The APL Orchard

ngn

4:16 AM

@Adám {⊃'+-|+'↓⍨2⊥5 4 2⊃¨⊂⌊∘⌽⍨' '=,⍵}⌺3 3

Adám

6:07 AM

@ngn OK, that has nothing to do with my answer. Will you post?

ngn

@Adám once I've golfed it and shown it to APL-ers, posting seems pointless...

Adám

@ngn OK, I'll attribute you then.

Cows quack

10:07 AM

@ngn how long is your apl solution to the finnish cinema challenge? mine is 55

Feeds

1:16 PM

The CXO, @mkromberg, talks about running Isolates on #AWS. #APLGermany2018-1

Tweeted by mbaas on April 17, 2018 at 9:50 AM

ngn

@Cowsquack I only have a proof-of-concept solution, about as long as yours: 51 bytes. I'll try to golf it better and I'll tell you my best byte count later today.

@Cowsquack hint: my current solution makes use of such primitives as ⍸ @ ⍣

ngn

3:01 PM

@Cowsquack I can do it in 42 bytes

Cows quack

@ngn as does mine :)

I got it to 53

Adám

@ngn Phew, wasn't easy to explain.

ngn

@Adám what's the ⍱ for? easier to explain?

Ven

@Adám Please do post the link here :-).

ngn

10

A: Render an ASCII maze

APL (Dyalog Unicode), 57 35 bytesSBCS –22 thanks to a novel solution by ngn. Anonymous tacit function taking a character matrix as argument. {⊃'+-|+'↓⍨2⊥5 4 2⊃¨⊂⍱∘⌽⍨' '≠,⍵}⌺3 3 Try it online! {…}⌺3 3 on each 3-by-3 neighbourhood, apply the following function: ,⍵ ravel (flatten) ' '≠ B...

Ven

3:15 PM

@ngn Thank you!

Adám

@ngn Yeah, I found inverting it so that the initial Boolean indicates being populated more natural. Btw. is using ⌊ instead of ∧ a K'ism?

ngn

@Adám yeah, sort of, I got that habit from k, but of course it's also just maths

Adám

@ngn If I was to start again, I don't even think I'd include ∧, but in favour of ×, not ⌊.

J. Sallé

6:49 PM

Hey @Adám, is this a postable solution or just a snippet? ''≡'(\w).*\1'⎕S'&'⊢<arg>

Adám

@J.Sallé That's a snippet, but remove ⊢<arg> and you have a proper tacit function.

J. Sallé

@Adám yeah I thought so, I'll just assign it to a variable

I'll need to add a bunch of bytes for case conversion though

Adám

⍞←'(\w).*\1'⎕S'\u&'⊢'abab'

Dyalog APL

@Adám  ABA

Adám

@J.Sallé Does ^ help?

J. Sallé

6:58 PM

@Adám yeah I was looking that up in Lesson 24 actually :p

Kinda confused by the challenge though

Adám

@J.Sallé Isograms?

J. Sallé

@Adám yeah, but I sorted it out

Adám

@J.Sallé How? The question is very unclear.

J. Sallé

that case folding doesn't help because I need to case-fold the argument, not the match itself

@Adám I interpreted the "case insensitivity" thing wrong the first time I read

@Adám ''≡'(\w).*\1'⎕S'&'(819⌶⊢)<arg> works, but I can't assign that as a function, right?

Nvm ''≡'(\w).*\1'⎕S'&'∘(819⌶) also works

Adám

@J.Sallé Yeah, that's fine.

@J.Sallé But that looks for a XYX pattern.

H.PWiz

7:07 PM

Would Variant help here?

J. Sallé

@Adám works for all the test cases though

Didn't test any more than that

Adám

@J.Sallé Those test cases are bogus.

J. Sallé

Yeah, it doesn't seem to work for 'ss'

I'll post it as is and then I'll try to fix it.

Adám

@J.Sallé Yeah, I think ⍠1 would do the trick, as @H.PWiz said.

J. Sallé

@Adám where exactly would I put that?

Adám

7:12 PM

@J.Sallé '(\w).*\1'⎕S'&'⍠1

J. Sallé

⎕←''≡'(\w).*\1'⎕S'&'⍠1⊢'ss'

Dyalog APL

@J.Sallé
0

J. Sallé

Doesn't seem to work

Oh no nevermind

It should return 0, not 1 >.>

J. Sallé

7:37 PM

12 people in the room! The TNB-link effect is indeed very strong.

moonheart08

The ~~reddit~~ TNB hug ~~of death~~

Adám

@all Hello everyone. Do you want a quick intro to APL?

Félix Gagnon-Grenier

yes!

Adám

@FélixGagnon-Grenier Do you know J already?

Félix Gagnon-Grenier

@J.Sallé in my case, I came here looking for a place where I could discuss J

@Adám No. A colleague linked a article about someone computing fractals in J. Fractals being a passion of mine, I couldn't help but try and learn J right away

going through the basics on jsoftware, I was looking for a place to exterioze all my surprises as I discover the paradigm

Adám

7:45 PM

@FélixGagnon-Grenier OK, J is a dialect of APL, but APL looks more like regular mathematical notation than J.

@FélixGagnon-Grenier You're welcome to discuss them here. Most of what can be said about J applies to APL too.

Félix Gagnon-Grenier

:)

Would you advise to learn APL first, or going for J would be an introduction to APL at the same time?

... I must say I'm quite unsettled by a language having non ascii characters

Adám

@FélixGagnon-Grenier I'm an APLer, personally, so I'm biased. For most purposes, I think APL is the better choice, as it allows you to easily use whichever programming paradigm you want, and because the notation is easier for humans to read.

dzaima

@FélixGagnon-Grenier I was too but it ends up being really useful

Adám

@FélixGagnon-Grenier Why? Many domain specific languages have their own symbols. Mathematics too.

dzaima

@FélixGagnon-Grenier IIRC J is what happened trying to remove non-ASCII from APL

Adám

7:52 PM

@FélixGagnon-Grenier Think of APL like this: It is a generalisation and harmonisation of conventional mathematical notation (esp. with focus on linear algebra), which happens to also be executable.

Félix Gagnon-Grenier

@Adám Oh indeed, I'm sure it's totally justified, and probably useful in the long run! It's just my initial reaction :)

Adám

@dzaima True, but not only that. J does contain some further generalisations and harmonisations, and a larger vocabulary, but I often find that it takes the good intentions too far, to the point where ease of use suffers.

@FélixGagnon-Grenier Let me teach you just a handful of APL built-ins (we call them primitives), and I guarantee you that you'll never forget what they mean. OK?

Félix Gagnon-Grenier

please do!

Adám

@FélixGagnon-Grenier ⌹ is matrix division/inversion. ? is random number(s). ⌈ is ceiling/max. ≤ is less-or-equal. ⍉ is transpose (axes).

@FélixGagnon-Grenier Want more?

Félix Gagnon-Grenier

do you alt + number the special characters?

hmmm... actually, I don't think I know enough to really know what to do with that. I'll continue reading, I'm sure I'll get questions

Adám

8:01 PM

@FélixGagnon-Grenier No, experienced APLers use a dedicated keyboard layout, but the IDEs provide multiple input methods. Also look here and here.

@FélixGagnon-Grenier Feel free to experiment. You can evaluate most APL expressions right here in the chat room by prefixing them with the two characters ⎕← as follows:

⎕← 3 1 4 ⌈ 2 7 1

Dyalog APL

@Adám
3 7 4

Pavel

@Adám Or anyone on Linux, 'cause we already have it preinstalled and the Dyalog GUI isn't even available.

Félix Gagnon-Grenier

⎕← x =: *:
⎕← 2x3

Dyalog APL

@FélixGagnon-Grenier
VALUE ERROR

Pavel

⎕←x←×
⎕←2 x 3

Dyalog APL

8:04 PM

@Pavel
VALUE ERROR

Adám

@FélixGagnon-Grenier That's J. This only works for single lines of APL. Also, you're trying to do 2 NOR 3 ‽

Félix Gagnon-Grenier

nah, I was trying to understand assignment, so my first idea was to try and break it

Pavel

* is Pow, × is multiplication

Félix Gagnon-Grenier

to my great surprise, x =: *: did not break

moreover, 2x3 when x =: *: gives 40.1711

Pavel

In APL, var ← val assigns val to var.

Assigning something to ⎕ prints it instead.

Adám

8:05 PM

@FélixGagnon-Grenier @Pavel If you want multiple statements for the bot, separate them with ⋄:

Pavel

I forgot how to type ⋄.

Adám

⎕←x←× ⋄ ⎕←2 x 3

Dyalog APL

@Adám
×
6

Félix Gagnon-Grenier

⎕← x =: *: ⋄ 2x3

Dyalog APL

@FélixGagnon-Grenier
VALUE ERROR

Adám

8:06 PM

@Pavel Top left button.

Félix Gagnon-Grenier

⎕← x =: *: ⋄ 2x3

Pavel

._. I knew that but accidentally had the shift key down

Dyalog APL

@FélixGagnon-Grenier
VALUE ERROR

Pavel

And typed a ⌺

@FélixGagnon-Grenier What are you doing

Adám

@FélixGagnon-Grenier Again, that's J, not APL.

@Pavel He's experimenting :-)

Félix Gagnon-Grenier

8:07 PM

> You can evaluate most APL expressions

Pavel

⎕←x←×⋄⎕←2 x 3

Félix Gagnon-Grenier

oh I completely misread that

Dyalog APL

@Pavel
×

Adám

@Pavel Don't forget ⎕← before everything you want printed.

Here is a NOR table:

⎕←∘.⍱⍨0 1

Dyalog APL

@Adám
1 0
0 0

Félix Gagnon-Grenier

8:08 PM

actually, I am trying to understand what happens when I run the J lines x =: *: and 2x3. it returns 40.1711 in my jconsole. I have no idea how to start thinking about it

Pavel

I suddently can't type ⋄

I'm hitting the same keys

But getting ⋄

Argh

It works now

Félix Gagnon-Grenier

lol

H.PWiz

@Adám *: is NAND rather than NOR

Pavel

⎕←x←× ⋄ ⎕←2 x 3

Dyalog APL

@Pavel
×
6

Adám

8:09 PM

@H.PWiz Yes, my bad, but in his case it doesn't matter.

Pavel

There we go

dzaima

@FélixGagnon-Grenier 2x3 alone seems to output 40.1711

Félix Gagnon-Grenier

oh, indeed

Adám

2x3 in J means 2 times e to the power of 3, which you can verify in APL:

⍞←2×*3

Dyalog APL

@Adám 40.17107385

Félix Gagnon-Grenier

8:12 PM

I naively assumed writing x would interpolate the value I had assigned to it

Adám

@FélixGagnon-Grenier J very often requires spaces to disambiguate. In APL, spaces are only required between identifiers.

@FélixGagnon-Grenier It does, but you must put spaces.

Félix Gagnon-Grenier

hmmmmmm

indeed, 2 x 3 returns the domain error given when writing 2 *: 3

Adám

@FélixGagnon-Grenier Yes, because 2 NAND 3 isn't meaningful. Same goes in APL:

⎕←2⍲3

Dyalog APL

@Adám
DOMAIN ERROR

Pavel

How does one do bitwise NAND?

Adám

8:15 PM

@Pavel Convert to binary, then NAND. It would be ambiguous otherwise, regarding which representation to use.

@FélixGagnon-Grenier Although technically speaking, A *: B is defined as -. A *. B, so the domain should really be all numbers.

Pavel

So <quad>DR 0

Dammit mobile chat

Adám

@Pavel You mean 11 ⎕DR?

Pavel

Yes

Adám

@Pavel That's one possibility, but then when you want to convert back, your datatype may overflow.

Pavel

So how would you do it

Adám

8:24 PM

⍞←2⊥⍲/2⊥⍣¯1⊢2 3

Dyalog APL

@Adám 1

Adám

@Pavel If I do my job right, we'll get the Under operator in 18.0, so you can write ⍲⍢(2∘⊥⍣¯1). Just tested it — yup, it works with my model for ⍢:

 Ü←{ ⍝ Under a.k.a. Dual
     ⍺←{⍵ ⋄ ⍺⍺}      ⍝ no ⍺: pass through
     ⍵ ⍵⍵{           ⍝ pass in original ⍵
         A←⍺             ⍝ modifiable array
         11::A⊣((⍺⍺)A)←⍵ ⍝ structural inversion on error...
         ⍺≢⍺⍺⍣¯1 ⍺⍺ ⍺:!# ⍝ ... or if imperfect inverse
         ⍺⍺⍣¯1⊢⍵         ⍝ try computational inverse
     }(⍵⍵ ⍺)⍺⍺{          ⍝ ⍺⍺, but:
         ⍺←⊢                 ⍝ no ⍺: pass through
         2=⎕NC'⍺⍺':⍺⍺⊣¨⍵     ⍝ if array: treat as scalar fn
         ⍺ ⍺⍺ ⍵              ⍝ else: just apply

⎕←Ü←{⍺←{⍵⋄⍺⍺}⋄⍵ ⍵⍵{A←⍺⋄11::A⊣((⍺⍺)A)←⍵⋄⍺≢⍺⍺⍣¯1 ⍺⍺ ⍺:!#⋄⍺⍺⍣¯1⊢⍵}(⍵⍵ ⍺)⍺⍺{⍺←⊢⋄2=⎕NC'⍺⍺':⍺⍺⊣¨⍵⋄⍺ ⍺⍺ ⍵}⍵⍵ ⍵} ⋄ ⎕←2 ⍲Ü(2∘⊥⍣¯1) 3

Dyalog APL

@Adám
{⍺←{⍵ ⋄ ⍺⍺} ⋄ ⍵ ⍵⍵{A←⍺ ⋄ 11::A⊣((⍺⍺)A)←⍵ ⋄ ⍺≢⍺⍺⍣¯1 ⍺⍺ ⍺:!# ⋄ ⍺⍺⍣¯1⊢⍵}(⍵⍵ ⍺)⍺⍺{⍺←⊢ ⋄ 2=⎕NC'⍺⍺':⍺⍺⊣¨⍵ ⋄ ⍺ ⍺⍺ ⍵}⍵⍵ ⍵}
1

Adám

@DyalogAPL Whoohoo!

Transcript for