The APL Orchard

@Razetime f⍣{0} will run forever, as will f⍣⊢ if f always returns 0. Infinite tail recursion works too: {code goes here ⋄ ∇⍵}. In a tradfn, you can be explicit about it:

:Repeat
    code
    goes
    here
:EndRepeat

Razetime

11:24 AM

there are english words in APL?

Adám

@Razetime Thanks. Fixed.

@Razetime Some APLs have keywords, but they all begin with : to preserve APL's "no reserved words".

Razetime

having reserved words would be blasphemous indeed

{(⎕TS[5]=0)^(⎕TS[6]=0)⋄{,'BONG'}¨⍳⎕TS[4]⋄∇⍵}

I tried this, and it's only displaying a single value

Adám

@Razetime This will terminate before reaching ∇⍵ because it returns the value from the first statement.

Razetime

hmm

Adám

Do you mean to print? (⎕←)

Razetime

11:29 AM

yes

{(⎕TS[5]=0)^(⎕TS[6]=0)⋄{⎕←'BONG '}¨⍳⎕TS[4]}

why isn't this printing anything?

Adám

@Razetime It begins by evaluating 0∧0 (most likely, at least) i.e. 0, which it then returns.

Razetime

will tail recursion work here?

Adám

Sure, but what are you actually trying to do?

Razetime

45

Q: Big Ben goes BONG

Your task is to create a program that is to mimic Big Ben (Twitter). Task Whenever a new hour has begun (according to your computer's local time or UTC), you have to output BONG repeated hour times (with a space between the words). For example, when it's 3 o' clock, you have to output BONG BONG...

this question

I linked it up there as well

Print BONG <hour> number of times whenever an hour is reached

loop forever

Adám

Ah, so you want to check all the time. But you should wait for a second after printing, or you'll end up printing multiple times per hour.

Razetime

11:35 AM

yeah lol

I was more worried that I didn't know how to make an infinite loop first

Adám

Also, you'd want a conditional, not just a statement.

Razetime

f←{(⎕TS[5]=0)^(⎕TS[6]=0)⋄{⎕←'BONG '}¨⍳⎕TS[4]⎕DL 1}

this look better?

or should the be a newline

Adám

Several issues: You'll still return on the first statement, and ⎕DL will return a value close to 1, which'll then pair with ⎕TS[4] and mess everything up.

Razetime

so I'm not fitting the output criteria in many ways lmao

Adám

Also, ⎕← will print to a new line.

And you'll have a trailing space.

Razetime

11:39 AM

so I need a full conditional

Adám

You could use ⍣

Razetime

does :If exist?

Adám

Of course, but only in tradfns. Dfns use guards, but I don't think that'll work well here.

Razetime

so, ⍬⍣{} ?

Adám

You need a function on the left and an expression on the right of ⍣

Razetime

11:42 AM

{⎕←'A'}⍣{0} this, then

Adám

Yeah, that'll print As forever.

Razetime

in the IDE it doesn't seem to do anything

on separate lines, correct?

Adám

Did you give it an argument?

@Razetime Not sure what you mean by that.

Razetime

I mean It will not print AAAAAAAA

but
A
A
A
A
A
A

Adám

Nope, but you can get that with ⍞← instead of ⎕←

Razetime

11:45 AM

⍞← for newlines, got it

Adám

But a much better approach is to construct the whole BONG phrase and output it at once.

⋄ 1↓(5×⎕TS[4])⍴' BONG'

Dyalog APL

@Adám BONG BONG BONG BONG

Adám

⋄ ⎕TS

Dyalog APL

@Adám 2020 9 1 4 45 52 117

Razetime

{,'BONG '}¨⍳⎕TS[4] this is what I was doing

Adám

11:47 AM

Sure, but that seems a bit convoluted when you could also do ⎕TS[4]⍴⊂'BONG '

Razetime

I think that's a good candidate for APLcart

Adám

What is?

Razetime

repeat array n times

this snippet ⎕TS[4]⍴⊂'BONG '

@Adám I guess this is bettwer because it keep only one space between all of the bongs

Adám

So it'd be Iv⍴∘⊂Y Array of shape Iv filled with repetitions of the entire array Y

Razetime

yeppo

The bind operator is for.. safety?

Adám

11:52 AM

to make it a single function

Razetime

⎕TS[4] is in 24 hours :/

First need to convert it to 12 hr

never mind I'm dumb

Marshall

@dzaima Fixed. Thanks!

TessellatingHeckler

3:12 PM

@Razetime you could compare 0 0≡⎕TS[5 6]

Razetime

3:46 PM

@TessellatingHeckler good idea. Then we can multiply that expression by 4 to determine what gets displayed.

Razetime

4:17 PM

⋄{{,'-'}¨⍳⍵}5

Dyalog APL

@Razetime - - - - -

Razetime

'-'⍴5

⋄'-'⍴5

Dyalog APL

@Razetime DOMAIN ERROR

Razetime

⋄5⍴'-'

Dyalog APL

@Razetime 0 ⍴ 0

Razetime

4:18 PM

⋄5⍴'-'

Dyalog APL

@Razetime -----

Razetime

hmm

⋄5⍴'-'⍪5⍴'-'

Dyalog APL

@Razetime -----

dzaima

@Razetime missing parenthesis around 1st 5⍴'-'?

(either way, ⍪ is equal to , for vector arguments)

i suspect that what's wanted is (5⍴'-'),[.5]5⍴'-' (or with ,[1.5])

Razetime

I'm trying to do this one.

what I've come up with is this:

⋄{⍪(⍪{'|',(⍵⍴' '),'\'}¨⍳⍵)((⍵+1)⍴'-')}5

Dyalog APL

4:30 PM

@Razetime
  | \
  |  \
  |   \
  |    \
  |     \
 ------

Razetime

(with ⎕IO←0)

problem is, the first 5 lines have spaces before them

dzaima

@Razetime doing ]box on would help seeing what's actually happening there

Razetime

ah, looks like the first five lines are enclosed

dzaima

@Razetime you probably don't want to use monadic ⍪ at all

Razetime

Well, I don't know what else I wanna use really

dzaima

4:34 PM

it creates a n×1 matrix whereas you want to end up with an n×m where m is the horizontal length; monadic ↑ is useful for that (taking a vector of vectors as an argument)

Razetime

⋄]box on⋄{({'|',(⍵⍴' '),'\'}¨⍳⍵)((⍵+1)⍴'-')}5

Dyalog APL

@Razetime
VALUE ERROR
  | \  |  \  |   \  |    \  |     \   ------

rak1507

{↑(⊂(⍵+1)⍴'-'),⍨'|'∘,¨(⍳⍵)⌽⍤↑¨⊂'\'} is a quick dfn I threw together

dzaima

@DyalogAPL something something no ]-commands in the bot

TessellatingHeckler

Razetime

4:38 PM

@rak1507 cool answer

TessellatingHeckler

An APLCart game :)

dzaima

@rak1507 (fwiw Razetime's solution can go to that size with simple changes; but there is at least one golfier approach)

rak1507

it breaks for 0 but a lot of commenters said it was an edge case so

¯\_(ツ)_/¯

Razetime

but it's in the question lol

dzaima

@Razetime that's still a 2-item vector of (a vector of vectors) and (a vector); you want to end up with a single matrix

Razetime

4:41 PM

hmm

@TessellatingHeckler you made this?

Adám

@TessellatingHeckler :-D Maybe I can add that to aplcart.info/quiz

Razetime

@dzaima I tried an enclose on the second vector and now it prints the pattern correctly, just not matching.

dzaima

@Razetime if you do that, you could use , instead of stranding to get a vector of the lines

TessellatingHeckler

@Razetime yep, I just thought of it in the context of "how could flashcards / spaced reptition learning help with APL?". Rumour has it that learning a foreign language, having flashcards of words on their own is less helpful than having a whole sentence with context, and some words missing - the context helps you remember the words where they are used and connected to other things. So, why not APL with bits missing, and some surrounding context as a prompt

Adám

@TessellatingHeckler How do you decide which ones can go?

TessellatingHeckler

4:45 PM

@Adám Has that been there all this time? :D

Razetime

@TessellatingHeckler I like it.

rak1507

ooh aplcart quiz

Adám

@TessellatingHeckler Only since May/June.

TessellatingHeckler

@Adám currently any APL symbol that isn't {}()[]

at least, that's the plan; I nicked the string from your javascript lb bookmarklet and cut it back a bit

might be good if it didn't choose ← or ⎕ either, they come up quite often

@Razetime :) I'd like it more if I could answer more of them!

Adám

@TessellatingHeckler Are × and ⌊ valid answers for "A⍰B Logical AND"?

Razetime

4:50 PM

I tried mixing, but it comes out like this:

⋄{↑((⍪{'|',(⍵⍴' '),'\'}¨⍳⍵)(⊂(⍵+1)⍴'-'))}5

Dyalog APL

@Razetime 0x0.st/iEGk.txt

TessellatingHeckler

@Adám it picks a row from the tsv, and cuts a symbol out, so it's only a valid answer if it's exactly what was in that spreadsheet row, it doesn't do any evaluation or anything clever

Adám

@TessellatingHeckler Right, but that is a bit unfair to the user. I had to deal with that in my crossword checkers.

Razetime

I don't know enough list manipulation to figure this out, needs a lot more trial and error

or a different function

Adám

@Razetime You just need to stack the bottom line below the matrix instead of mixing them.

Razetime

4:51 PM

@rak1507 how do you get spaces without the character?

dzaima

@Razetime you are mixing a vector of vectors of vectors there, whereas here you definitely want to execute ↑ on a vector of vectors of characters. (or, from a different angle, forget about the last row for now)

rak1507

↑

n↑'whatever' pads with spaces

dzaima

@Razetime as I suggested previously, you also don't want stranding (stranding being array creation by juxtaposition, i.e. as in (1 2)(3 4); you essentially have {↑(…)(⊂…)}, but you don't want to make that 2-item vector. you could insert a , between )(, but there are golfier ways around that problem)

TessellatingHeckler

@Adám for now, the user will just have to grumble about how unfair it is

dzaima

@Razetime also no monadic ⍪!

rak1507

4:56 PM

⍪ is pointless ↑ does what you want most likely

↑

Razetime

@dzaima awwww

but it's so cute

rak1507

not ↓

Adám

@Razetime Btw, once you have the matrix, you can simply append a single - below it, and it'll be extended to all columns.

rak1507

ooh

Adám

@rak1507 You know you can edit your messages for a short time?

rak1507

4:57 PM

I do now

dzaima

@Razetime you can still use it, just not monadically

Razetime

⋄ {{⎕←⍵}¨({'|',(⍵⍴' '),'\'}¨⍳⍵),⊂(⍵+1)⍴'-'}5

Dyalog APL

@Razetime Illegal code

Razetime

this is what I ended up with

Adám

@Razetime Don't output. You want to assemble the result.

rak1507

4:58 PM

{'-'⍪⍨↑'|'∘,¨(⍳⍵)⌽⍤↑¨⊂'\'}

Adám

@rak1507 ,[1] is the same as ⍪ (for matrices).

rak1507

ah

that then

Razetime

so I have a vector of strings, but it's all on a single line

rak1507

⋄{'-'⍪⍨↑'|'∘,¨(⍳⍵)⌽⍤↑¨⊂'\'}0

Dyalog APL

@rak1507 -

rak1507

5:00 PM

:/

Adám

(I've got a 20-byte flat solution with no loops.)

Razetime

but you're Adám that's unfair

rak1507

Of course you have

Adám

Except the 0 case :-(

Razetime

oh lol

everyone falls to the 0 case

It's a bit late here, Imma sleep

dzaima

5:01 PM

@Razetime because it's wrong!

Razetime

hard luck lol

good night to y'all

24 with the 0 case.

gn

○/

@Razetime as the answer, ↑ is exactly what you want in place of {⎕←⍵}¨

Razetime

5:02 PM

@Adám I'm assuming those 4 bytes are ⍵>0⋄

dzaima

@Adám (trying to figure out where that 1 byte went)

Adám

@Razetime No, my solution is a train, not a dfn. (Also, that'd be ⍵>0:)

Razetime

oh well

Adám

@dzaima What, you've got 21?

dzaima

@Adám yeah (dfn)

Adám

5:05 PM

Wow, that probably means you're packing in a lot more action, since you loose 2 on the braces and you need ⍵ here and there.

dzaima

@Adám "here and there" ≡ once

Adám

My dfn equivalent has ⍵ thrice.

Ooh, just hit 19 without 0 case. Completely new approach. So, 23 including 0.

So many interesting ways to attack this one

TessellatingHeckler

@Adám Soon you could add "tsv passes 0.5MB"; currently at 0.493MB. (I've been wondering how much the TIO example links will bulk it up).

dzaima

still at a 21 byte train (plus 2 23-byters)

Adám

@TessellatingHeckler :-D But not really a milestone of interest to the user.

dzaima

5:13 PM

@dzaima (fwiw it's 18 in Extended/dzaima; 21 as a train in Dyalog)

Adám

@TessellatingHeckler The average TIO link is about 100 bytes, so probably no more than 200kB.

@dzaima Interesting. My solutions wouldn't benefit from any extensions. Is that with or without 0 support?

dzaima

@Adám no 0 support

Adám

You produce a matrix, right?

dzaima

yeah

yay 20 byte train (regular Dyalog)

Adám

Wanna know how I add 0 support to a train in 4 bytes? Adding it in 3 bytes to a dfn is obvious, I presume.

dzaima

5:20 PM

@Adám ×⍴∘⊂?

Adám

@dzaima No, that'll give an enclosed result for the normal cases, and still output whitespace (forbidden) for 0. But close.

dzaima

@Adám ×⊢⍤/?

Adám

Almost. That'll still print a newline.

rak1507

come on lets see it already, we're on the edge of our seats!

Adám

OK: Spoiler! (updated for performance)

dzaima

5:24 PM

@Adám ooh, smart. my solutions

TessellatingHeckler

"DESCRIPTION : Is Dv a valid Finnish social security number? (10=≢Dv)" - I wonder if this came from FinnAPL

Adám

@TessellatingHeckler Of course it did. There's also "Syllabisation of a Finnish word Dv"

TessellatingHeckler

⋄ (⊢⊂⍨1,1↓≢↑2</'aeiouyäö'∊⍨⎕C)¨ 'hauska' 'tutustua'

Dyalog APL

@TessellatingHeckler haus ka tu tus tua

Adám

@dzaima Mine

dzaima

5:38 PM

@Adám oh so i had another 20-byter by replacing ' ' with ''

@dzaima (s/another/your)

Adám

⋄ (⊢⊂⍨1,1↓≢↑2</'aeiouyäö'∊⍨⎕C)'Kolmivaihekilowattituntimittari'

Dyalog APL

@Adám Kol mi vai he ki lo wat ti tun ti mit ta ri

Adám

Apparently, the genitive plural of that is kolmivaihekilowattituntimittarienkolmivaihekilowattituntimittareidenkolmivaihekilowattituntimittareitten. Now you know!

I suspect an error in the declension algorithm.

Marshall

Long but cool BQN triangle maker.

Adám

5:59 PM

@Marshall Seems very convoluted. Surely you can translate one of ours?

Marshall

@Adám I disagree. It builds four sections and puts them together. Of course I can translate (" \|-"⊏˜·(⊢∾3¨∘⊏)⥊⟜2∾˘=⌜˜∘↕ is shorter, for example), but BQN will be longer because it won't put fills everywhere automatically.

Adám

@Marshall Why doesn't "-"∾˜"|"∾˘∾˜⥊"\ "/˜1∾⊢ work?

Marshall

@Adám You're using strings instead of characters in a lot of places, and the last ∾ won't automatically extend a scalar anyway (why would it?).

Adám

@Marshall Characters are in single-quotes?

Marshall

@Adám Yes.

Adám

6:12 PM

@Marshall '|'∾˘∾˜⥊"\ "/˜1∾⊢ gives me "TypeError: undefined has no properties" or sometimes "l is undefined"

ngn

⋄⎕io←0⋄' \|-'[3⍪⍨2,∘.=⍨⍳5]

Dyalog APL

@ngn INDEX ERROR

ngn

⋄⎕io←0⋄' \|-'[3⍪⍨2,∘.=⍨⍳5]

Dyalog APL

@ngn
|\
| \
|  \
|   \
|    \
------

Marshall

@Adám Negative rank isn't working on atoms for some reason. (<'|') in place of '|' works.

Marshall

6:25 PM

@Adám Two character fix; might have to wait a few seconds for Github to update.

∾˘⟜'-'⌾⍉'|'∾˘"\ "⊏˜↕≠⌜↕ works in dzaima/BQN but not in the JS version. It relies on invertible Under since the left operand to ⌾ changes its argument's shape, and only structural Under (with insufficient shape checking, it seems) is implemented so far.

"\ |-"⊏˜·∾˘⟜3⌾⍉2∾˘↕≠⌜↕ gets it to 22 characters.

rak1507

8:33 PM

A string variable called month has been assigned the value ‘April’ and another
string variable called year has been assigned the value ‘2019’ as shown
below.
Line 1 DECLARE month INITIALLY "April"
Line 2 DECLARE year INITIALLY "2019"
Line 3
The variable shortDate is to be assigned the value ‘Apr19’ using substring
operations.
Using a programming language of your choice write line 3.
this is a question in a past paper I have to do for homework, do you think I could get away with doing shortDate ← ,3 ¯3↑⍤0 1↑ month year? lol

Adám

¯2?

You'd have an extra space, no?

rak1507

Oh yeah good point, that's annoying to fix

¯2 wouldn't work

Adám

@rak1507 How about ∊3 ¯2↑¨month year or ∊3 ¯2↑¨⊂month,year?

rak1507

yeah forgot that ↑ pads with spaces for a moment

Adám

Or go tacit with ∊3 ¯2↑¨∘⊂, taking month as left argument and year as right argument.

∊3 ¯2↑ᐵ, in dzaima/APL.

rak1507

8:40 PM

dzaima APL is cool, it's crazy to me that basically everyone here has their own APL-ish language haha

like, learning APL was too easy so they decided to write their own language

Adám

∊3 ¯2↑⍥0 1, in Extended

Marshall

@rak1507 dzaima claims to sometimes learn APL best by implementing it.

Adám

@Marshall I think ngn also did it for that reason.

ngn

@Adám yep, it's easier to learn that way :P

rak1507

Fair enough, I have no idea where I'd even start with something like that, tokenisation and stuff seems very complicated

Adám

8:42 PM

@Marshall How would you do the assignment in BQN?

@rak1507 That's not really the hard part. APL-like languages are easy for a computer to parse. See e.g. an APL parser in APL here.

rak1507

whoever made that had far too much time

that's just excessive

Adám

(Actually, it is a general purpose parser, not just for APL.)

rak1507

dfns.dyalog.com/n_lisp.htm

that is insane

how do people do this stuff

Marshall

@Adám Probably ∾3‿¯2↑¨month‿year. Or just ∾⟨3↑month,¯2↑year⟩?

ngn

@rak1507 are you familiar with scheme?

rak1507

8:49 PM

not really

Adám

@Marshall There's no pair, only ≍○< right? ∊3 ¯2↑¨⍮ in Extended/dzaima becomes ∾3‿¯2↑¨≍○< but then you might as well do ∾3‿¯2↑⚇0 1∾

@Marshall ,→∾ right?

ngn

@rak1507 it's a very simple dialect of lisp. it was made for playing with interpreters and language stuff.

rak1507

yeah, it's still very impressive that someone could write a full parser for it in APL

Marshall

@Adám That's right. I think joining only once is cleaner though.

Adám

@rak1507 I'd find it more impressive if someone could write a parser for a full programming language in, say, C.

rak1507

8:53 PM

Fair enough

Marshall

@Adám No, the , in ⟨3↑month,¯2↑year⟩ is a list separator—identical to ⋄.

Dyalog APL

@Marshall ``

Adám

@Marshall Oh, I misread the angle brackets as round parentheses.

@rak1507 Have you heard of co-dfns? It is written in Dyalog APL and compiles dfns to C.

@Marshall I still think that should be ; and ‿ should be ,

rak1507

no I haven't, that sounds crazy, I feel proud of myself solving phase 1 problems and people do this stuff lmao, got a long way to go

Adám

@rak1507 What operating system do you use?

rak1507

8:56 PM

windows + linux on a vm

Adám

@rak1507 Both of which are mostly written in C. Now that's impressive ;-)

rak1507

Obviously the most impressive operating system is TempleOS though

Adám

@rak1507 Impending kOS…

rak1507

crazy stuff

TessellatingHeckler

@rak1507 Computer Science college courses spend a good deal of time on lexing, parsing, tokenising, computer language grammars, and the thing you get next (Abstract Syntax Trees). If you want to follow through how it works in general, craftinginterpreters.com is a free book that guides through building a small language interpreter in Java

rak1507

9:07 PM

Thanks for the link, will definitely check that out!

ngn

@Adám "everything we know about kos"

TessellatingHeckler

(it's trying to be a casual-practical sort of book with real working examples, not a dense academic thing with "left as an exercise for the reader" all over)

Adám

@rak1507 There's also this unfinished series by RGS.

rak1507

'define a doubly-linked list of heap-allocated strings' when this is on the first page I don't think I'll get very far with my current very limited C understanding

RGS

@Adám felt like I was reading a novel, even got goosebumps at the end

@Adám yup, have been thinking about what to put in for the 3rd post, probably gonna make smaller posts from now on so that I can post them more frequently

cannadayr

10:12 PM

"Arthur succeeded in getting it to run, but eventually got drowned in device drivers."

i am shocked and amused.

a better strategy might be to target a specific hw platform.

(most of the kernel code isn't the scheduler or malloc, its all the device drivers - it is not surprising thats the wall they hit)

like kOS on RISCv or something

Transcript for