7 ms·

On being startled by a programming language

66 points by sbierwagen 16y ago

sofuture 16y ago
J is awesome. GPL'd J is even better! Some of us are trying to organize our efforts here: https://github.com/openj/core https://github.com/openj/core We don't intend on mucking with the code much at this point, just patching together a better build system, etc.
- jameskilton 16y ago
  I think I'm more amazed at the C code. Is that obfuscated or a decompiled dump? Or is that how the core of J is really written?
  - sofuture 16y ago
    Hah! That is how the code is really written. It's... a unique style, shared amongst the APL languages. I'm pretty sure there are about two people that can read and write that fluently: Roger Hui and Arthur Whitney.
    - swah 16y ago
      Btw, is this related in anyway to Arthur?
      - kragen 16y ago
        Apparently Arthur Whitney wrote a one-page interpreter of an APL dialect in C (in the same style!) which inspired Ken Iverson to write J. http://keiapl.org/rhui/remember.htm#40 http://keiapl.org/rhui/remember.htm#40
        swah 16y ago
        He is also the author of the K language.
    - jfager 16y ago
      As a wholly unaffiliated, self-selected, and entirely unofficial spokesperson, I'm happy to bestow this lifetime achievement award on behalf of the IOCCC.
      - narrator 16y ago
        People who like APL do not often code with readability as their number 1 priority. In fact, to code in APL one must have an outright disdain for the idea that someone else could easily understand their code.
        gnosis 16y ago
        With that kind of attitude, they might enjoy programming in Brainfuck or Malbolge even more.
        silentbicycle 16y ago
        No, we just have a different idea of what's readable. APL was created to have a mathematical notation that made problem solving easier. Don't believe me? See Iverson's Turing Award lecture,"Notation as a Tool of Thought" (http://www.jsoftware.com/papers/tot.htm http://www.jsoftware.com/papers/tot.htm). "I did pure mathematics in school, but later I was a teaching assistant for a graduate course in computer algorithms. I could see that the professor was getting killed by the notation. He was trying to express the idea of different kinds of matrix inner products, saying if you have a directed graph and you're looking at connections, then you write this triple nested loop in Fortran or Algol. It took him an hour to express it. What he really wanted to show was that for a connected graph it was an or-dot-and. If it's a graph of pipe capacities, then maybe it's a plus-dot-min. If he'd had APL or K as a notation, he could have covered that in a few seconds or maybe a minute, but because of the notation he couldn't do it." - "A Conversation with Arthur Whitney" (http://queue.acm.org/detail.cfm?id=1531242 http://queue.acm.org/detail.cfm?id=1531242)
        gchpaco 16y ago
        APL is actually not inherently so bad, because at least the operators have some vague relationship to what they're doing and are all distinct. I can't read J at all because in attempting to embed the APL operator space in ASCII they went in for multiple token operators that can be modified (so =, =., and =: are all different, and actually quite different IIRC). A better way of doing this is Fortress, which tries to use mathematical notation when possible but doesn't try to compress it as small as possible.
    - robinhouston 16y ago
      It looks as though Kevin Lawler can hold his own in that style, too. https://github.com/kevinlawler/kona/blob/master/k.c https://github.com/kevinlawler/kona/blob/master/k.c
    - Locke1689 16y ago
      That is one of the most frustratingly idiotic things I have ever seen. Any incentive I had to try this language is now gone and I want to erase the memory of its construction from my mind.
      - sofuture 16y ago
        How could you look at an APL-family language and then be totally outraged that the implementation took quite a bit stylistically from the language itself?
        billswift 16y ago
        Conventional C style is there to improve its readability and maintainability. Applying APL style to C code is sort of like trying to use Russian grammar with English vocabulary, you might be able to puzzle out the meaning, but it is a stupid waste of time.
        silentbicycle 16y ago
        Actually, given that a substantial part of the codebase involves looping simple operators over vectors, it makes sense to use macros to make the looping implicit. Code like this needs to be read more slowly than usual, because unlike much C code, the information level isn't low enough that you can read it lines at a time. Math papers have all these weird sigma symbols and stuff, too; you can't read just them as fast as you would read "for (i=0; i<N; i++) {" for the millionth time. Likewise, rather than writing "for (i=0; i<N; i++) {" yet again, K just calls that ' and moves on. In the source, there's a macro called DO, which is used extensively. DO(N, Blk) -> for(i=0, int _n=N; i<_n; i++) { Blk } . The APLs are about having a notation that doesn't constantly get in the way of thinking about your actual problem.
        Locke1689 16y ago
        In practical implementations we call this code write-only. It's funny that you bring up mathematical notation because mathematical notation is completely optimized for write-only. Mathematics is so obsessed with one letter variable names that they had to gather letters from other alphabets to continue their opaque tradition.
  - kragen 16y ago
    That's how it's really written. Take a look at the source of A+ sometime.
- gmartres 16y ago
  Thanks for that! I've been trying to use the libraries found at http://www.jsoftware.com/trac/base/browser/trunk http://www.jsoftware.com/trac/base/browser/trunk (svn co http://www.jsoftware.com/svn/base/trunk/ http://www.jsoftware.com/svn/base/trunk/), but I can't seem to be able to load anything: % jconsole |file name error | 0!:0 y load 'packages/misc/fndisplay.ijs' |value error: load | load'packages/misc/fndisplay.ijs' (I tried with absolute paths too) Any idea what I might be missing? And do you know the license of the libraries?
cousin_it 16y ago
Haha, I remember feeling the same way about K many years ago.
3pt14159 16y ago
Yesterday I was testing something in ruby. Basically the return value was a constant * rand. Yeah, 0.6360787621690106 and 0.6360787023690106 were the two results. I was like what's going on? until I noticed the middle of the answer was different. Felt like nearly winning the lottery.
- swah 16y ago
  I'd FUUUUUU on that.
losethos 16y ago
I wrote down BMP file palette reordering.
dfan 16y ago
I'm trying to figure out what is so startling about this, actually. The { function just seems to return the element at the given index, or if you pass in a collection of indices, maps that operation onto the collection. That is convenient operator overloading but not mind-blowing.
- sofuture 16y ago
  True, that is what's happening. The cool thing is that dimensional (in J terms, it's 'rank') overloading is inherent to the language. Most operators have implicit dimensional looping or respond different to differently ranked inputs. Take for example integers (i.): i.3 0 1 2 i.3 3 0 1 2 3 4 5 6 7 8 (That's kind of a dumb example, but I think it at least hints at the expressive power of rank.)
dcolgan 16y ago
I've been taken aback by some of the concepts of Haskell recently as well. Hooray for obscure but insanely powerful languages!
numeromancer 16y ago
OMG! Zoinks!
raymondh 16y ago
Other than using an operator for a function name, there's no real rocket surgery in this post: >>> def index(x, y): 'Dyadic function indexes y by x' return [y[i] for i in x] >>> p = list(range(22)) >>> shuffle(p) >>> p [1, 20, 19, 15, 9, 14, 6, 8, 2, 4, 16, 18, 10, 3, 0, 13, 12, 11, 7, 17, 5, 21] >>> index(p, p) [20, 5, 17, 13, 4, 0, 6, 2, 19, 9, 12, 7, 16, 15, 1, 3, 10, 18, 8, 11, 14, 21] >>> index([1]*22, p) [20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20] >>> index(p, [1]*22) [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]
rbanffy 16y ago
I remember APL from my college years. APL is the archetypal write-only programming language. Even with 20-to-1 comment-to-code ratio, it was easy to get confused with code written the previous morning. By you. From what I see, the most important development is using plain ASCII. Too bad it's still unpronounceable. Dictating APL code to another student was endless fun. We even invented our own names for some of the symbols... Having said that, APL was exquisitely powerful and concise. Yet, I have to wonder if there is space for an exquisitely concise write-only computer language today.
forkandwait 16y ago
I would like to coin a phrase: "seduced by syntax". I think humans LOVE syntax; and syntax is often loveable, crystallizing a new viewpoint into a linguist pattern that allows this viewpoint to be diffused by our amazing hardwired capabilities for language. J does that. That being said, sometimes being seduced, while REALLY FUN for a little while, turns out to REALLY MESSED UP in the long run. I think J and APL are like that too -- just break up with him, really.... For another metaphorical take: I think Python and C are like English -- boring, with a lot of fun syntax stripped out, but very, very useful. C++ is like Mandarin -- completely complicated, but so much literature and history is written in it that you have to take it seriously. Shell is like the pidgin languages. And J/APL ... they are like Quileute (the language spoken by Jacob-the-Werewolf's tribe): completely difficult, arcane, counter intuitive, and only spoken by like six other weirdos (ie don't bother). All these languages are seductive, though. J programmers love to exchange emails like "I found a way to calculate pi to the 80th decimal with only 4 characters" ... as if that is remotely useful. We have J code in use in our office -- its fun, if you like speaking Klingon, but it is a complete waste of time and sweat for any pragmatic applications. If you want to do complex weird matrix stuff in a few lines, learn linear algebra and tensor operations for chrissake, which actually have some utility in the real world. And don't be fooled by APL/J types telling you their language is mathematical -- it isn't, in the sense that you can't actually derive new results with it; it is just difficult (good math makes thing easier). Sorry for the rant. I wish the damn language would disappear, and we could remember it wistfully like SNOBOL, Ada, SAS, and other bad experiments in seductive syntax.
- gruseom 16y ago
  Syntax to me is superficial. What APL/J/K are doing is much deeper: programs are short sequences of composable operators.
  - ldh 16y ago
    I get your point, but syntax ceases to be superficial if it prevents you from expressing what you want to say cleanly. As a daytime Java programmer, discovering the freedom from syntax that Lisps allow you has been revelatory.
billrobertson42 16y ago
Intentional pun in the blog? "This shit is crazy! Let me show you a log."
- sbierwagen 16y ago
  :>

66 points by sbierwagen 16y ago

J is awesome. GPL'd J is even better! Some of us are trying to organize our efforts here: https://github.com/openj/core https://github.com/openj/core We don't intend on mucking with the code much at this point, just patching together a better build system, etc.

jameskilton 16y ago
I think I'm more amazed at the C code. Is that obfuscated or a decompiled dump? Or is that how the core of J is really written?
- sofuture 16y ago
  Hah! That is how the code is really written. It's... a unique style, shared amongst the APL languages. I'm pretty sure there are about two people that can read and write that fluently: Roger Hui and Arthur Whitney.
  - swah 16y ago
    Btw, is this related in anyway to Arthur?
    - kragen 16y ago
      Apparently Arthur Whitney wrote a one-page interpreter of an APL dialect in C (in the same style!) which inspired Ken Iverson to write J. http://keiapl.org/rhui/remember.htm#40 http://keiapl.org/rhui/remember.htm#40
      - swah 16y ago
        He is also the author of the K language.
  - jfager 16y ago
    As a wholly unaffiliated, self-selected, and entirely unofficial spokesperson, I'm happy to bestow this lifetime achievement award on behalf of the IOCCC.
    - narrator 16y ago
      People who like APL do not often code with readability as their number 1 priority. In fact, to code in APL one must have an outright disdain for the idea that someone else could easily understand their code.
      - gnosis 16y ago
        With that kind of attitude, they might enjoy programming in Brainfuck or Malbolge even more.
      - silentbicycle 16y ago
        No, we just have a different idea of what's readable. APL was created to have a mathematical notation that made problem solving easier. Don't believe me? See Iverson's Turing Award lecture,"Notation as a Tool of Thought" (http://www.jsoftware.com/papers/tot.htm http://www.jsoftware.com/papers/tot.htm). "I did pure mathematics in school, but later I was a teaching assistant for a graduate course in computer algorithms. I could see that the professor was getting killed by the notation. He was trying to express the idea of different kinds of matrix inner products, saying if you have a directed graph and you're looking at connections, then you write this triple nested loop in Fortran or Algol. It took him an hour to express it. What he really wanted to show was that for a connected graph it was an or-dot-and. If it's a graph of pipe capacities, then maybe it's a plus-dot-min. If he'd had APL or K as a notation, he could have covered that in a few seconds or maybe a minute, but because of the notation he couldn't do it." - "A Conversation with Arthur Whitney" (http://queue.acm.org/detail.cfm?id=1531242 http://queue.acm.org/detail.cfm?id=1531242)
      - gchpaco 16y ago
        APL is actually not inherently so bad, because at least the operators have some vague relationship to what they're doing and are all distinct. I can't read J at all because in attempting to embed the APL operator space in ASCII they went in for multiple token operators that can be modified (so =, =., and =: are all different, and actually quite different IIRC). A better way of doing this is Fortress, which tries to use mathematical notation when possible but doesn't try to compress it as small as possible.
  - robinhouston 16y ago
    It looks as though Kevin Lawler can hold his own in that style, too. https://github.com/kevinlawler/kona/blob/master/k.c https://github.com/kevinlawler/kona/blob/master/k.c
  - Locke1689 16y ago
    That is one of the most frustratingly idiotic things I have ever seen. Any incentive I had to try this language is now gone and I want to erase the memory of its construction from my mind.
    - sofuture 16y ago
      How could you look at an APL-family language and then be totally outraged that the implementation took quite a bit stylistically from the language itself?
      - billswift 16y ago
        Conventional C style is there to improve its readability and maintainability. Applying APL style to C code is sort of like trying to use Russian grammar with English vocabulary, you might be able to puzzle out the meaning, but it is a stupid waste of time.
        silentbicycle 16y ago
        Actually, given that a substantial part of the codebase involves looping simple operators over vectors, it makes sense to use macros to make the looping implicit. Code like this needs to be read more slowly than usual, because unlike much C code, the information level isn't low enough that you can read it lines at a time. Math papers have all these weird sigma symbols and stuff, too; you can't read just them as fast as you would read "for (i=0; i<N; i++) {" for the millionth time. Likewise, rather than writing "for (i=0; i<N; i++) {" yet again, K just calls that ' and moves on. In the source, there's a macro called DO, which is used extensively. DO(N, Blk) -> for(i=0, int _n=N; i<_n; i++) { Blk } . The APLs are about having a notation that doesn't constantly get in the way of thinking about your actual problem.
        Locke1689 16y ago
        In practical implementations we call this code write-only. It's funny that you bring up mathematical notation because mathematical notation is completely optimized for write-only. Mathematics is so obsessed with one letter variable names that they had to gather letters from other alphabets to continue their opaque tradition.
- kragen 16y ago
  That's how it's really written. Take a look at the source of A+ sometime.
gmartres 16y ago
Thanks for that! I've been trying to use the libraries found at http://www.jsoftware.com/trac/base/browser/trunk http://www.jsoftware.com/trac/base/browser/trunk (svn co http://www.jsoftware.com/svn/base/trunk/ http://www.jsoftware.com/svn/base/trunk/), but I can't seem to be able to load anything: % jconsole |file name error | 0!:0 y load 'packages/misc/fndisplay.ijs' |value error: load | load'packages/misc/fndisplay.ijs' (I tried with absolute paths too) Any idea what I might be missing? And do you know the license of the libraries?

cousin_it 16y ago

Haha, I remember feeling the same way about K many years ago.

3pt14159 16y ago

Yesterday I was testing something in ruby. Basically the return value was a constant * rand. Yeah, 0.6360787621690106 and 0.6360787023690106 were the two results. I was like what's going on? until I noticed the middle of the answer was different. Felt like nearly winning the lottery.

swah 16y ago
I'd FUUUUUU on that.

losethos 16y ago

I wrote down BMP file palette reordering.

dfan 16y ago

I'm trying to figure out what is so startling about this, actually. The { function just seems to return the element at the given index, or if you pass in a collection of indices, maps that operation onto the collection. That is convenient operator overloading but not mind-blowing.

sofuture 16y ago
True, that is what's happening. The cool thing is that dimensional (in J terms, it's 'rank') overloading is inherent to the language. Most operators have implicit dimensional looping or respond different to differently ranked inputs. Take for example integers (i.): i.3 0 1 2 i.3 3 0 1 2 3 4 5 6 7 8 (That's kind of a dumb example, but I think it at least hints at the expressive power of rank.)

dcolgan 16y ago

I've been taken aback by some of the concepts of Haskell recently as well. Hooray for obscure but insanely powerful languages!

numeromancer 16y ago

OMG! Zoinks!

raymondh 16y ago

Other than using an operator for a function name, there's no real rocket surgery in this post: >>> def index(x, y): 'Dyadic function indexes y by x' return [y[i] for i in x] >>> p = list(range(22)) >>> shuffle(p) >>> p [1, 20, 19, 15, 9, 14, 6, 8, 2, 4, 16, 18, 10, 3, 0, 13, 12, 11, 7, 17, 5, 21] >>> index(p, p) [20, 5, 17, 13, 4, 0, 6, 2, 19, 9, 12, 7, 16, 15, 1, 3, 10, 18, 8, 11, 14, 21] >>> index([1]*22, p) [20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20, 20] >>> index(p, [1]*22) [1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1]

rbanffy 16y ago

I remember APL from my college years. APL is the archetypal write-only programming language. Even with 20-to-1 comment-to-code ratio, it was easy to get confused with code written the previous morning. By you. From what I see, the most important development is using plain ASCII. Too bad it's still unpronounceable. Dictating APL code to another student was endless fun. We even invented our own names for some of the symbols... Having said that, APL was exquisitely powerful and concise. Yet, I have to wonder if there is space for an exquisitely concise write-only computer language today.

forkandwait 16y ago

I would like to coin a phrase: "seduced by syntax". I think humans LOVE syntax; and syntax is often loveable, crystallizing a new viewpoint into a linguist pattern that allows this viewpoint to be diffused by our amazing hardwired capabilities for language. J does that. That being said, sometimes being seduced, while REALLY FUN for a little while, turns out to REALLY MESSED UP in the long run. I think J and APL are like that too -- just break up with him, really.... For another metaphorical take: I think Python and C are like English -- boring, with a lot of fun syntax stripped out, but very, very useful. C++ is like Mandarin -- completely complicated, but so much literature and history is written in it that you have to take it seriously. Shell is like the pidgin languages. And J/APL ... they are like Quileute (the language spoken by Jacob-the-Werewolf's tribe): completely difficult, arcane, counter intuitive, and only spoken by like six other weirdos (ie don't bother). All these languages are seductive, though. J programmers love to exchange emails like "I found a way to calculate pi to the 80th decimal with only 4 characters" ... as if that is remotely useful. We have J code in use in our office -- its fun, if you like speaking Klingon, but it is a complete waste of time and sweat for any pragmatic applications. If you want to do complex weird matrix stuff in a few lines, learn linear algebra and tensor operations for chrissake, which actually have some utility in the real world. And don't be fooled by APL/J types telling you their language is mathematical -- it isn't, in the sense that you can't actually derive new results with it; it is just difficult (good math makes thing easier). Sorry for the rant. I wish the damn language would disappear, and we could remember it wistfully like SNOBOL, Ada, SAS, and other bad experiments in seductive syntax.

gruseom 16y ago
Syntax to me is superficial. What APL/J/K are doing is much deeper: programs are short sequences of composable operators.
- ldh 16y ago
  I get your point, but syntax ceases to be superficial if it prevents you from expressing what you want to say cleanly. As a daytime Java programmer, discovering the freedom from syntax that Lisps allow you has been revelatory.

billrobertson42 16y ago

Intentional pun in the blog? "This shit is crazy! Let me show you a log."

sbierwagen 16y ago
:>