Hy – A Lisp-flavored Python

https://hy.readthedocs.io/en/stable/

by tosh • 9 years ago

625 213 9 years ago

hy.readthedocs.io

asimjalis4• 9 years ago

I have been using Hy as an alternative to Python. It is delightful to use. Highly recommended.

cheez• 9 years ago

What kind of projects have you used it for?

rcarmo2• 9 years ago

My blog/wiki engine is written in an older version of Hy:

https://github.com/rcarmo/sushy

I'm currently re-writing it in Python 3.6 because Hy has taken a hard stance in backwards compatibility and deprecated "let" (with a replacement macro, but still) and isn't tackling async (both for understandable, but sad reasons), so it doesn't really work for me anymore.

But while it did, I loved it to bits. Python with a LISP syntax is just wonderful, and if someone ever finds a way to add back the stuff I like, I'm more than willing to deal with the quirks.

asimjalis4• 9 years ago

My main use cases: Doing data analysis using Numpy, Pandas, sklearn. Calling AWS API using Boto. Calling GDAX API using their Python code samples.

I also use it for small personal scripts.

My first choice for one-off scripts would have been Clojure. But the long startup time for Clojure apps makes it unusable for quick command line scripts. Hy gives us the best of both worlds.

elf_m_sternberg• 9 years ago

There are a few projects on my github (elfsternberg) that I've used Hy for.

One thing that I've also done is write the first draft in Hy, and then used Hy2Py to generate the version that I'm going to publish. This does result in a lot of compiling-by-hand, but the end product is often surprisingly robust.

On the other hand, it's also very (!) un-pythonic. Hy discourages classes in favor of closures, like Scheme, so I end up with a lot of nested sub-functions. My project git-linter looks like that. Hy encourages highly functional thinking, and git-linter is just that: the inputs are things like the configuration file, available linters, the command line, and (depending on the command) either the output of a "git-status --porcelain" or just "find . -type f"; the output is a report on the various outputs of the linters. It's a very straightforward map/reduce/filter, so no object-orientated code at all was required.

To an engrained Pythonista who must make a class for every last step, this land-of-verbs approach tends to look strange.

JadeNB3• 9 years ago

using Hy …. Highly recommended.

I think that you are not getting in the proper punning spirit of the language.

taneem• 9 years ago

The online demo is amusing/creative: https://try-hy.appspot.com/

nathancahill• 9 years ago

This is my go-to for toy projects over the last couple years. Great execution all around. Be aware, if you don't like puns, this isn't the language for you.

sn9• 9 years ago

So you're saying that if you don't like puns, you should get off your Hy horse?

kerkeslager2• 9 years ago

Nothing so aggressive. He's just saying they hold puns in Hy esteem.

alschwalm2• 9 years ago

Hy is a great project. One unfortunate thing, though, is that no one has been able to successfully implement `let`, which makes writing idiomatic lisp almost impossible. I'm curious if anyone can actually prove it can't be done or is just very difficult (or too slow to be useful).

rcarmo2• 9 years ago

This. See my other comment for a sizable Hy example and more info, but the lack of "let" mostly killed it for me.

nnq4• 9 years ago

but the lack of "let" mostly killed it for me

Can you give an example of code that is improved by let significantly? ...and can't be rewritten in a more readable way without it?

nerdponx4• 9 years ago

What is it about `let` that makes it so hard to implement?

alschwalm2• 9 years ago

There is a description of the problem here: https://github.com/hylang/hy/issues/1056

I think it really comes down to python having limited scoping capabilities (I think you can only really make new scopes with function declarations), this can maybe be hacked around, but at too high a performance cost to be usable.

thaumasiotes• 9 years ago

I'm curious too. Conceptually, `let` defines a function of n variables and then invokes it on a list of n values:

    (let ((a 3)
          (b 5)
          (c -1))
      (* a (+ b c)))

    def «gensym»(a,b,c):
        return (a * (b + c))
    «the-gensym»(3,5,-1)

It seems workable purely in terms of the code transformation. But, I haven't tried.

tuturto• 9 years ago

I actually wrote a short blog post about this just recently: https://engineersjourney.wordpress.com/2017/08/01/history-of...

In short, let kept running into trouble with yield, exceptions, breaking out of loops and such. It would have been really tricky (probably close to impossible) to get everything working correctly.

shuzchen• 9 years ago

Hy actually used to have a let (but a broken one) and it was removed in a recent version. For context on this see https://github.com/hylang/hy/issues/1056

aidenn02• 9 years ago

Is there something about Hy's lambda that makes implementing let with it not possible?

For those unaware, the transformation is roughly:

  (let ((foo bar) ...) Y) -> ((lambda (foo ...) Y) bar ...)

[edit]

Reading around it seems that it breaks several pythonic control-transfer functions that treat functions specially (e.g. yield from inside a let would be surprising).

Without knowing much about Hy's semantics, it should still be doable by rewriting variable names. That's likely not doable in a robust manner via a macro though, so would have to be part of Hy's transformation from sexpr to ASTs.

kzisme• 9 years ago

Shout out to @paultag (Author/Creator of Hy) - he's an awesome guy who has written some really cool stuff.

boxy_brown• 9 years ago

This, Paul is someone I look to as a model open source contributor in so many ways.

https://twitter.com/nedbat/status/861696880790704128

How many people can say they have worked with [Sally] Yates, and have written exemplary Dockerfiles?

paultag3• 9 years ago

Thank you so much. I'm deeply humbled and extremely grateful for this. It's the nicest thing I've read all week, and really brightened my day up. Thank you!

paultag3• 9 years ago

yoodenvranx• 9 years ago

Hy

This must be the worst submission title in the history of HN.

asimjalis4• 9 years ago

It is possibly the shortest.

hk__23• 9 years ago

I got curious and I checked; there have been a few one-character submissions so it’s not the shortest.

     ⊥: https://news.ycombinator.com/item?id=7530548
     {: https://news.ycombinator.com/item?id=7729399
     M: https://news.ycombinator.com/item?id=4367006

In total (at least in the [fh-bigquery:hackernews.stories] BQ dataset), there have been 15 non-deleted submissions with a one-character title.

winter_blue• 9 years ago

Since Python doesn't support multi-line lambdas, how do they support the analog in LISP? It's almost necessary for any dialect of LISP to support lambdas (i.e. functions) that contain a LISP 'do' [1] which lets you group statements. Do they chain together expressions/statements with continuations or something?

[1] http://www.lispworks.com/documentation/lw60/CLHS/Body/m_do_d...

gkya3• 9 years ago

IIRC in a talk the author said that it just generated a function with a made-up unique name, like an internal def in another function.

serhei• 9 years ago

That's pretty similar to how languages like Scala with higher-order features translate down to the JVM bytecode (which is subject to certain restrictions).

joejev• 9 years ago

Python lambdas are just syntactic sugar for creating a function object. The ast can just build a named function and use it in the needed expression.

kazinator11• 9 years ago

However, that approach seems rather backwards compared to having an anonymous function primitive which is used as the basis for the lambda one-line-only sugar, as well as named functions (which simply establish a binding between a name and the anonymous thing).

kbp4• 9 years ago

It's almost necessary for any dialect of LISP to support lambdas (i.e. functions) that contain a LISP 'do' [1] which lets you group statements.
[1] http://www.lispworks.com/documentation/lw60/CLHS/Body/m_do_d...

'do' in most Lisps and Schemes is essentially an extended for loop, not a blocking construct. Of course, you could write

    (do ((once nil t)) ; Scheme: (do ((once #f #t))
        (once)
      (thing-1)
      (thing-2)
      ...)

or something, but that's silly.

The blocking construct is in CL called PROGN and in Scheme called begin. Lambdas in CL get what's called an "implicit PROGN" around their bodies, so you can put however many forms you want inside and the value of the last is returned. It works the same in modern Scheme lambdas.

Calling PROGN/begin "do" is solely a Clojureism as far as I know.

kazinator11• 9 years ago

Calling PROGN/begin "do" is solely a Clojureism as far as I know.

Also an ANSI-CL-LOOP-ism:

   (loop for x in '(1 2 3) do (print x) (print x))
   1
   1
   2
   2
   3
   3

Omit the do and you have a syntax error.

tills13• 9 years ago

hy more like why

etiam2• 9 years ago

There's even a couple of kernels for IPython/Jupyter. The best I've seen so far is https://github.com/Calysto/calysto_hy

I'd like to move to Hy for much more of what I currently do in Python, but so far I've been too lazy to find/create good editor support for it. For actual projects of some size the loss of context help, documentation, various completions, etc seems like too high a price to pay.

zitterbewegung• 9 years ago

I think the highest cost is really collaboration with others. If you ever want help with your project or if people even want to use it you are dealing with only Hy users and disregarding regular Python users. I do like the concept of Hy though.

etiam2• 9 years ago

It's restrictive that way, no doubt.

On the other hand, I suspect that as subsets go, Hy programmers would be a good one.

And I'm not sure the impact on getting people to use it would necessarily be that bad. It's just an extra import hy away to use modules written in Hy just as usual from Python, after all.

thro1237• 9 years ago

Also, in terms of useful debug/error messages. That was the biggest drawback of clojure.

partingshots• 9 years ago

Now CS61A can finally go back to teaching purely in Scheme again.

kronos29296• 9 years ago

The entire thing is full of puns really too much of puns related to hy. Cool project. Python joined Erlang and Java with a lisp flavour of its own with Hy. Hope we get a let expression soon.

smaili• 9 years ago

Excerpt:

Hy is a wonderful dialect of Lisp that's embedded in Python.
Since Hy transforms its Lisp code into the Python Abstract Syntax Tree, you have the whole beautiful world of Python at your fingertips, in Lisp form!

dvdt• 9 years ago

One fun application: I used Hy to control my liquid handling robot,

https://twitter.com/dtsao/status/891038101706989568

cuspycode• 9 years ago

I used Hylang to program an Arduino Yún a few years ago, via cross-compilation. This was a very nice experience, since it allowed me to use the great Python ecosystem without having to be limited by Python the language.

udkl• 9 years ago

While both have their own personalities, has anyone done a side by side syntax and feature comparison of clojure and Hy ?

ealhad2• 9 years ago

“Keep in mind we’re not Clojure. We’re not Common Lisp. We’re Homoiconic Python, with extra bits that make sense.” — Hy Style Guide

That said, Hy takes a lot from Clojure.

flavio817• 9 years ago

That said, Hy takes a lot from Clojure.

Oh oh... That scares me. Clojure -as a Lisp-like language running on the JVM- is a good idea but i have some reservations regarding how that idea got implemented.

didibus4• 9 years ago

Features are very different, because the features of Hy are the features of Python. So any comparison of Clojure vs Python will do.

The syntax though is very similar.

IgorPartola4• 9 years ago

As a heavy user of Python, and someone who grew up with the "curly braces" languages, I have a question for y'all. Is this really readable?

    (setv result (- (/ (+ 1 3 88) 2) 8))

Or rather is it more readable than

    result = ((1 + 3 + 88) / 2) - 8

I just... Do you just get used to this, or is it something that you have to keep struggling with? Especially given that the latter is how we do math the rest of the time?

evilduck• 9 years ago

It's just training. Buy and use a RPN calculator and lisp suddenly seems normal.

(To answer your question directly, yes it's readable. You learn to look past the syntax in lisp the same as you stop reading semi-colons and curly braces in other languages)

gkya3• 9 years ago

M-x calc!

xelxebar• 9 years ago

Or if you're a fan of cli tools, check out dc!

chx• 9 years ago

Normal? It's upside down! I am a die hard RPN fan, started with a HP 32S when I was 14 years old almost three decades ago, upgraded to a HP 48G a few years old (well, kind of a 48G - 48GX hybrid, a friend of mine and I hacked it to 512K SRAM, gosh, hand soldering those little TSOP legs, brrr, we just stacked them chips and ran the data legs together, those were the days).

But this is alien and hard to read, RPN should be _reverse_ polish notation, you know? And + 1 2 3 is weird because it makes + a three argument operation, or I guess in this case, variable number of arguments. 1 2 3 + + is easy to read and understand :)

jwr• 9 years ago

From a long-time Lisper (Common Lisp, and now Clojure): not, this is not comfortably readable.

That said, I very rarely write code like this. I tend to split operations over multiple lines, so that you don't spend so much time digging into and out of nesting levels. Prefix notation is not ideal for reading math. People have been writing infix math notation libraries for CL for exactly this reason.

It's not a showstopper, though. Also, I do find that sometimes having + and < to be of arbitrary arity is really convenient and makes for nicer code.

kazinator11• 9 years ago

People have been writing infix math notation libraries for CL for exactly this reason.

People have writing these notation libraries, but then not actually using them.

Infix libraries for CL mostly exist to end arguments; "we have infix; Lispers choose not to use it, and you will understand why soon enough".

mthomas2• 9 years ago

I find that both are equally (un)readable. However, for expressions that are that long, I'm inclined to write:

    (setv result (-> (+ 1 3 88) 
                     (/ 2) 
                     (- 8)))

jamescostian• 9 years ago

I am not a lisper, but that looks pretty cool. The non-lisp ways of writing that would either require adding parens like GP or writing things like "result /= 2", but that means repeating "result" or whatever variable many times.

cle• 9 years ago

This is even better than in-fix notation.

lliamander2• 9 years ago

Indeed, I like that better than either typical prefix or infix nested expressions.

What do you advise in the case the expression tree is a little bit more complex?

For example: (setv result (- (/ (+ 1 3 88) 2) (* 8 (+ 3 2)))

lliamander2• 9 years ago

Also, I've seen lisps that obviate the need for the outermost parentheses (I think by using significant whitespace). In that case, the above example would have as many parens as the python example (for those who are keeping count).

btschaegg• 9 years ago

Still being an absolute beginner at lispy syntax, I'll have to ask you the same question though.

While I see the awesomeness in threading macros, I'm usually left pondering a while on where the result actually goes in the next s-expression. How long did it take you to get used to that? Would you prefer someting like `as->` for readability?

swiley• 9 years ago

I'd imagine it's like FORTH: If you read enough of it (it only takes a few days of reading) it becomes clear.

throwaway135634• 9 years ago

In my experience, the hardest part about reading Forth is recognizing when a word doesn't use RPN because of metaprogramming

ifelsehow• 9 years ago

As @jwr says, there are different paradigms in Lisp, such that you would never really write code like this.

6581• 9 years ago

I have been using RPN calculators for all my life (except when I was forced to use infix calculators in school). Your first example is at least as readable to me as the second one. Just read it from inside to outside instead of from left to right.

dunham• 9 years ago

In school, RPN occasionally caused trouble for me as I could work through an entire problem (leaving values on the stack) without writing down any intermediate steps.

deathanatos• 9 years ago

This is the main reason Lisp has just never grown on me. I'm certain — and other responses demonstrate — that the readability can certainly be improved here. Yes, a programming language needs nothing more than parens. But dedicating a tastefully chosen set of sigils seems to make a lot of difference, in my opinion.

kbp4• 9 years ago

I understand this point of view if most of your programs are just doing arithmetic, but once you start using functions and procedures the differences are pretty minimal.

kazinator11• 9 years ago

In the dialect called TXR Lisp, I have introduced precisely that: some tastefully chosen syntactic sugars which do not disturb the structure of the surrounding Lisp. I believe I made a wise choice by not targetting arithmetic. My syntactic inventions target areas like working with arrays, string processing, OOP and higher order functions.

Small taste:

  1> (let ((x "abc") (y "def"))
       (swap [x 0..2] [y 2..:])
       (list x y))
  ("fc" "deab")

  2> (mapcar (ret `@1-@2`) "abc" "def")
  ("a-d" "b-e" "c-f")

  3> '#"lorem ipsum dolor sit"
  ("lorem" "ipsum" "dolor" "sit")

  4> (let ((a "dolor")) ^#`lorem ipsum @,a sit`)
  #`lorem ipsum dolor sit`

  5> (eval *4)
  ("lorem" "ipsum" "dolor" "sit")

  7> (defstruct (point x y) nil (x 0) (y 0))
  #<struct-type point>

  8> (new point)
  #S(point x 0 y 0)

  9> (new (point 1 2))
  #S(point x 1 y 2)

  10> (new point x 2 y 3)
  #S(point x 2 y 3)

  11> (let ((a (new (point 10 20))))
        ^(,a.x ,a.y))
  (10 20)

  12> (let ((a (new (point 10 20))))
        `<@{a.x}, @{a.y}>`)
  "<10, 20>"

  13> (defmeth point distance (p) (sqrt (+ (* p.x p.x) (* p.y p.y))))
  (meth point distance)

  14> (new (point 3 4)).(distance)
  5.0

  15> (defmeth point distance-to (p q)
        (let ((r (new (point (- q.x p.x)
                             (- q.y p.y)))))
          r.(distance))) 
  (meth point distance-to)

  16> (new (point 3 4)).(distance-to (new (point 20 30)))
  31.0644491340181

  19> (let ((pts (build (dotimes (i 10) (add (new (point (rand 10)
                                                         (rand 10))))))))
        pts)
  (#S(point x 5 y 6) #S(point x 1 y 1) #S(point x 1 y 3) #S(point x 5 y 7)
   #S(point x 2 y 8) #S(point x 7 y 3) #S(point x 1 y 1) #S(point x 2 y 4)
   #S(point x 5 y 6) #S(point x 5 y 3))

  20> (mapcar .(distance) *19)
  (7.81024967590665 1.4142135623731 3.16227766016838 8.60232526704263
   8.24621125123532 7.61577310586391 1.4142135623731 4.47213595499958
   7.81024967590665 5.8309518948453)

  21> (mapcar (ret @1.(distance-to @2)) *19 (cdr *19))
  (6.40312423743285 2.0 5.65685424949238 3.16227766016838 7.07106781186548
   6.32455532033676 3.16227766016838 3.60555127546399 3.0)

cutler2• 9 years ago

It's not merely a matter of syntax or taste. Lisp's code-as-data and advanced macros are made possible by its s-expressions.

jonahx2• 9 years ago

Do you just get used to this

Short answer: yes, and in far less time than you'd think. The difference between the examples you cite, specifically, is trivial, and if it's a sticking point for you I'd suggest (and I'm not trying to be catty at all) that you broaden your experience with different languages. I mean, languages like J can become readable with enough experience -- stuff like the above is just a blip on the syntactical radar screen.

Optimizing syntax for readability is a worthy goal, but experience (both with the language in question and languages in general) is an almost insurmountable confound here, and ultimately your goal should be to effortlessly "see past" surface differences so you can concentrate on the deeper facilities the language provides for expression.

IgorPartola4• 9 years ago

Heh. I am pulling this example straight from Hy's tutorial. They are the ones claiming that this syntax is superior to Python's. I do have a hard time with parsing it in my brain.

Long ago, I internalized what I believe to be true about programming languages: they should be optimized for the humans using them, not for the machines that run the programs or the compiler authors who create them. This leads to less bugs. Maybe your goal is to see past syntax to understand language semantics. My goal is to get code shipped with as few bugs as possible as quickly as possible. Python and ES let me do that pretty well. Also don't meant to be catty, to each their own interests.

dustingetz• 9 years ago

In clojure you write it like this, so the parens indicate scope. No need to mutate an environment here. dunno about hy

    (let [result (- (/ (+ 1 3 88) 2) 8)]
      ; result defined inside the expression only
      )

cannonpr• 9 years ago

I used to be mostly a Python guy, fell in love with Clojures syntax, you get used to it surprisingly fast, and it's caused me less headaches than 'interesting' pythonic syntax. That having been said, I still love Python too.

erikpukinskis• 9 years ago

I would never use a variable name like result, it's meaningless. And the inner values should be labeled too, with their own variable names. Any compiler worth its salt will optimize the names away.

kazinator11• 9 years ago

There are advantages:

Long Lisp expressions break into multiple indented lines according to a couple of very easy rules:

   (- (/ (+ 1 3     ;; sideways tree [-] --- [/] --- [+] -- 1 3
            8 8)    ;;                   |       |       `- 8 8
         2)         ;;                   |        ` 2
      8)            ;;                    `- 8

by contrast, there is any satisfactory, canonical way to break up infix into multiple lines (with excellent support from editors like Emacs and Vim). Thus, Lisp expressions can be longer while remaining understandable. (Which is important because entire function definitions, for instance, are really one big expression.)

There is no ambiguity in Lisp expressions. The parentheses are not optional grouping punctuators, but operators denoting operator application (analogous to the parentheses in f(x, y)). You never, ever have to grapple with "what is the precedence/associativity of this?" or "I know the precedence/associativity of this, but did the person who wrote this intend it that way?"

Sufficiently long expressions are hard to read in any notation. Just like in any language, Lisp programs break down complex calculations with intermediate variables:

   (let ((bytes (+ 1 3 8 8)) ;; what is this term? Oh, bytes
         (words (/ bytes 2)) ;; why divide it by two? Ah 16 bit words
         (size (- words 8))) ;; Why subtract? Account for some header or special offset.
     ...)

> Do you just get used to this

What happens is that you have to get used to working with big, nested prefix expressions: because that's what functions are. A 200 line function in Lisp is a 200 line nested expression. You find that this it is perfectly manageable and you even like it. You get used to structuring code that way, and then that all applies to arithmetic expressions also, which are just more of the same thing. Given that you're able to cope with big functions made up of that sort of syntax, you are then not fazed by some five node arithmetic subexpression.

breck16• 9 years ago

I'm a Lisper (and Pythoner) and strongly agree, the first one is not readable. I would say it's gross. It boggles my mind that Lispers haven't taken that problem seriously. I have taken it very seriously and have been working on it for years.

I just started publishing my results.

http://breckyunits.com/the-flaw-in-lisp.html

What I've discovered is that the parentheses are completely unnecessary in Lisp, and in fact cause much harm by making Lisp harder to read and maintain in the long run. Drop the parentheses in Lisp by switching to whitespace/geometry for syntactic structure and you suddenly have the dream language.

Your example above would be something like:

  setv result
   -
    /
     + 1 3 88
     2
    8

You might think that looks a little strange too at first, but the neat thing about that code is you can write it on a piece of graph paper, draw lines around each word, connect the indented nodes, and you'd have a tree visualization of your syntax code without doing a single transformation of the source.

anaphor• 9 years ago

The problem is that it requires all of your functions to have a fixed arity I think, or at least a lower bound on the arity, or you use newlines as part of the syntax. Check out this https://shriram.github.io/p4p/

kod4• 9 years ago

Yeah, whitespace sensitive languages are a great idea...

Until you permanently lose user data because of a pull request that contained whitespace changes.

Yes, I've seen this happen in production, yes it could have been prevented in other ways... but why invite disasters like that upon yourself?

dragonwriter2• 9 years ago

Drop the parentheses in Lisp by switching to whitespace/geometry for syntactic structure and you suddenly have the dream language.

This has been “discovered” many times, but somehow this dream language never approaches even Lisp in popularity. (Though a loosely similar transformation of it's basic ML-based syntax is used in Haskell to reduce parens, but that's not a Lisp base.)

blain_the_train2• 9 years ago

Take a look an Parinfer https://shaunlebron.github.io/parinfer/. Which essential makes clojure a whitespace language with indent mode.

gnaritas4• 9 years ago

by switching to whitespace/geometry for syntactic structure and you suddenly have the dream language.

Nope, you've just made it awful, white-space with syntactic meaning is just god-awful. It's the worse thing about Python, there's a reason it's not a more popular language feature.

amelius2• 9 years ago

I'm using this same approach in a side project. The idea is very old. See e.g. [1] and the references therein.

[1] https://srfi.schemers.org/srfi-119/srfi-119.html

flavio817• 9 years ago

I'm a Lisper (and Pythoner) (...) It boggles my mind that Lispers haven't taken that problem seriously. I have taken it very seriously and have been working on it for years. I just started publishing my results. http://breckyunits.com/the-flaw-in-lisp.html

Honestly, there is no flaw in the parentheses. You just need to write more Lisp.

The parentheses are one of the best things in Lisp.

zeveb• 9 years ago

I'd say that they're about equally (un)readable. What do you think of:

    (setf result (make-instance 'foo :bar (baz quux)))

vice:

    result = new Foo(bar=baz(quux))

The first is simple to write a parser for, but the second is … evil.

dragonwriter2• 9 years ago

Prefix notation is something you get used to; it's not as easy as infix for operations that use the same infix operators with the same meaning and priority in the domain (e.g., math) but real programming languages often use operators that aren't standard math operators and have precedence and/or associativity quirks with the operators they use which increase overhead. Prefix notation is basically everything-is-a-function, which I think is ultimately less overhead overall, though still more for simple math, than infix.

Indentation can help, too.

jrs95• 9 years ago

I find it difficult to read if I just try and read it left to right, but if I simply go directly to the innermost operation its not really much more difficult for me. And I’m not a lisp user, either.

brudgers2• 9 years ago

Neither is readable because both are filled with magic numbers. With experience the Lisp becomes more readable when formatted conventionally:

  (setv result (- (/ (+ 1 3 88)
                      2)
                  8))

Line breaking the Python is possible but unconventional and requires additional symbols.

As expressions become more complex, languages with infix operators are faced with the choice between some form of operator precedence and strict left to right (or right to left) application. It is hard to look at:

  6 * 89 & 57 >> 2**2 - 1

and see the implicit parenthesis. And maybe not having to figure out where the implicit parentheses go in complex cases is part of the attraction of Lisp syntax. Lisp code does not require going to the rule book as often.

mythrwy• 9 years ago

Once you are in the parentheses line breaks in python are just fine, no additional symbols needed.

  result = ((( 1 +
 
          3 +
         
          8) /
         
          2)
 
          * 6)

works.

Also works as a way to deal with long strings without the \.

  long_string = ('hello'
      'really'
      ' really'
      ' long'
      ' line'
      )

cgag• 9 years ago

I find it pretty readable.

gkya3• 9 years ago

Maths in lisp is like learning a language with a different alphabet, it's confusing at first, then becomes natural. I myself prefer lisp's syntax especially for maths and parenthesise heavily in other languages because in you can never know how tightly they follow mathematical association rules, and that can be a source for really annoying bugs. Generally the total absence of syntactical ambiguity in lisp is really relieving (tho lots of complexity is shifted into macro expansion, which is less intriguing for me as macroes allow easy inspection whereas core language constructs not so much).

benji-york• 9 years ago

you can never know how tightly they follow mathematical association rules

You might be interested in the way Pony (https://www.ponylang.org/) handles mathematical order of operation rules: it doesn't. Runs of like-operator are allowed, but any complex expression has to have all the operations grouped by parentheses so as to remove any ambiguity.

such_a_casual4• 9 years ago

Prefix notation is a problem because nothing beats the familiarity of:

    x = 10

But prefix notation is superior. You don't have to invent an order of operations out of thin air. All your functions take several arguments, not just 2. So + is now sum, - is difference, / quotient, > greater than order, etc. They aren't special. they're just like everything else.

It'll never be as familiar, but it's smarter and makes writing code easier. For example

    "hello " + name + ", welcome to the " + place +"."
    + "hello " name ", welcome to the " place "."

The reality is you want functions that work on many things, not just 2, even when you're doing math.

    1 + height + x + floor-height
    + 1 height x floor-height

infix notation does nothing to make code clearer

kazinator11• 9 years ago

Anyway, you want string interpolation for the former example anyway:

  1> (let ((name "Alice") (place "The Palace"))
       `Hello @name, welcome to @place!`)
  "Hello Alice, welcome to The Palace!"

string interpolation is compatible with Lisp designs; it is self-contained and so doesn't "disturb" the surrounding syntax with issues of precedence and associativity. The above example is from TXR Lisp which has this built in. For Common Lisp, there is the cl-interpol and other packages, and other dialects have their own solutions.

jlarocco2• 9 years ago

It takes some getting used to, but eventually the Lisp style is just as readable as the non-Lisp style.

For actual coding (in an editor vs reading in a browser) it's even easier with things like matching parenthesis highlight, jump to matching brace, and rainbow parenthesis mode.

ohyes2• 9 years ago

I was never any good at math in the first place, order of operations always screwed me up, still does. Order of operations actually makes no sense at all (if you really think about it). So yeah, lisp is easier for me.

ohyes2• 9 years ago

Compare to:

1 + 3 + 88 / 2 - 2

Or (1 + 3 + 88) / 2 - 2

Or (1 + 3 + 88) / (2 - 2)

If it comes down to it, I'm going to fully parenthesize the stupid thing anything else is difficult to parse. If I'm using all the parens, prefix is fine.

IgorPartola4• 9 years ago

Since I can't edit, I'll reply to my own comment. This is not my example, but one I pulled straight from Hy's tutorial at https://hy.readthedocs.io/en/stable/tutorial.html#basic-intr.... I don't think I could have dreamt up code like this.

didibus4• 9 years ago

I got used to it, now I almost have the reverse problem.

Having said that, sometime for math, infix operators for complex equations are easier, and its easy to write a macro in Lisp to let you do that too:

(setv result (infix ((1 + 3 + 88) / 2) - 8))

endgame• 9 years ago

I've only written a small amount of elisp but I've found that while you don't actually do all that much arithmetic manipulation, the ease of manipulating S-Expressions outweighs the price of entry.

asimjalis4• 9 years ago

Here is how I would write that expression in Hy.

    (-> 1 (+ 3) (+ 88) (/ 2) (- 8))

Here is how I would write the assignment, using my definition of def2.

    (defmacro def2 [value variable] `(def ~variable ~value))
    (-> 1 (+ 3) (+ 88) (/ 2) (- 8) (def2 result))

dmoney2• 9 years ago

As a Python-lover who previously played with Common Lisp and Clojure, lispy syntax took less getting used to than having no end tags, but I'm still not used to prefix math.

torrent-of-ions• 9 years ago

Yes, it is. I can't say anything about being "more readable" but English is presumably "more readable" than isiXhosa for you. Reading is a skill that has to be learnt and practised. You were not born with the ability to read English or infix notation, you just got used to them.

klibertp3• 9 years ago

I have a blog post about this: http://klibert.eu/posts/tools_for_lisp_syntax.html (with pictures: http://klibert.eu/statics/images/lisp-editor-features1.png).

Basically, in my opinion, what makes a difference in readability is the tooling a certain kind of syntax enables. As you can see on my "Languages" page I have been trying out new languages for quite a while now, and so I've seen all levels of syntactic helpers. The syntax itself is actually a secondary concern: it's really not that important if your editor can easily generate it, manipulate it, and display while hiding parts of it for you.

However, how easy it is to write such tooling for a language depends on the kind of syntax it uses. Languages with more complex grammars are generally harder to support properly in the editor. Of course, how good the tools are, depends also on a level of effort invested in making them so - so relatively complex, but popular, languages may still have better editor support.

Anyway, back on topic of Lisp: yes, you get used to the prefix notation to the point that - after a long while, mind you - you're able to open a Lisp file in Notepad and still easily read it. No one does it, though, unless forced for some reason - things like:

  - blink the matching paren, display parens as rainbow colored
  - jump to matching paren
  - jump to inside next expression
  - reliably select current block of code
  - fix indentation automatically when copy&pasting code
  - split ( (..|..) -> (..)|(..) ), join, transpose, extend the current expression
  - convolute, which means making the nested (inner) expression the outside one

all of this makes Lisp readable, and probably more readable than some other languages with much more complex grammars. Lisp is not the only language with such a "feature" - recently I found Idris to be very good in this regard, both Erlang and (even more so) Elixir are similar, some (popular) brace languages too.

Then there's a thing about macros and trivial AST manipulation in Lisp, arguably enabled by the syntax being as it is. You can have pattern-based macros in any language, actually (see Nim, or Dylan), but the raw AST being identical to the code makes certain kinds of imperative macros much easier to write. Again, take a look at Nim: is has imperative macro facility, but for every piece of syntax you want to introduce you need to call a special constructor.

a-saleh2• 9 years ago

Well, when using clojure and needed to deal with lot of math I usually used some sort of infix macro, i.e. [1].

Otherwise, I wasn't really bothered by the brackets. And most of the time I would be using some sort of threading macro anyway [2], such as -> mthomas suggested in one of the responses.

[1] https://github.com/rm-hull/infix [2] https://clojure.org/guides/threading_macros

pvdebbe• 9 years ago

The thing is that usually basic algebra like this occurs somewhat seldomly in code. The rest of the function calls are readable, this is the (only) outlier to readability.

flavio817• 9 years ago

Do you just get used to this, or is it something that you have to keep struggling with?

You get easily used to parenthesis and you start loving them, since the auto-indenting (on any good Lisp editor) can make you have very readable code. In fact, Common Lisp, having so much versatility in control flow operators (and, in general, a ton of handy helper functions), lets you write really readable code, and often that is the case. So highly readable, that the downside is that the common Common lisper (pun intended) feels no need to add comments in the code...

As for your formula

(setv result (- (/ (+ 1 3 88) 2) 8))

It can be written in a more readable way. As user "brudgers" wrote above:

   (- (/ (+ 1 3 88)    
         2)         
      8)

... and it gets more evident.

As for "result", usually in Lisp you don't need to have such a variable since the last expression returned by a function automatically becomes the return value of the function.

platform• 9 years ago

prefix notation mimics of how we call functions

   setv result  (my_function  arg1, arg2)

to me this is as readeable as

   set result= my_function( arg1, arg2)

I realize that specifically arithmetic expressions in prefix notation look foreign, however I think most of the code are function calls -- and that's very normal.

If I would to think of an example where LISP notation is un natural -- it would be the if-statements.

eggy• 9 years ago

I am a polyglot, so no, I don't find it unreadable. I would probably structure it differently.

I always wonder why Lisp never ceases to get singled out for parentheses when each language uses parentheses, braces, semicolons, colons, pipes, etc...

I program in C, Forth\Factor, Lisp/Scheme, Python, J, APL, F#, C#, Asm, Basic, Clojure and others. Each has their syntax and differences, and it just takes getting used to them.

I like Hy, but as the documentation says it is not Lisp or Clojure, but Homoiconic Python with that guiding and constraining its implementation. A lot of fun, really. I only get as much pleasure from J for play.

kazinator11• 9 years ago

I always wonder why Lisp never ceases to get singled out for parentheses

POSIX shell:

   FOO=$(basename $(dirname $(cat file)))

Make:

   OBJS=$(foreach x,$(WHATEVER),\
           $(addprefix $(call BLAH,arg),....))

rthomas6• 9 years ago

As a Python user who read about 60% of SICP, yes it really is readable.

The trick is to realize that in lisps open-parenthesis denotes something more than a grouping. ( means "function call", and a change in scope.

So if it really was written in Python, it would look something like this:

  setv(result, sub(div(add(1, 2, 88), 2), 8))

Really, this is clearer what's happening under the hood than infix notation. + is a function, and so is =. Why is it called the way that it is, other than convention? It's just syntactic sugar. Doing it with prefix notation makes every function have the same syntax, which makes nesting functions easier to look at and reason about.

evanwolf• 9 years ago

Next on the wishlist: an APL inside Python.

wodenokoto• 9 years ago

Will this give me lisp in jupyther notebooks?

mmaul• 9 years ago

You can already have lisp (common) in jupyter notebooks via cl-jupyter. Here is an example https://github.com/mmaul/clml.tutorials/blob/master/CLML-Win...

If you want to use Hy in a note book you will need a jupyter kernel for it and I belive there is one called hy_kernel

amelius2• 9 years ago

Is this like clojure, but for python instead of java?

kirbyfan64sos2• 9 years ago

Yup, although Hy tries to stay much closer to Python's semantics than Clojure to Java's.

gcoda• 9 years ago

Reason I am trying it - best "hello world" example ever.

(print "I was going to code in Python syntax, but then I got Hy.")

souenzzo2• 9 years ago

Hy – A Lisp-flavored Python

Related Stories