3 ms·

A self-infecting datatype in Lisp

52 points by gnosis 15y ago

guard-of-terra 15y ago
This looks ubercool until you realise that Clojure does this for a living with its seq protocol. Then it's just neat.
lispm 15y ago
Nested DEFUNs is not the best idea...
- AUmrysh 15y ago
  As a CL newbie, what would you suggest be used instead? Also, why would you not use nested defuns?
  - gumbo72 15y ago
    Also interested in this
  - draven 15y ago
    labels? http://www.lispworks.com/documentation/HyperSpec/Body/s_flet_.htm http://www.lispworks.com/documentation/HyperSpec/Body/s_flet... My guess: the nested defun is not only defined in the enclosing defun scope as it may seem, but in the whole current package. http://www.lispworks.com/documentation/HyperSpec/Body/m_defun.htm http://www.lispworks.com/documentation/HyperSpec/Body/m_defu... "Defines a new function named function-name in the global environment" Is it only a bad idea because it's bad style or is there more to it?
draven 15y ago
I'm learning Haskell ATM and it looks similar to lazy lists in WHNF, as at each step you get an element and a "thunk" that can be used to compute the next step. The code for the cycle Haskell function is really simple: http://hackage.haskell.org/packages/archive/base/latest/doc/html/src/GHC-List.html#cycle http://hackage.haskell.org/packages/archive/base/latest/doc/...
- eridius 15y ago
  This is similar to one of my favorite Haskell "tricks": Tying the Knot. It's kind of a mindfuck the first time you do it, but it also feels really really cool. http://www.haskell.org/haskellwiki/Tying_the_Knot http://www.haskell.org/haskellwiki/Tying_the_Knot
SeanLuke 15y ago
Since he mentioned self-infecting Lisp functions (which seems to me to be unrelated to his main topic but whatever), I should mention that I put together a Lisp self-infecting "virus" (so to speak) a long time ago. My original version -- which was beautiful and I have conveniently forgotten -- used a quine to deliver the payload. But here's a version I whipped up for HN a few weeks back which gives you the idea. ;; run this on SBCL (sb-ext:unlock-package :common-lisp) (let (syms) (do-symbols (s :common-lisp) (if (fboundp s) (push s syms))) (labels ((worm (sym) (let ((next (elt syms (random (length syms)))) (old (symbol-function sym))) (format t "~%Worm attacks ~a, next target is ~a!~%" sym next) (setf (symbol-function sym) (lambda (&rest args) (when next (worm next) (setf next nil)) (apply old args)))))) (worm '+))) ;; now try adding some numbers (and so on)

52 points by gnosis 15y ago

guard-of-terra 15y ago

This looks ubercool until you realise that Clojure does this for a living with its seq protocol. Then it's just neat.

lispm 15y ago

Nested DEFUNs is not the best idea...

AUmrysh 15y ago
As a CL newbie, what would you suggest be used instead? Also, why would you not use nested defuns?
- gumbo72 15y ago
  Also interested in this
- draven 15y ago
  labels? http://www.lispworks.com/documentation/HyperSpec/Body/s_flet_.htm http://www.lispworks.com/documentation/HyperSpec/Body/s_flet... My guess: the nested defun is not only defined in the enclosing defun scope as it may seem, but in the whole current package. http://www.lispworks.com/documentation/HyperSpec/Body/m_defun.htm http://www.lispworks.com/documentation/HyperSpec/Body/m_defu... "Defines a new function named function-name in the global environment" Is it only a bad idea because it's bad style or is there more to it?

draven 15y ago

I'm learning Haskell ATM and it looks similar to lazy lists in WHNF, as at each step you get an element and a "thunk" that can be used to compute the next step. The code for the cycle Haskell function is really simple: http://hackage.haskell.org/packages/archive/base/latest/doc/html/src/GHC-List.html#cycle http://hackage.haskell.org/packages/archive/base/latest/doc/...

eridius 15y ago
This is similar to one of my favorite Haskell "tricks": Tying the Knot. It's kind of a mindfuck the first time you do it, but it also feels really really cool. http://www.haskell.org/haskellwiki/Tying_the_Knot http://www.haskell.org/haskellwiki/Tying_the_Knot

SeanLuke 15y ago

Since he mentioned self-infecting Lisp functions (which seems to me to be unrelated to his main topic but whatever), I should mention that I put together a Lisp self-infecting "virus" (so to speak) a long time ago. My original version -- which was beautiful and I have conveniently forgotten -- used a quine to deliver the payload. But here's a version I whipped up for HN a few weeks back which gives you the idea. ;; run this on SBCL (sb-ext:unlock-package :common-lisp) (let (syms) (do-symbols (s :common-lisp) (if (fboundp s) (push s syms))) (labels ((worm (sym) (let ((next (elt syms (random (length syms)))) (old (symbol-function sym))) (format t "~%Worm attacks ~a, next target is ~a!~%" sym next) (setf (symbol-function sym) (lambda (&rest args) (when next (worm next) (setf next nil)) (apply old args)))))) (worm '+))) ;; now try adding some numbers (and so on)