size (fn)
  (let* ((current-font-name (frame-parameter nil 'font))
         (decomposed-font-name (x-decompose-font-name current-font-name))
         (font-size (string-to-int (aref decomposed-font-name 5))))
    (aset decomposed-font-name 5 (int-to-string (funcall fn font-size)))
    (set-frame-font (x-compose-font-name decomposed-font-name))))

(defun my-inc-frame-font-size ()
  (my-alter-frame-font-size '1+))

(defun my-dec-frame-font-size ()
  (my-alter-frame-font-size '1-))

(global-set-key (kbd "C-+") 'my-inc-frame-font-size)
(global-set-key (kbd "C-=") 'my-inc-frame-font-size)
(global-set-key (kbd "C--") 'my-dec-frame-font-size)

This is a bit less hacky, but still disgusting. The code fetches the font name (which curiously comes in the XLFD notation) from the frame, converts it into an array, extracts the font size, manipulates it, puts it back into the array, converts it to a font name and sets the frame’s font to it. You can find this snippet and many more in my, so if you haven’t already, give it a look to find more goodies!

Trapping Attackers With Nyan Cat


In case you haven’t done it yet, I strongly recommend you to give telnet a try. If you don’t have a telnet client ready, head over to instead and enjoy the pretty pictures.

I’ve wondered for quite some time whether it would be possible to run the same thing on a SSH server. It turns out that it’s not too hard to do as not only Kevin Lange’s creation can be run in a TTY, but OpenSSH allows you to do something else than giving you a shell after a successful authentication attempt.

First of all you’ll need to install and test the nyancat program:

% pacman -S nyancat
$ nyancat

To restrict the impact of the public-facing service, I decided to create a new user for it and run a separate SSH daemon with its own config and service file:

% useradd -m -s /bin/sh anonymous
% cp /usr/lib/systemd/system/sshd.service /etc/systemd/system/nyanpot.service
% cp /etc/ssh/sshd_config /etc/ssh/nyanpot_sshd_config

Relevant changed bits in the service file:

Description=OpenSSH Honeypot
ExecStart=/usr/bin/sshd -D -f /etc/ssh/nyanpot_sshd_config

The SSH config is a bit special as it locks out everyone who isn’t the anonymous user:

Port 22
PermitRootLogin No
PermitTTY No
PasswordAuthentication No
X11Forwarding No
AllowTcpForwarding No
Match User anonymous
    PasswordAuthentication Yes
    PermitTTY Yes
    ForceCommand nyancat

You can test the service with systemctl start nyanpot.service and logging in (ideally from a different system) as the anonymous user. If everything works fine, enable the service permanently with systemctl enable nyanpot.service. My honeypot is available via ssh (PW: anonymous). Enjoy!

Brave New World


Update: Finally figured out the layout after digging a bit more into the sources, it’s a QWERTY-UK (see devices/rpi2/uspi/include/uspios.h). Looks like I’ll have to modify the bundled USPI library to include a QWERTY-US layout before I can make any progress on keyboard remapping in Lisp…

I believe I’ve found an even greater time sink than writing Lisp interpreters for fun. Long time ago, I’ve read an encouraging blog post on the future of the LispM, not expecting to find an implementation of the ideas presented therein. Turns out I was wrong about that. Meet Interim OS!

In case you’re wondering why you should possibly care:

  • Small and readable codebase (most of the code is device drivers for the Raspberry Pi)
  • Simple to hack on
  • Plan9-style APIs
  • Minimal Lisp dialect
  • Runs on your favorite desktop OS in hosted mode, that is, safely contained to a terminal with the ability to spawn graphical windows
  • Runs on bare metal (Raspberry Pi 2)

Getting it to run in hosted mode is simple enough, so I won’t explain it here. Booting on bare metal however is a different story, so here we go:

$ git clone
$ cp interim/docs/interim-0.1.0-rpi2.tgz ./
$ bsdtar -xf interim-0.1.0-rpi2.tgz # cry me a river
% mkdir /media/boot
% mount /dev/sdXN /media/boot
% cp release-rpi2/* /media/boot/
% rm /media/boot/cmdline.txt
% umount /media/boot
  • Plug in the SDHC card, a HDMI monitor and a USB keyboard
  • Optionally: Plug in a network cable and/or a USB mouse
  • Power up

You’ll be greeted by a “Welcome to Interim OS” and dropped into a promptless shell. If you’re unlucky, the chosen resolution may be unreadable, so feel free to retry this process a few times. The keyboard layout is hardcoded and somewhere between QWERTY-US and QWERTZ-DE, something I intend to fix soon. For basic usage instructions, type (bytes->str (load "/sd/hello.txt")) and hit the enter key. Happy hacking!

On Minimalism


Update: There is a CL merge request that supports SBCL among other implementations, as expected it beats the picolisp implementation in speed by a factor of 2x. Still, I’m fine with being second place :)

I’ve implemented MAL for the third time by now, this time in PicoLisp, a language priding itself on its implementation simplicity. While it clearly is a Lisp dialect, it has foregone a good amount of classic Lisp design choices in favor of terse code. Despite this, there are practical inclusions for writing application software, like the GUI system and a distributed database implementation with a Prolog-style query language. Other interesting features are an unobtrusive OOP system, a FFI for C and Java, live debugging utilities, pattern matching and more in a 1MiB tarball.

You might wonder why I’d be up for learning yet another Lisp dialect, after having learned Emacs Lisp, Clojure and Scheme. Furthermore, Scheme already claims to take minimalism as language design principle and of course there are more obscure Lisp dialects, like Arc and newLISP. I can only blame a friend who told me about this fascinating talk given by the PicoLisp author demonstrating the abilities of his programming language. The overall picture my friend painted was that of a bizarro world where a crazy German ignored all established rules for a Lisp dialect, walking the fine line between insanity and practicability. Most surprisingly though was that he used his own invention to write business applications and succeeded in making a living off it. Naturally I was intrigued and kept PicoLisp on my backlog of things to play with.

My implementation is a bit smaller than the Emacs Lisp one, is the first one to actually make use of GNU readline and went for a purely Lisp tokenizer as I couldn’t figure out how to use PCRE for this task. It also appears to be the fastest one out of all Lisp family implementations. This might change though once the “clisp” implementation gains support for using SBCL instead of CLISP…

Regarding oddities, here’s an incomplete list:

  • No lambda. You pass a quoted list instead.
  • Quote returns more than only its first argument. However (quote 1 2 3) is equivalent to '(1 2 3) so you’ll most likely not ever notice…
  • No macros. Functions can instead not evaluate their arguments, your job is to evaluate them as needed.
  • No strings. String syntax creates so-called “transient” symbols. Additionally to doubling as string replacement, these are not equal to other symbols with the same contents and are therefore used to avoid name clashes in macro-like functions.
  • No implicit closures. If you need to capture something from the environment, you must do this explicitly and have the choice between mutable and immutable ones.
  • Unknown symbols evaluate to NIL instead of throwing an exception.
  • Indentation (and even pretty-printing) is a lot easier than in classic Lisp dialects as it’s basically about increasing the depth by three spaces for each level instead of lining up parentheses.
  • Closing parentheses that do not belong to the current line are separated by spaces. For convenience’s sake, a closing bracket is interpreted as “super paren” and closes all remaining parentheses.
  • The style guide has an interesting solution to the problem of local variables potentially shadowing built-in functions: Capitalized identifiers!
  • NIL is not only equivalent to the empty list, but to the empty string as well. This bit me when wrapping readline as an empty string couldn’t be discerned from NULL with the naïve approach…
  • Error handling is close to non-existing. If you screw up things too much, the interpreter will segfault on you. This is not considered a bug.
  • Identifiers for built-in functions are very short and at times cryptic[1]. Clojure got nothing on that!
[1]read does not parse a string into a S-expression, str does. The result of this cannot be handed to eval either, you’ll need to run it instead. I could go on with this for a while…

Small-time Patching


Update: Added the improved backtrace

Update: Merged!

Today #emacs reminded me of an oddity in Emacs I’ve sort of learned to live with: Backtraces are, well, see for yourself:

Debugger entered--Lisp error: (wrong-type-argument number-or-marker-p t)
  +(1 t)
  eval((+ 1 t) nil)
  eval-expression((+ 1 t) nil)
  call-interactively(eval-expression nil nil)

I can live with errors printed as a list. What I can’t live with is none of the call stack lines being printed as a S-Expression… To fix this, one must dive a bit deeper than usual as only the debugger porcelain is implemented in Emacs Lisp. Its workhorse is backtrace in eval.c:

DEFUN ("backtrace", Fbacktrace, Sbacktrace, 0, 0, "",
       doc: /* Print a trace of Lisp function calls currently active.
Output stream used is value of `standard-output'.  */)
  union specbinding *pdl = backtrace_top ();
  Lisp_Object tem;
  Lisp_Object old_print_level = Vprint_level;

  if (NILP (Vprint_level))
    XSETFASTINT (Vprint_level, 8);

  while (backtrace_p (pdl))
      write_string (backtrace_debug_on_exit (pdl) ? "* " : "  ");
      if (backtrace_nargs (pdl) == UNEVALLED)
          Fprin1 (Fcons (backtrace_function (pdl), *backtrace_args (pdl)),
          write_string ("\n");
          tem = backtrace_function (pdl);
          Fprin1 (tem, Qnil);   /* This can QUIT.  */
          write_string ("(");
            ptrdiff_t i;
            for (i = 0; i < backtrace_nargs (pdl); i++)
                if (i) write_string (" ");
                Fprin1 (backtrace_args (pdl)[i], Qnil);
          write_string (")\n");
      pdl = backtrace_next (pdl);

  Vprint_level = old_print_level;
  return Qnil;

Despite the rather unusual look and weird naming, it’s not too hard to find the culprit. Most of the time is spent inside a loop that walks through the call stack, accesses the top-most function and args with backtrace_function and backtrace_args, prints a lisp object with Fprin1 (which is just another way to use prin1 from C code) and writes out normal strings with write_string. Qnil refers to the global nil symbol, tem is a naming convention for a temporary variable. It should be sufficient to print the opening paren first, then the function, a space and proceed normally from that point:

tem = backtrace_function (pdl);
write_string ("(");
Fprin1 (tem, Qnil); /* This can QUIT.  */
write_string (" ");

Time to recompile Emacs with make, boot the binary with src/emacs -Q, and trigger a backtrace with M-: (+ 1 t). Unfortunately that does not yield a prettier backtrace yet, but rather a Search failed: "\n debug(". As the debugger is busted, I had to resort to ag to find where exactly the breakage occurs. It’s not too hard to fix, mind you, all you have to do is to patch debugger-setup-buffer in debug.el to search for "\n (debug" instead.

The result is the following teensy patch:

diff --git a/lisp/emacs-lisp/debug.el b/lisp/emacs-lisp/debug.el
index 22a3f39..4020620 100644
--- a/lisp/emacs-lisp/debug.el
+++ b/lisp/emacs-lisp/debug.el
@@ -279,7 +279,7 @@ That buffer should be current already."
   (goto-char (point-min))
   (delete-region (point)
-              (search-forward "\n  debug(")
+              (search-forward "\n  (debug")
               (forward-line (if (eq (car args) 'debug)
                                      ;; Remove debug--implement-debug-on-entry
                                      ;; and the advice's `apply' frame.
diff --git a/src/eval.c b/src/eval.c
index 72facd5..e32e7a1 100644
--- a/src/eval.c
+++ b/src/eval.c
@@ -3409,8 +3409,9 @@ Output stream used is value of `standard-output'.  */)
      tem = backtrace_function (pdl);
-     Fprin1 (tem, Qnil);   /* This can QUIT.  */
      write_string ("(");
+     Fprin1 (tem, Qnil);   /* This can QUIT.  */
+     write_string (" ");
        ptrdiff_t i;
        for (i = 0; i < backtrace_nargs (pdl); i++)

And an IMHO vastly improved backtrace:

Debugger entered--Lisp error: (wrong-type-argument number-or-marker-p t)
  (debug error (wrong-type-argument number-or-marker-p t))
  (+ 1 t)
  (eval (+ 1 t) nil)
  (eval-expression (+ 1 t) nil)
  (funcall-interactively eval-expression (+ 1 t) nil)
  (call-interactively eval-expression nil nil)
  (command-execute eval-expression)

Not sure whether to bother submitting this… Let me know what you think!