6 ms·

That looks like lisp?

by Kinnard 11y ago

adamnemecek 11y ago
Programming in lisp is basically manipulating the ast.
- PeCaN 11y ago
  Not just "basically", that actually is the so-called ‘Zen of Lisp’. With something like paredit, your editing isn't text editing so much as manipulating the AST in a very literal sense. Then you can write macros to do that for you, since programmers are lazy and Lisp programmers are particularly so.
Mikeb85 11y ago
It's because Lisp is basically an abstract syntax tree.
chumich1 11y ago
You are right! Lisp is homoiconic which basically means that the language mirrors the AST https://en.wikipedia.org/wiki/Homoiconicity https://en.wikipedia.org/wiki/Homoiconicity
- Kinnard 11y ago
  Hmmm, I am learning a lisp now I've been reluctant to dive in but this makes me all the more willing
  - yjgyhj 11y ago
    It's simpler than it seems, and how strange people make it out to be held me off for a long time. There are cons cells. A cons cell is basically an untyped tuple. The syntax for a cons cell containing a number look like this: (1 nil) (where nil is the zero byte (I think, please correct me if I'm wrong)). You can nest them, like this: (1 (2 (3 nil))). Because nobody can be bothered to type those parens, the part of the compiler/interpreter called the Reader compiles syntax like this (1 2 3) to this (1 (2 (3 nil))). After being parsed by "the reader" (you could also call it "parser"), those nested tuples/cons cells are sent to eval function. The eval function takes the first element of the list (aka `car` in lisp-speak. My mnemonic is that a is to the left of the keyboard, therefore means first element). That first element is being regarded as the function, and the second element in the cons cell is the argument to the function. The function to get the second item in the cons cell is `cdr`. I remember that by having `d` being to the right on my keyboard. (car 1 2) => 1. (cdr 1 2) => 2. (cdr 1 (2 (3 nil))) => (2 (3 nil)). That is how things are nested. I digress. The eval function evaluates every symbol (aka atom) in these nested cons cells, and looks up the corresponding meaning of them in a big table of defined symbols. So the + symbol might match an addition function, the string-join symbol might match to a function for joining strings. What is returned from eval you could say is the true AST, which still looks pretty darn similar to the lisp syntax. The function is then sent to the apply function, which applies the argument to the function. Remember that the argument can be a nested set of cons cell, so it can nest infinitely. That's it! I'm no lisp expert, so if someone is, please correct me. But that's the gist of it. Syntax is read by the read function, the symbols are interpreted (lookup in a big key-value object) by the eval function, these function gets apply-ed with their argument. Very few primitives are needed by build a system from that. Here is a list of the primitives needed - these can be implemented in binary, C, assembly, Java, Javascript or whatever: http://stackoverflow.com/a/3482883 http://stackoverflow.com/a/3482883. BONUS SYNTAX: If you want a eval to skip over a cons cell (and of course it's children), you can prepend it with a `'` symbol. So (cdr 1 '(this is never going to be evaled, but is just a list)) => (this is never going to be evaled, but is just a list) This is called a 'quote'. Any Javascript array is basically working like a quoted list. If the first item in a javascript array was a function, and you applied the cdr of that array as argument to that funciton, it would be like using lisp with an not-quoted list. BONUS VIDEO: This is just lovely. I'm not American, and do not have a CS degree, so I feel like attending a place like this is a far dream. But happy to be able to enjoy it over the interwebs. Incredibly thankful to the giants of computing on whose shoulders we can stand https://www.youtube.com/watch?v=2Op3QLzMgSY&list=PLF4E3E1B72A58B492 https://www.youtube.com/watch?v=2Op3QLzMgSY&list=PLF4E3E1B72...
    - dragonwriter 11y ago
      A note on notation: The standard notation for a cons cell with car "x" and cdr "y" is (x . y), not (x y). The latter is a list, equivalent of (x . (y . nil)). So (1 2 3) is equivalent to (1 . (2 . (3 . nil)), not (1 (2 (3 nil))). The latter is a nested list, which is equivalent to (1 . ((2 . ((3 . (nil . nil)) . nil)) . nil))
      - yjgyhj 11y ago
        Good catch, thank you. To anyone who wants to play around with lisp, I recommend downloading the emacs text editor. Just open it, navigate around with the arrow keys. Type an s-expression, and place the point after the ")". Then type Ctrl-x Ctrl-e, and the expression will be sent thru read -> eval -> apply. Fun to just play around with. (+ 1 2)_ ;; _ means the cursor, ";" means a comment
    - yjgyhj 11y ago
      BONUS FUN: If you want to play around with lisp interactively, I reccomend checking out the program Emacs. It's a little lisp interpreter written in C, that comes with a text editor and such. Download it and run it. When you write some lisp code into the editor, place your cursor after the expression and hit Ctrl-x Ctrl-e. That is read, eval and apply what the expression, and display the result in the bottom of them window. Here is a screenshot of me doing that with this code (apply (quote +) (quote (1 2))) ;; http://i.imgur.com/SrhMxS8.png http://i.imgur.com/SrhMxS8.png Play around with that - it's fun enough for an evening. Some other fun code to run is this (reduce '+ '(1 2 3 4)) (defun say-hello (&optional name) ;; a wild lambda appeared! (if (stringp name) (concat "hello " name) "I'm a lonely program")) (say-hello "Martin") (say-hello)
  - bsaul 11y ago
    Your remark made me curious, because i initialy felt the same. Then, i thought, well i still find the original swift much more readable, so isn't it like prefering to code in ASM because you're seen compiled code result ?
lispm 11y ago
No, it looks like some kind of s-expression version of the Swift AST. It could have been in XML format or something else. Lists seem to be popular for it - somehow. Not everything which uses a notation similar to s-expressions has anything to do Lisp.
- outworlder 11y ago
  True, but it could be turned into executable Lisp code without much effort. As could a XML format version of it. Besides, square brackets is the most common representation of lists. It's difficult to find round brackets denoting lists outside of Lisp and dialets. I doubt that's a coincidence. Given that Apple already uses Scheme in its sandbox implementation [1], it wouldn't be too outlandish to assume those are related somehow. The notation is a bit off, but it's nothing a few macros wouldn't solve. [1] http://lemonodor.com/archives/2007/10/sexprs_in_leopard.html http://lemonodor.com/archives/2007/10/sexprs_in_leopard.html
  - lispm 11y ago
    > True, but it could be turned into executable Lisp code without much effort Unlikely. > Given that Apple already uses Scheme in its sandbox implementation [1], it wouldn't be too outlandish to assume those are related somehow. The notation is a bit off, but it's nothing a few macros wouldn't solve. They both use parentheses. They are related. Somehow.

That looks like lisp?

by Kinnard 11y ago

That looks like lisp?

adamnemecek 11y ago

Programming in lisp is basically manipulating the ast.

PeCaN 11y ago
Not just "basically", that actually is the so-called ‘Zen of Lisp’. With something like paredit, your editing isn't text editing so much as manipulating the AST in a very literal sense. Then you can write macros to do that for you, since programmers are lazy and Lisp programmers are particularly so.

Mikeb85 11y ago

It's because Lisp is basically an abstract syntax tree.

chumich1 11y ago

You are right! Lisp is homoiconic which basically means that the language mirrors the AST https://en.wikipedia.org/wiki/Homoiconicity https://en.wikipedia.org/wiki/Homoiconicity

Kinnard 11y ago
Hmmm, I am learning a lisp now I've been reluctant to dive in but this makes me all the more willing
- yjgyhj 11y ago
  It's simpler than it seems, and how strange people make it out to be held me off for a long time. There are cons cells. A cons cell is basically an untyped tuple. The syntax for a cons cell containing a number look like this: (1 nil) (where nil is the zero byte (I think, please correct me if I'm wrong)). You can nest them, like this: (1 (2 (3 nil))). Because nobody can be bothered to type those parens, the part of the compiler/interpreter called the Reader compiles syntax like this (1 2 3) to this (1 (2 (3 nil))). After being parsed by "the reader" (you could also call it "parser"), those nested tuples/cons cells are sent to eval function. The eval function takes the first element of the list (aka `car` in lisp-speak. My mnemonic is that a is to the left of the keyboard, therefore means first element). That first element is being regarded as the function, and the second element in the cons cell is the argument to the function. The function to get the second item in the cons cell is `cdr`. I remember that by having `d` being to the right on my keyboard. (car 1 2) => 1. (cdr 1 2) => 2. (cdr 1 (2 (3 nil))) => (2 (3 nil)). That is how things are nested. I digress. The eval function evaluates every symbol (aka atom) in these nested cons cells, and looks up the corresponding meaning of them in a big table of defined symbols. So the + symbol might match an addition function, the string-join symbol might match to a function for joining strings. What is returned from eval you could say is the true AST, which still looks pretty darn similar to the lisp syntax. The function is then sent to the apply function, which applies the argument to the function. Remember that the argument can be a nested set of cons cell, so it can nest infinitely. That's it! I'm no lisp expert, so if someone is, please correct me. But that's the gist of it. Syntax is read by the read function, the symbols are interpreted (lookup in a big key-value object) by the eval function, these function gets apply-ed with their argument. Very few primitives are needed by build a system from that. Here is a list of the primitives needed - these can be implemented in binary, C, assembly, Java, Javascript or whatever: http://stackoverflow.com/a/3482883 http://stackoverflow.com/a/3482883. BONUS SYNTAX: If you want a eval to skip over a cons cell (and of course it's children), you can prepend it with a `'` symbol. So (cdr 1 '(this is never going to be evaled, but is just a list)) => (this is never going to be evaled, but is just a list) This is called a 'quote'. Any Javascript array is basically working like a quoted list. If the first item in a javascript array was a function, and you applied the cdr of that array as argument to that funciton, it would be like using lisp with an not-quoted list. BONUS VIDEO: This is just lovely. I'm not American, and do not have a CS degree, so I feel like attending a place like this is a far dream. But happy to be able to enjoy it over the interwebs. Incredibly thankful to the giants of computing on whose shoulders we can stand https://www.youtube.com/watch?v=2Op3QLzMgSY&list=PLF4E3E1B72A58B492 https://www.youtube.com/watch?v=2Op3QLzMgSY&list=PLF4E3E1B72...
  - dragonwriter 11y ago
    A note on notation: The standard notation for a cons cell with car "x" and cdr "y" is (x . y), not (x y). The latter is a list, equivalent of (x . (y . nil)). So (1 2 3) is equivalent to (1 . (2 . (3 . nil)), not (1 (2 (3 nil))). The latter is a nested list, which is equivalent to (1 . ((2 . ((3 . (nil . nil)) . nil)) . nil))
    - yjgyhj 11y ago
      Good catch, thank you. To anyone who wants to play around with lisp, I recommend downloading the emacs text editor. Just open it, navigate around with the arrow keys. Type an s-expression, and place the point after the ")". Then type Ctrl-x Ctrl-e, and the expression will be sent thru read -> eval -> apply. Fun to just play around with. (+ 1 2)_ ;; _ means the cursor, ";" means a comment
  - yjgyhj 11y ago
    BONUS FUN: If you want to play around with lisp interactively, I reccomend checking out the program Emacs. It's a little lisp interpreter written in C, that comes with a text editor and such. Download it and run it. When you write some lisp code into the editor, place your cursor after the expression and hit Ctrl-x Ctrl-e. That is read, eval and apply what the expression, and display the result in the bottom of them window. Here is a screenshot of me doing that with this code (apply (quote +) (quote (1 2))) ;; http://i.imgur.com/SrhMxS8.png http://i.imgur.com/SrhMxS8.png Play around with that - it's fun enough for an evening. Some other fun code to run is this (reduce '+ '(1 2 3 4)) (defun say-hello (&optional name) ;; a wild lambda appeared! (if (stringp name) (concat "hello " name) "I'm a lonely program")) (say-hello "Martin") (say-hello)
- bsaul 11y ago
  Your remark made me curious, because i initialy felt the same. Then, i thought, well i still find the original swift much more readable, so isn't it like prefering to code in ASM because you're seen compiled code result ?

lispm 11y ago

No, it looks like some kind of s-expression version of the Swift AST. It could have been in XML format or something else. Lists seem to be popular for it - somehow. Not everything which uses a notation similar to s-expressions has anything to do Lisp.

outworlder 11y ago
True, but it could be turned into executable Lisp code without much effort. As could a XML format version of it. Besides, square brackets is the most common representation of lists. It's difficult to find round brackets denoting lists outside of Lisp and dialets. I doubt that's a coincidence. Given that Apple already uses Scheme in its sandbox implementation [1], it wouldn't be too outlandish to assume those are related somehow. The notation is a bit off, but it's nothing a few macros wouldn't solve. [1] http://lemonodor.com/archives/2007/10/sexprs_in_leopard.html http://lemonodor.com/archives/2007/10/sexprs_in_leopard.html
- lispm 11y ago
  > True, but it could be turned into executable Lisp code without much effort Unlikely. > Given that Apple already uses Scheme in its sandbox implementation [1], it wouldn't be too outlandish to assume those are related somehow. The notation is a bit off, but it's nothing a few macros wouldn't solve. They both use parentheses. They are related. Somehow.