The APL Orchard

Razetime

02:04

tio.run/##SyzI0U2pTMzJT9dNrShJzUtJTfn/v/… why is it saying 'Boolean singleton value required'?

Marshall

@Razetime The value used for a guard has to be a boolean singleton (a 1-element array containing 0 or 1). You probably want 0<⍵: instead of ⍵:.

Razetime

thanks a lot

ngn

07:01

@cannadayr tbh i don't understand why they had to replace all of linux

Adám

07:28

@ngn What's the alternative? Partial replacement? How'd that work?

ngn

@Adám keep the original problematic drivers, comply with the gpl by publishing the rest of your kernel changes (which as far as i understand consist mostly of removing code), keep the k language and apps proprietary

Razetime

09:13

@dzaima is there a way to draw arcs in APLP5?

rak1507

This is very annoying why does ↑ pad with 0s rather than spaces

⋄ ↑(⊂⊣∘⍕,'*'⍴⍨⊢∘≢)⌸,+/¨⍳6 6

Dyalog APL

@rak1507 0x0.st/iE5l.txt

Razetime

my logic says that instead of spaces it has been padded with the empty string

and apl is printing them as 0's with spaces between them

maybe just remove all the zeros in the final result

rak1507

it still adds spaces in between characters too

not sure what's going on

with -style=max they seem to have +s on the border, whatever that means

Razetime

try ]box on

it's probably packaging each * separately

rak1507

09:19

yeah I have ]boxing on -style=max

Razetime

ok so is the tally converted to char?

rak1507

Ah it isn't, that's the problem

Razetime

⋄⍪(⊂⊣∘⍕,'*'⍴⍨⊢∘≢)⌸,+/¨⍳6 6

rak1507

Alright fixed, thanks for the help @Razetime :)

Dyalog APL

@Razetime 0x0.st/iE5k.txt

rak1507

09:25

Ended up with {↑⍕(⊂⊣,'*'⍴⍨⊢∘≢)⌸,+/¨⍳⍵}

Ah damn it, not properly formatted

Razetime

oh, changed to dfn

I used the forbidden unary ⍪

rak1507

I should probably do some physics just now, rather than APLing

Razetime

that one dosn't have pormatting either lol

⋄{↑⍕(⊂⊣,'*'⍴⍨⊢∘≢)⌸,+/¨⍳⍵ ⍵}6

Dyalog APL

@Razetime
 2 *  3 **  4 ***  5 ****  6 *****  7 ******  8 *****  9 ****  10 ***  11 **  1
      2 *

rak1507

need a ⍤

Razetime

09:28

now only it get it on 12 lines

{↑⍕⍤(⊂⊣,''⍴⍨⊢∘≢)⌸,+/¨⍳⍵ ⍵}6 has the same output as ⍪(⊂⊣∘⍕,''⍴⍨⊢∘≢)⌸,+/¨⍳6 6

cya anyway

have fun with physics

dzaima

@Razetime i don't think so

Razetime

cool, no problem

Adám

09:50

@Razetime Notice how your *s caused italic text!

rak1507

dyalog.com/blog/2018/09/… the fact problem 3 doesn't work with ]boxing on annoys me slightly

Razetime

10:33

@Adám oh sorry about that lmao

Adám

11:32

@rak1507 Well, it does work, but looks different. You can always add monadic ⎕FMT to make it "work".

rak1507

anyone here happen to know haskell? don't think there's a haskell chat room :/

dzaima

11:56

@rak1507 there was, but it's frozen

Dyalog APL

12:30

<kritixilithos> @rak1507 i know a tiny bit, also the ppcg discord has a channel for 'of monads and men' although that is inactive as well

chrispsn

12:45

@cannadayr k9 will (apparently) eventually make it to risc-v, so who knows

2

Adám

@rak1507 chat.stackoverflow.com/rooms/110313/…

rak1507

13:22

Is there a way to append (1 2 3) to (4 5 6) so you get 2 3 ⍴ 1 2 3 4 5 6? I feel like ⍪ can probably do it but I'm struggling to figure out how

ngn

@rak1507 you should make 1 2 3 a 2d array before you catenate 4 5 6 below it

⋄(1 3⍴1 2 3)⍪4 5 6

Dyalog APL

@ngn
1 2 3
4 5 6

rak1507

Yeah, is there an easy way of doing that with variable length arrays?

ngn

@rak1507 you could also use "mix": ⋄↑(1 2 3)(4 5 6)

Dyalog APL

 @ngn
1 2 3
4 5 6

dzaima

13:26

@rak1507 also 1 2 3,[.5]4 5 6

Bubbler

⋄ 1 2 3 ,[.5] 4 5 6

Dyalog APL

@Bubbler
1 2 3
4 5 6

ngn

right, "laminate"

rak1507

↑,⍥⊂ I guesss this works

Oh right

,[.5]

Bubbler

But: ⋄ 1 2 ,[.5] 4 5 6

Dyalog APL

13:27

@Bubbler LENGTH ERROR

ngn

@rak1507 apl arrays are always rectangular (not "ragged"), so you must fill the gaps with something. if 0 is ok as a filler - "mix" (↑) does exactly that.

Bubbler

@rak1507 So that's (↑,⍥⊂) the right way for different lengths

(Filler is 0 for numeric arrays and space for char arrays)

Adám

@rak1507 I like ⋄ 1 2 3 ⍉⍤,⍥⍪ 4 5 6 because it looks so cute.

Dyalog APL

@Adám
1 2 3
4 5 6

ngn

13:43

i got a mysterious upvote on this 4-year-old answer and i noticed it can be shortened. mini-challenge: can anyone get to 16 bytes?

my pronoun is monicareinstate

the question was answered 2 hours ago and then 1 hour ago, so I guess it's not mysterious

ngn

@mypronounismonicareinstate mystery solved :) challenge still stands

Adám

@mypronounismonicareinstate Welcome to the Orchard, btw. Intested in APL now?

my pronoun is monicareinstate

14:03

not really (I occasionally join this room silently, but I do not know APL)

Adám

@mypronounismonicareinstate We can fix that.

TessellatingHeckler

⋄ ]runtime -compare '{0}¨3 3⍴⍳9' '≠⍨3 3⍴⍳9'

Dyalog APL

@TessellatingHeckler VALUE ERROR

Adám

@TessellatingHeckler No user commands for the bot (and you can't copy in cmpx either).

TessellatingHeckler

{0}¨3 3⍴⍳9 → 3.5E¯7 |   0% ⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕
≠⍨3 3⍴⍳9   → 4.6E¯7 | +31% ⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕

then

I was predicting that not using a d-fn and each, and only using primitives, would be faster. And it isn't. How come?

ngn

14:10

@TessellatingHeckler {0} is an idiom

Adám

@TessellatingHeckler {0}¨ is an idiom

ngn

even better

TessellatingHeckler

ooohkay

dzaima

interestingly enough

Adám

Operators have an overhead to set up.

@dzaima However:

      a←30 300⍴⍳9
      ]runtime -c {1}¨a =⍨a 1⍨¨a

  {1}¨a → 5.5E¯4 |    0% ⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕
  =⍨a   → 1.2E¯7 | -100%
  1⍨¨a  → 3.7E¯7 | -100%

TessellatingHeckler

14:15

set the array up first:

M←300 300⍴⍳9
]runtime -compare '{0}¨M' '≠⍨M'

{0}¨M → 4.4E¯7 | 0% ⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕
≠⍨M → 4.5E¯7 | +2% ⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕

dzaima

was trying to set up more proper tests (also with factorio not running in the background..) but seems y'all have got that covered

Adám

Interesting:

      a←⍳15
      ]runtime -c {2}¨a 1+=⍨a 2⍨¨a

  {2}¨a → 1.7E¯6 |   0% ⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕
  1+=⍨a → 3.0E¯7 | -83% ⎕⎕⎕⎕⎕⎕⎕
  2⍨¨a  → 3.3E¯7 | -80% ⎕⎕⎕⎕⎕⎕⎕⎕

I think this is enough reason to allow A¨ and A⍤k

Razetime

@dzaima +1 for mentioning factorio

dzaima

beat that, Dyalog /s

Razetime

java power... or something

what is Dyalog's IDE written in?

Adám

14:23

@dzaima Interesting. Your array of 1s is int, right? Not bits.

@Razetime C.

Razetime

hmm should be faster then right

dzaima

@Razetime as a hint, a 1e100 item array would have the same performance

Adám

@dzaima Oh, 2⍨ not 2⍨¨

Razetime

oh

dzaima

@Adám eh, same thing

(≢a is 2147483647 as it's truncated to an int, not that the specific size changes anything anyways)

Marshall

14:28

@ngn Nailed it (and found a BQN bug in the process; hard refresh if it doesn't work).

Razetime

nice pillow

ngn

@Marshall yes, you nailed it!

Razetime

Are those quotes supposed to be there?

Marshall

...but then I looked at other answers and Moris beat me to it by four years.

ngn

@Marshall huh! i hadn't noticed that

moris's code is exactly the same as what i had

Marshall

14:33

@Razetime In the output, yes. That's how it indicates a character array, since otherwise something like "1 2"≍"3 4" would be confusing.

Razetime

14:45

interesting

so there a command for removing those then

Adám

15:02

Meet The 2020 APL Problem Solving Competition Winners!

Marshall

15:13

@Razetime No. The formatter's not really part of the language; it's just how results are shown. In dzaima/BQN you can use •← to print a string directly, but the online version doesn't have any mechanism like that yet.

dzaima

@Marshall in dzaima/BQN all printing/formatting is equal and all of it depends on global state (i.e. )oneline, )quote, •pp). i think having a •-something for direct printing would be good

Adám

@dzaima •i and •o would be nice, also as inverses, so you can do F⌾•i 0 to get input, transform it, and output it.

dzaima

@dzaima (but with that i still think •← for print-debugging (so full formatting) without need to worry about spaces (and, though unsupported currently, allowing printing functions/modifiers directly in-place) is wonderful)

Adám

@dzaima Wouldn't an operator that is a no-op with the side-effect of printing be better for debugging? Like {⎕←⍺⍺ ⍵} or maybe {⍺⍺ ⎕←⍵} in APL.

dzaima

@Adám edited in a counter-argument as you posted that :p

Adám

15:20

@dzaima True, but it is really awkward to output intermediary values while a train is running.

dzaima

@Adám true, i've been annoyed by that. {•←𝕨𝔽𝕩} does seem useful for that

cannadayr

@chrispsn very interesting do you have any links you could send me?

ngn

15:33

@cannadayr are you in the google group? have you tried k9?

cannadayr

@ngn i am not and have not.

ngn

15:53

@cannadayr see the links at the top of shakti.com

Adám

16:37

@cannadayr In general, you can find links about APL-related subjects on aplwiki.com

rak1507

17:04

damn the fact the winner didn't know any APL before and was able to win shows some serious dedication, I assumed everyone who got prizes would've been APLing for years

Adám

@rak1507 I don't think I'd have a chance in the competition, and I've been doing APL for over 3 decades.

rak1507

Interesting, what do you think makes people likely to do well in it if not experience?

ngn

people with decades of experience tend not to be students :D

rak1507

I forgot it was a student competition, fair enough, I guess doing CS at uni or something helps

Adám

@rak1507 No, that'll probably lessen your chances.

rak1507

17:13

It'd probably help more than doing any other subject though

Adám

@ngn When I has 2 decades of APL experience, I was still a student :-)

2

@rak1507 Mathematical inclination, I'd say, and attention to detail.

rak1507

17:38

Fair enough

Adám

@ngn could win the professional prize if he wanted to.

Jeff Zeitlin

@Adám - As far as I am concerned, when it comes to programming languages, regardless of how many years of using the language I have, I am always and forever a student... :)

Adám

@JeffZeitlin Regarding the language, sure, but I was actually in a school.

Jeff Zeitlin

@Adám - OK... but then, there's always SHK... the School of Hard Knocks... :)

user41805

18:01

@rak1507 also, 3/4 of the winners mentioned ppcg, woot

rak1507

18:12

~~what is ppcg~~

Dyalog APL

<kritixilithos> =programming puzzles and code golf, a former name for this site

rak1507

Oh cool

@Adám in your bounty thing, what's the reason for 'The challenge was posted before 2020'?

Adám

@rak1507 New challenges get more attention anyway and I didn't want to simply reward all new APL answers, but rather try to get APL to "catch up" on older ones. It has the additional benefit for the site as a whole, to get some focus on older challenges with untapped potential.

rak1507

Fair enough!

rak1507

18:44

Can anyone find a golfier way to do:

⋄ ↓⍉2(⊥⍣¯1)1 2 4 8 7 14

Dyalog APL

@rak1507 DOMAIN ERROR

rak1507

oh ffs

↓⍉2(⊥⍣¯1)1 2 4 8 7 14
┌───────┬───────┬───────┬───────┬───────┬───────┐
│0 0 0 1│0 0 1 0│0 1 0 0│1 0 0 0│0 1 1 1│1 1 1 0│
└───────┴───────┴───────┴───────┴───────┴───────┘

formatting is butchered but you get the idea

Adám

@rak1507 (⊥⍣¯1)→⊥⍣¯1⊢

@rak1507 Ctrl+K!

rak1507

I did ctrl K

Maybe I didn't do it correctly

ngn

@rak1507 1 2 4 8 7 14 -> ⎕a⍳'BCEIHO'

Adám

18:47

'ABDHGN' if ⎕IO←1

rak1507

clever, thanks

Tbh I don't think this is where the byte saving will be important

Adám

Same length but simpler: ⋄ (4⍴2)∘⊥¨⎕A⍳'ABDHGN'

ngn

@Adám delenda est

Dyalog APL

@Adám 15 30 60 120 105 210

Adám

Meh, I mean ⋄ (4⍴2)∘⊤¨⎕A⍳'ABDHGN'

Dyalog APL

18:48

@Adám 0 0 0 1 0 0 1 0 0 1 0 0 1 0 0 0 0 1 1 1 1 1 1 0

rak1507

nice

ngn

⋄2⊥⍣¯1⊢5 11 19 34

Dyalog APL

@ngn DOMAIN ERROR

Adám

@ngn bot is paranoid.

ngn

⋄↓2⊥⍣¯1⊢5 11 19 34

Dyalog APL

18:51

@ngn DOMAIN ERROR

ngn

@Adám what is it afraid of?

rak1507

How can I flatten a vector of vectors?

ngn

@rak1507 ∊

rak1507

⋄10|(↓⍉2⊥⍣¯1⊢⎕A⍳'ABDHGN')(-,⍥⊂+)¨⍨⊂4 5 6 7

Dyalog APL

@rak1507 DOMAIN ERROR

rak1507

18:52

∊ flattens it all

Adám

@rak1507 ⊃,/

rak1507

Thanks!

great now I need to apply this function recursively, time to kill my PC

Adám

@rak1507 Pro-tip for when nobody is around: APLcart.

rak1507

Ah yeah

Adám

@rak1507 Why kill your PC? If you forget a stop condition, you can always interrupt.

@ngn Tight loops that cannot be interrupted with a weak interrupt, and bugs in the inverse code that crash the interpreter.

I've since improved the bot's code, but it hasn't been updated yet.

ngn

18:56

@Adám this message seems to contradict your previous message :)

Adám

@ngn I know it seems that way, but a user can issue strong interrupts while an APL green thread can only issue weak interrupts to kill its fellow green thread without killing itself.

ngn

@Adám green threads are very ethical

rak1507

hmm currently this algorithm that I'm writing would potentially need a a vector of 12^12 other vectors

ngn

@rak1507 what are you solving?

rak1507

codegolf.stackexchange.com/questions/69187/… but please don't beat me to it <3

I think I'll go in reverse order

ngn

19:08

@rak1507 i don't care much about rep

rak1507

My current thoughts are a function that takes something like (0 0 0 0)(0 0 0 1)(0 0 0 2)... the current paths and returns all new possible paths from that

And then use the power operator or something to apply it to 0 0 0 0 15 times or however many it takes to generate all 10000

ngn

@rak1507 for the thing we golfed, shouldn't you include 4-vectors with 2 consecutive 1s too?

rak1507

And then some sort of search and just print it out

TessellatingHeckler

@ngn I think you have a pretty good reputation already

rak1507

@ngn good point, up to 3 didn't see that

only 3 more characters

ngn

19:15

⋄⎕io←1⋄↓↑∪,(-⍳4)∘.↑,\3⍴1

Dyalog APL

@ngn
 1 0 0 0  0 1 0 0  1 1 0 0  0 0 1 0  0 1 1 0  1 1 1 0  0 0 0 1  0 0 1 1  0 1 1
      1

rak1507

⋄↓⍉2⊥⍣¯1⊢⎕A⍳'ABCDFGHLN'

Dyalog APL

@rak1507 DOMAIN ERROR

rak1507

one byte less

ngn

@rak1507 usually ⎕io← isn't counted

rak1507

19:18

ah

fair enough

Got any bright ideas for the main meat of the program?

ngn

@rak1507 not yet

ngn

20:18

@rak1507 i have an idea now

rak1507

oh cool

ngn

compute the matrix of neighbours - 0000..9999 along the first axis and what we golfed before (with their negatives too) along the second axis

then compute the distances from 0000 to each of 0000..9999

rak1507

The distances don't matter, all that matters is the sequence right?

ngn

then use recursion or maybe ⍣ to recover the path from the input to 0000

rak1507

Ah fair enough

ngn

20:20

@rak1507 yeah, it's just an intermediate step

rak1507

sounds pretty tough

(-,⍥⊂+)

I was using this for plus or minus

ngn

@rak1507 not trivial, but i guess it should be doable in <100 bytes

rak1507

Yeah

ngn

20:33

⋄⎕io←0⋄x←9871⋄ m←10⊥¨10|(,⍳4⍴10)∘.+,∘-⍨↓↑∪,(-1+⍳4)∘.↑,\3⍴1⋄t←{⍵⌊1+⌊/⍵[m]}⍣≡0,⍴⍨9999⋄1↓¨⍕¨1e4+1↓{×⍵:⍵,∇a[⊃⍋t[a←⍵⌷m]]⋄0}x

Dyalog APL

@ngn 9881 9991 0001 0000

ngn

x input, m matrix of neighbours, t distances

rak1507

21:28

(ignore the crap code) tio.run/##SyzI0U2pTMzJT///… why does this error?

Works fine for me :/

ngn

@rak1507 probably because tio uses an old version of dyalog

rak1507

how frustrating

Ah using ∪ doesn't account for different paths to the same number

is there a way to get the elements with unique left parts?

ngn

@rak1507 can't we output any shortest path?

rak1507

Yeah, as long as they're equally short, but atm I'm getting cycles, so when I go to find the path, I could find a longer one

If that makes any sense haha

ngn

@rak1507 what do you mean by "elements with unique left parts"?

rak1507

21:39

┌─────────────────┬─────────────────┬─────────────────┬─────────────────┬─────────────────┬─────────────────┬─────────────
│┌───────┬───────┐│┌───────┬───────┐│┌───────┬───────┐│┌───────┬───────┐│┌───────┬───────┐│┌───────┬───────┐│┌───────┬────
││0 0 0 1│0 0 0 0│││0 0 0 9│0 0 0 0│││0 0 1 0│0 0 0 0│││0 0 9 0│0 0 0 0│││0 0 1 1│0 0 0 0│││0 0 9 9│0 0 0 0│││0 1 0 0│0 0
│└───────┴───────┘│└───────┴───────┘│└───────┴───────┘│└───────┴───────┘│└───────┴───────┘│└───────┴───────┘│└───────┴────
└─────────────────┴─────────────────┴─────────────────┴─────────────────┴─────────────────┴───────────────…

my stuff looks like this, where the right is what produced what's on the left

so I was thinking I'd backtrack at the end

ngn

@rak1507 so, group by left part, and pick one from each group?

rak1507

yeah something like that

ngn

@rak1507 train: (⊃¨{⊂⊃⍵}⌸⊢)

rak1507

amazing thanks

f⍣12 gets 10000! amazing

time to write the backtracking and then I think I'm basically done

Adám

22:13

@rak1507 Polyfill.

TessellatingHeckler

23:19

⋄ (' '∘≠(/)⊢)'a b'

Dyalog APL

@TessellatingHeckler SYNTAX ERROR

TessellatingHeckler

SYNTAX ERROR is "The function does not take a left argument", but why? The tree view of the train looks like I intend, there is no left argument for my train (intentionally), there is a left argument for not-equal and it does take one

this is going to turn out to be something I've seen before, I can feel it

⋄ {(' '≠⍵)/⍵}'a b'

Dyalog APL

@TessellatingHeckler ab

TessellatingHeckler

perfect

⋄ ((' '≠⊢)/⊢)'a b'

Dyalog APL

@TessellatingHeckler 1

TessellatingHeckler

23:24

broken

⋄ ((' '≠⊢)(/)⊢)'a b'

Dyalog APL

@TessellatingHeckler 1

TessellatingHeckler

isolate the reduce so it's compress/replicate ... still no

Bubbler

@TessellatingHeckler Slashes don't translate well to trains

You can use ⊢⍤/ though

⋄ ((' '≠⊢)⊢⍤/⊢) 'a b'

Dyalog APL

@Bubbler ab

Bubbler

Also (' '≠⊢) → ' '∘≠

⋄ (' '∘≠⊢⍤/⊢) 'a b'

Dyalog APL

23:30

@Bubbler ab

TessellatingHeckler

@Bubbler But what's wrong with it when the slash is isolated inside parens to avoid it being treated as an operator?

Bubbler

@TessellatingHeckler Isolated slash alone is still a slash syntax-wise

So the previous workaround to force it to a function was (/⍨⍨)

which makes / an operand to an operator

TessellatingHeckler

:o I could have sworn I've used that before; I see it, in the -trains=tree view it's binding as an operator, the tree view doesn't look like I intend

Thank you, Bubbler

Bubbler

Relevant golfing tip

Btw, if the sole purpose is to remove spaces, you can just ~∘' '

⋄ (~∘' ') 'a b'

Dyalog APL

@Bubbler ab

TessellatingHeckler

23:37

@Bubbler "without space" 😃 that's beautiful

Transcript for