The APL Orchard

nathan rogers

15:09

Now that I'm visualizing it, template on the left seems funny

Especially as you're likely to load multiple functions from the same file

File loads x

Seems nicer than the other way

dzaima

sum

nathan rogers

Yay!

dzaima

it's slow though, probably just from copying the array back and forth

nathan rogers

15:25

:(

dzaima

did the speeding up (using direct mapping in JNA). 5x better for increment, not any better for sum
https://dzaima.github.io/paste/#0hY/BSsNAEIbv@xTD9tI2ZZOtliixF@klIILorXjYJqssJBvIbkAUwVPBmxfPFU969wl8lDyJM2lEKYh72B3@/5udfwbgtfOFWYmM6Ruvawt51awKDcZmw77MR3DHAGrtG/IDmbD7Hdo15dBYPwY1wU4PhbbbpqKy19vLwRyiBKWrqgaCwXQKPkfEYxEEI3DBHNTSXCY/Ex3NYwPIb5UpVXh8dooeBYR2/QRciDALcQXaRLiKw0mlcuCLjlhwYlOsJJtSKXXcPrxemFLzTuWMSxlbx/80D3oTd/xnIBHpGNIOV8i262cZdYeUc/pg86Ym7eOLajfvbF8c0plpKckPPj/Urkbt35HIx0SR2JtNyz7wL3v3ewovYgS/AA#y0nMS9dQdwr0U9e05krLzEnNS8xN1VAvSCwuSdVLKswDCgMA

(this is another option for syntax)

anyway gotta go do other stuff

Marshall

16:00

@dzaima dzaima speed is the only way to describe this development.

Will look closer after lunch.

dzaima

@Marshall it's not published anywhere yet though

Marshall

Too slow on the edit...

This is the function I'd most like to replicate, and after that I'd probably work on FFTW bindings.

dzaima

@dzaima now it is. really gotta go now though

currently with I I* D D* and only the ability to return I D (except by mutating argument arrays probably, which is very bad)

Wezl

16:36

new poultreM release, now supports German I think, among others.

Sign here if @Wezl should stop annoyingly posting updates about his fonts

nathan rogers

@Wezl do you have demo pictures?

I wanna see it before bothering to install it

Wezl

wait a sec

nathan rogers

i have not 1, but 2 bajillion fonts already installed

Wezl

I don't have any example texts right now showing the new accented characters, but I have plenty of terminal shots

Marshall

@dzaima I'm getting ImplementationError: java.lang.ClassNotFoundException: BQNative.Load when I try load←•JLoad"BQNative.Load". Is there an extra step I need to do to build tools/ or something?

Wezl

16:44

nathan rogers

I have been looking for just this sort of tiny blocky font, but the aliasing on the slashes is murder

and so is the hook for l and i

Wezl

feel free to fork it :)

nathan rogers

I can't unsee "jambda"

dzaima

@Marshall by using the ./REPL in the base directory, load should already be defined

nathan rogers

@dzaima what REPL stuff is there?

@Wezl fib either looks like fub or fjb

dzaima

16:47

@nathanrogers the ./REPL in BQNative just loads the jar file in

Marshall

@dzaima Nope. A grep for "Load" is only turning up •JLoad and ClassLoader, so I think there's a missing file.

Latest commit I have is 14d3e2b.

dzaima

@Marshall BQNative is a whole separate repository

Marshall

@nathanrogers Possibly not what you're looking for but Fairfax is a very good bitmap font.

nathan rogers

@Marshall I have that one

its superscripts are too gaunt and tiny for me

dzaima

@dzaima (for now at least. The jar files of JNA alone are 8 times the size of dzaima/BQN, so including them in the base language seems very bad)

nathan rogers

16:51

I really like fixedsys excelsior for K, but for bqn, I need other unicode support, which is fine in vscode, but i can't in microsoft terminal

rak1507

chat.stackexchange.com/transcript/message/57339451#57339451 is relevant again

nathan rogers

feh

Marshall

@rak1507 But, like, you're fixing the fact that some characters don't look that good by using a character set where almost none of them make sense. Replacing a tooling problem with a whole lot of language problems.

Who do you think has spent more effort trying to cram primitives into their chosen character set, me or Whitney?

nathan rogers

Right, then there's the overloading problem

and how to "read" without knowing the types of the arguments

rak1507

@Marshall I think whitney is more careful about thinking what to add

nathan rogers

16:56

I think K has lost the plot on notation. It just wants to be short

It doesn't strive for clarity, and you can tell by his style of english, and the overabundance of k documentation

rak1507

overabundance?

nathan rogers

/sarcasm

Marshall

@rak1507 Which puts me in second place among the practical language designers of the world?

rak1507

@Marshall I'm not ngn, I don't think k is the perfect divine language created by the gods

nathan rogers

CMC: en.wikipedia.org/wiki/Flood_fill Bucket Fill

Wezl

16:59

@nathanrogers as an algorithm, it probably doesn't fit much with APL

nathan rogers

 board←(8 8⍴⍳64)∊4 12 19 26 25 61 53 46 39 40

@Wezl what

rak1507

@nathanrogers do you not have to specify a starting point too?

nathan rogers

I think its fine to just pass in a random starting point and just fill that

@rak1507 so in the board I passed in if you give it 11 it should fill the uper left corner

rak1507

ok

nathan rogers

@wezl what are you on? github.com/carkat/APLProjects/blob/master/raytracer/…

@Wezl github.com/carkat/APLProjects/blob/master/fun/mazeGenerator.apl

algorithms

why do "algorithms" not fit with APL?

Wezl

17:05

imperative algorithms

like the algorithm to sum a vector is "variable s, for each element: s is s plus that element, return s"

nathan rogers

{s+⍵}¨vec

?

Marshall

Regarding caution, the last time I made a significant breaking change to a primitive was when I introduced Group (⊔) last June, which was before I had anything resembling a real implementation (the insignificant change was when I made a few primitives enclose atom arguments instead of returning them unchanged). The only other thing I'm slightly considering is removing the multidimensional case from Window (↕). I think any extra caution would easily qualify as paranoia.

nathan rogers

and wezl, in APL, that's just s + vec

user

@nathanrogers I thought it was +/vec

nathan rogers

no its s + each item he said

sorry, s IS s + each item

s + +/vec

idk, its pseudocode ayway

lol when the notation is shorter than the pseudocode

user

17:09

I think by variable s, they mean int s;, or int s = 0;, so the s + is probably not needed.

Wezl

no, my point is in APL it would be {s←0⋄{s+←⍵}¨⍵⋄s} but that's obviously not idiomatic APL.

nathan rogers

it also wouldn't work

which is also stupid

user

You could use a Tradfn with those while loops to make it look more like C, I guess.

nathan rogers

because its just s + +/ vec

Wezl

so imperative "for each" doesn't usually make sense for APL

nathan rogers

17:10

right

because ¨

like you don't need for each because you can just evaluate...
s + +/ ⍵

(s + 1) + (s + 2) + (s + 3)...
is just
s + sum 1 2 3...

which is s + +/1 2 3...

so that isn't a limitaiton of APL

that's a limitation of general programming languages

user

I think you two both agree, you're just not understanding each other :)

nathan rogers

no he's saying that algorithms are not idiomatic in APL

which I think is a silly thing to suggest

user

"imperative algorithms"

Algorithms are a very general term, so I doubt that was what Wezl was suggesting.

nathan rogers

right, what does that mean? of course you aren't going to write imperitive code in APL

right, like that maze generator I linked isn't even recursive

it's iterative

and also isn't imperative

user

You can write imperative code in APL, iti just won't be idiomatic.

nathan rogers

17:13

github.com/carkat/APLProjects/blob/master/fun/mazeGenerator.apl

user

(I'm not an expert, though, so feel free to correct me)

nathan rogers

@user I would say it isn't idiomatic "modern APL" but I would say it is idiomatic "corporate and money making APL"

user

@nathanrogers Right, and those are idiomatic solutions, but translating an imperative algorithm step by step wouldn't be so nice.

nathan rogers

most large scale, in production, real world money making APL is written with imperative code

@user that's because APL wants you to think in arrays with notation, and not in loops and mutation

user

And I agree with that

It's because APL encourages natural-looking notation instead of boilerplatey and unintuitive loops and mutation that an imperative algorithm translated directly from C would be unidiomatic, even if there's another algorithm that does the same thing that would be more natural in APL.

Let me find some ugly APL code :P

dzaima

17:16

@nathanrogers the original message said "[flood fill] as an algorithm", not algorithms in general

Adám

@Wezl Please don't stop.

2

Wezl

@Adám rats, you foiled my evil plans

Adám

I need the updates to prompt me to update the link on APL Wiki.

Wezl

also the picture shows it all blurry and unhinted

user

Is this how you'd sum an array with a loop in APL (warning: ugly code)?

rak1507

17:24

@nathanrogers {⍺{(~⍺)∧{∨/(,⍵)[1 4 5 6 8]}⌺3 3⊢⍵}⍣≡2=2@⍵⊢⍺} can't find any clever way doing a cross shaped stencil

Wezl

@Adám now your name gets the respect it deserves

rak1507

that's an interesting m

Wezl

@rak1507 it looks better (fits better) this way than the obvious ~(5 5⍴⍳25)∊6 8 11 13 16 18 21 23

rak1507

fair enough

Marshall

@dzaima So the mutating arguments thing does work, and I was able to call my C filtering function and get accurate results. A sane option might be to use & instead of * to indicate an argument might be changed and return a list of all the new values if there are any of these. Although the user should also be able to specify that a pointer might change but they don't want results back so this needs more work.

dzaima

17:32

@Marshall right, something like that was what i had in mind

Marshall

Probably obvious but I should point out that a void return type also needs to be supported (I'm using I and get an arbitrary number back).

dzaima

uh yeah

Adám

@Wezl Uh, *Adán*‽

dzaima

@Marshall what would a void method return? @?

Wezl

@Adám that's an á and an m

6 mins ago, by rak1507

that's an interesting m

Adám

17:35

@Wezl I saw that, but you can hardly call that m respectful of my name.

Wezl

but I forgot to hint the lowercase diacritics, so wait don't use the new release

Marshall

@dzaima Or an empty list. Maybe it would make sense to return a list by default but have a character indicating that only one result is expected and it can be returned alone.

Wezl

is privately offended at the hate thrown at m

Marshall

@Wezl I didn't know "privately" meant "in italics"...

rak1507

I wonder how a more m-like m would look

~(5 5⍴⍳25)∊6 8 11 13 16 18 21 23 seems ok to me maybe

Wezl

17:39

@Marshall I'd use subscript but I can't. the italics is for a different reason

@rak1507 I don't want a boring font ::/

rak1507

what about ' █'[~(5 5⍴⍳25)∊11 13 16 18 21 23] or something

Marshall

@dzaima I guess another useful feature is to say you want to pass in an uninitialized array of a given shape. So the argument is the shape and the result includes an array of that shape (hopefully initialized). Maybe I⥊ or I&⥊ makes sense.

Wezl

@rak1507 I considered it. I think using horizontal stroke variation is best for emphasis, e.g. BQN's blackboard bold w is ' █'[~(5 5⍴⍳25)∊1 3 6 8 16 17 18]

rak1507

how does that look in context?

Wezl

IDK my font editor breaks on characters off the BMP

but blackboard bold letters don't mix well in general ::(

Adám

17:48

@rak1507 Btw, this looks better if you add 2/ in front.

rak1507

trying to stick within the 5x5

Adám

@rak1507 Just for display purposes, as characters are about twice as high as wide.

rak1507

Ahh, fair enough

nathan rogers

@rak1507 what argument does this take?

rak1507

board as left argument, ⊂start as right

Adám

17:50

@Adám Wow, now over 150 registrations.

nathan rogers

this bleeds

what is your start board?

rak1507

@nathanrogers wdym

nathan rogers

with the board I shared, and the start position ⊂2 2 the fill bleeds into the middle space

instead of filling just the top left

1 1 1 0 1 1 1 1
1 1 1 0 1 1 1 1
1 1 0 1 1 1 1 1
0 0 1 1 1 1 1 1
1 1 1 1 1 1 0 0
1 1 1 1 1 0 0 0
1 1 1 1 0 0 0 0
1 1 1 1 0 0 0 0

rak1507

oh, ⎕IO←0

nathan rogers

yes

rak1507

17:53

start definitely ⊂2 2 not 2 2?

nathan rogers

   board{⍺{(~⍺)∧{∨/(,⍵)[1 4 5 6 8]}⌺3 3⊢⍵}⍣≡2=2@⍵⊢⍺}⊂2 2

rak1507

works for me

nathan rogers

what is your board?

@user yes

dzaima

@Marshall yeah. I think it's acceptable to infer whether to output a list of returned objects or just the return number from whether there's precisely one thing to return. (still at the "really gotta go now though" state so not writing impls just yet)

nathan rogers

 board←(8 8⍴⍳64)∊4 12 19 26 25 61 53 46 39 40

rak1507

17:54

same thing you posted, board←(8 8⍴⍳64)∊4 12 19 26 25 61 53 46 39 40

nathan rogers

yeah that's the board I'm using

rak1507

you're definitely using ⎕IO←0?

nathan rogers

oh

⊂2 2 is on the line

so of course it bleeds

no, anything in the middle bleeds too

⊂4 4 fills middle and top left

      ⎕IO
0

rak1507

oh, the board has a hole in it

nathan rogers

?

rak1507

17:57

I'm confused why the code is different

can you do ⍸board

nathan rogers

     ⍸board
┌───┬───┬───┬───┬───┬───┬───┬───┬───┬───┐
│1 4│2 4│3 3│4 1│4 2│5 7│5 8│6 6│7 5│8 5│
└───┴───┴───┴───┴───┴───┴───┴───┴───┴───┘

rak1507

that's what you'd get with ⎕IO←1

nathan rogers

yes

      ⎕IO←0
     ⍸board
┌───┬───┬───┬───┬───┬───┬───┬───┬───┬───┐
│0 3│1 3│2 2│3 0│3 1│4 6│4 7│5 5│6 4│7 4│
└───┴───┴───┴───┴───┴───┴───┴───┴───┴───┘

sorry I was trying ⎕IO←1 for a second to see if maybe it was backwards for no reason

rak1507

ok, got it, you declared the board with ⎕IO←1, I used ⎕IO←0 so the boards were different

nathan rogers

oh :P

rak1507

17:59

but you're right, my code is wrong for some reason

nathan rogers

lolo

rak1507

no, you are right but for a different reason, ⊂2 2 is on the line, but my code is also wrong because it wasn't working for a different board when it should have

ah, that's because I can't count, in this notepad document I have open I have '0 1 2 4 5 6 7 8 9', must have just forgotten about 3s existence

nathan rogers

:O

3 IS a crowd after all

rak1507

the correct code should be {⍺{(~⍺)∧{∨/(,⍵)[1 3 4 5 7]}⌺3 3⊢⍵}⍣≡2=2@⍵⊢⍺}

nathan rogers

:C

oh ok

you could just put an "on the line check" before this if it were going into production

but otherwise, pretty slick

user41805

18:02

i'm sure there's a clever scan magic solution

can't find it either yet

what's the shortest way to transform 0 1 0 2 0 0 1 0 1 to 0 1 0 2 2 2 1 0 1?

Marshall

@user41805 What's the idea here? Repeat every 2 until the next non-zero number?

user41805

something like (⌽∘⍉transform)⍣4⍣≡, ngn always does this

0 1 0 2 0 0 1 0 1 to 0 1 0 2 2 0 1 0 1 would also work (and of course with potentially different choice of numbers, or even the opposite direction)

@Marshall oh i misunderstood, i meant for the bucket fill problem

Marshall

@user41805 So you'd want 1 0 3 0 0 1 2 0 to go to 1 0 3 3 3 1 2 2?

user41805

no, i'm only concerned with this in the context of the bucket fill challenge, where i assume we only have three types of points of interest: walls, non-walls, fill

Marshall

There's something similar in the BQN compiler for propagating expression roles through parentheses. My strategy would be to get (for each value) the index of the most recent non-zero, then select with that index and merge in only the 2s from the selection result.

rak1507

18:11

{∊{⍵∨2×2∊⍵}¨w⊂⍨1=w←1,⍵} this is pretty terrible

Marshall

The tricky part is index-of-most-recent-nonzero, which is (⌈\⊢×⍳∘≢)0≠⍵.

Looks like {⎕IO←0⋄⍵⌈2(⊢×=)⍵[(⌈\⊢×⍳∘≢)0≠⍵]} works; probably can be shortened.

nathan rogers

@rak1507 pretty cool
https://tio.run/##RZDNTsJQEIX3fYqzKyQY6Q8VTEzEUNMGpKbUGEOIKVCxsWlJKQtCcOHCILHGje5cqAvZqxseZ16k3hbRm5ubzMw3d84ce@ht9Se2FwyShB6edINuH4ug@AWTYIye7cMd@EHoILp0R/Bc3wHHdQM77IORyJVRpviL4k9FztN8IWxRvJQhiBAqEBWIJSgCShJkBVIFcpHj2BTWyesW9BaKODRMqEfH1lkB1WYNQpZo6E2V35CWxkB2T7WqBUtTcWBUzRoahlFvMbKu7m7QTBfHzoXreSyeUrya5q7Zy7R9TGm@3M4VKP7OtwUwSShhpzOj@5UEiRZvrDCj@J3uXsU9cZ9FWW4140aRHUbnPSf7kxY3DOe40BmNvSj1YO1GOhL/ZKriVz@f2TkMXT9C17P9q8zGTVVTTTXdLl3MVFsnDetvcdDznG@vB3WS5Ac

rak1507

nice

nathan rogers

you got me on the rightmost part of the function though

fill←{⍺{...{...}⍣≡2=2@⍵⊢⍺}

is this f⍣≡2=2@⍵ applied to ⍺, or is this f⍣depth of 2= (2@⍵)⍺

I don't read that

rak1507

f⍣≡ is f applied until the result doesn't change

nathan rogers

18:25

right but its part of some other thing

rak1507

that is applied to 2=2@⍵⊢⍺

(which is just a hacky way of saying 'a matrix of the size of a with a 1 at ⍵')

nathan rogers

is that 2 = (2@⍵)⍺ or is it (2 = 2)@

rak1507

first one

nathan rogers

oh so its empty, but for a value at the index

which could have been anything

I get it

rak1507

something like (⍴⍺)↑⍸⍣¯1,⊆⍵ is a more literal way of implementing it

or 1@⍵≠⍨⍺

Wezl

18:28

favorite guillement poll

rak1507

is this the bqn »

Wezl

yes

rak1507

probably 3

Wezl

what? there is no 3

there are only 3 choices

Adám

@Wezl ermineii.github.io/paste/#081VQ8FXg8vVV8PXlArIUfKFsEFMBAA#S0/…

Wezl

18:32

ok that's choice number 3

nathan rogers

@Adám does ⍣≡ only work as an idiom? or if trying to implement it literally, you'd need to cache the previous result each time?

or do some kind of 2 element list each time with (current previous)

rak1507

no, look up the docs for the power operator

nathan rogers

i mean specifically power limit

right, so I don't see how you'd do it without something like

{((current_result) (⊃⍵))}⍣{≡/⍵}

because each time you'd have to check if the current matches the previous, right?

rak1507

in f⍣g, g is dyadic

nathan rogers

so my question is: is f⍣≡ an interpreter idiom to save the user from caching stuff, or??

but (f Y) g Y will never be true, because what you pass into the function is a matrix of 0's with a single 1

and that isn't your target

rak1507

18:42

oh right, I think it's an idiom

user

It wouldn't be too hard to implement, right?

nathan rogers

"Notice that power calls its dyadic right operand g with the next (f Y) and current (Y) values of the iteration as left and right arguments. In particular, if g is = or ≡, the result is sometimes termed a fixpoint of f."

oh I see

it is an idiom then

rak1507

not necessarily

nathan rogers

it says so right there

rak1507

@Adám is ⍣≡ an idiom?

says nothing about it being an idiom

nathan rogers

18:43

it says that g is passed the current Y

rak1507

yes? it is so normally too

nathan rogers

yes that's an idiom

rak1507

what

Wezl

APL: where some things are objectively idioms

nathan rogers

it isn't passed the result of f, its passed current Y

rak1507

18:44

that's just how ⍣ works

user

@nathanrogers Something like fix←{⍵(⍺⍺{res←⍵⍵ ⍺⍺ ⍵⋄res≡⍺:res⋄⍵∇res}⍺)(⍺ ⍺⍺ ⍵)}?

rak1507

run 1+∘÷⍣{⎕←⍺ ⍵ ⋄ ⍺≡⍵}1 @nathanrogers

nathan rogers

I get it already

rak1507

ok

nathan rogers

I just assumed that g always gets passed the current generation of f

This is cool
https://tio.run/##RZFPSwJBGMbv8yme2yopuatuKgQZbimaG7oRIRKrbrq07Mq6EiJ2iAizNroUXTpUh7yXFz/OfBGbWf8Ew/DOM7/3fed9Ru9a0dZAt5z2YkGfXgoqvXuOgfrvGDh9NHUbZtt2XANex@zBMm0DhDQc3W2BkQilkKL@L/V/5ESYjidilPrTBEQJYhqSDCkJWUQyjoSMeBqJGCGsC8sUChoKVcRwoFagHB1rZxFkyzmIgVAqlBVhTWp5BrJ1ms9q0PIK9tVsJYeSqharjCwqmTUavIuQtmHzKUiv41yxYLiqc2jYGQgRfvtwy/fxG5NZsCUyYQWBvo6FGvVnda75sxEhF6ZlBWX8@ZCXROiahWza7yEdT7dDEYaFayLYkEhipz6ij/M44nTyyfOp/0XvP6RdaY@dAm0@Ij1Pd73zphEUppMbhhPiGr2@5XFXl/7yvvgnCdl4JwQf1HVN20PD0u3L4GPWt3mlonC/uFUVpXpS0jZWLqdbNqovFn8

Adám

18:52

@rak1507 In what way? It certainly isn't special-cased by the interpreter in any way.

@nathanrogers I don't understand either of your two questions.

nathan rogers

moot

@rak1507 the more I look at this the more clever it is

I was approaching it via the + sign of cardinal directions in a 2d matrix 4 2⍴1 0 ¯1 0 0 1 0 ¯1

Adám

@Marshall Do you remember why we chose to group by order of base-10 magnitude for summing floats exactly? Why specifically 10⍟ and not some other base?

nathan rogers

but I was going about it recursively and couldn't figure out how to backtrack back to the beginning to check for fills in other directions

Marshall

19:20

@Adám I don't think I made that decision? I don't remember anything like that.

Wezl

CMP: What's your favorite unicode character?

Adám

@Wezl ⎕ucs 65038

@Marshall OK, well we wrote it together. I guess the base doesn't really matter.

Wezl

@Adám I can't manually add that to my font ::(

but if I add text versions of all the emojis, maybe it will make the same effect

nathan rogers

My favorite unicode

(•_•)

( •_•)>⌐■-■

(⌐■_■)

Adám

⍨

nathan rogers

19:28

⍥

(⌐■_■)–︻╦╤─

@Marshall please make this a legit expression in BQN

Wezl

wish granted

fixed, now the square is the right width

nathan rogers

⌐■_■ i is a 3 train with an atop, negate the ︻ window of ╦ pipes ╤ to ─ who even knows

make it happen

Wezl

@nathanrogers if I add box drawing characters to my font you better pay twice what you're paying now

nathan rogers

t(･_- )t

nothing drop of 0?

done @Wezl

\(•‿•)/ expand nothing joined iwth nothing compress?

¬(・。・¬)
Negate nothing bound with nothing atop negate x

aka. The Fonz

nathan rogers

19:54

@Marshall the new ⍉ @ngn the new +

(╯°□°）╯︵ ┻━┻

Wezl

I haven't immediately noticed any bugs or warts

nathan rogers

20:19

@Wezl

Adám

@nathanrogers Please try to stay on topic.

ngn

@chrispsn thanks! i was trying to remember that one..

ngn

20:47

@Marshall "almost none of them make sense" - chars mean what you (the language designer) want them to mean. wherever there is structure, there can be interpretation.

@nathanrogers "documentation" - don't conflate ngn/k (no docs) and k (for instance k4 has good docs)

Wezl

21:02

@ngn and it gets worse, much worse, when different language designers disagree

ngn

21:25

@Wezl my fav unicode char: "\0" :)

coltim

@ngn I would have figured (U+200B ZERO WIDTH SPACE)

nathan rogers

@Wezl I always liked the visual symmetry of ⍋⍒ , <> ⌈⌊ and others, but I never liked the mental asymmetry with things like ↑↓, ⌽ ,⍣,- +, ×÷, or ⊃⊂

ngn

@coltim haha, wouldn't that be a good glyph for "nothing"? :D

coltim

@ngn a good notation should elucidate what underlies. a zero width space is both present and not, depending on one's perspective

nathan rogers

21:41

What? @coltim

You told me "something" betrayed and murdered my father... But "nothing" betrayed and murdered my father... because he's still alive, and unbetrayed

rak1507

oh we're on to the nothing stuff again

'nothing' is a perfectly decent name for what it does runs away

nathan rogers

@rak1507 OH so that was SARCASM...

rak1507

no it wasn't

ngn

@rak1507 let's take it to the next level and talk abou

nathan rogers

@rak1507 i mean earlier

rak1507

21:45

@nathanrogers when?

nathan rogers

link

rak1507

@ngn ?

nathan rogers

@rak1507 htisn

ngn

@rak1507 it's backspace, also known a

nathan rogers

no, backspace is not a value

it isn't even nothing

rak1507

21:46

@nathanrogers no, the discussion about language implementation was interesting, the nothing one was just funny

nathan rogers

I think its a real discussion that needs to be had

coltim

@nathanrogers I can't claim to understand all the subtleties here, but would (‿x) better explain things?

ngn

@nathanrogers why not make it a value, a first clas

nathan rogers

no

@ngn I've got to say to you on the matter

Adám

If you want to discuss the design of BQN's "Nothing" and/or J's "Cap", please make a room on TopAnswers for it.

nathan rogers

21:47

it's called "kappa" like they KNEW

ngn

@Adám but we were discussin

nathan rogers

@Adám to see here.

Adám

It is unnecessary noise, and not on-topic here. You've been warned.

nathan rogers

discussions about array language design...

is OFF topic?

Adám

Yes.

nathan rogers

21:49

wut

pretty sure... that's like 90% of the topic

Adám

> Learn, teach, ask, code, golf, & discuss.

nathan rogers

pretty sure pedantry is the EXCLUSIVE PERVIEW of array language users

Adám

Doesn't mention "design".

ngn

@nathanrogers "apl array-manipulation j k tips" - only 3÷5 of the topic :)

nathan rogers

"discuss" array languages... right?

rak1507

21:51

there's really nothing to discuss anyway ;)

nathan rogers

@rak1507 gets it

either we discuss nothing or there's nothing to discuss

Adám

room topic changed to The APL Orchard: apl.chat ― Learn, teach, ask, code, golf, & discuss usage. See apl.wiki/APL_Orchard for access and info, aplcart.info for simple how-to questions. [apl] [array-manipulation] [j] [tips]

Wezl

@Adám no fun :(

rak1507

spot the difference

nathan rogers

right, so dzaima and marshall gotta just bounce

no bacon allowed

Wezl

21:52

I come here to discuss implementation and make ascii-art. Practically nothing else.

nathan rogers

right

coltim

maybe we have to phrase things as questions?

nathan rogers

bye-bye wezl, nice knowing you

rak1507

@Adám I feel like you can discourage all the random rubbish without completely making K off topic... do people really spam here with K anyway? Seems like a strange (and rash) decision removing it from the tags

Adám

Announcement: k is now off-topic (other than in direct comparison to APL); use the k tree. Non-APL language design is off-topic too; open a discussion on TopAnswers.

nathan rogers

21:54

lol

rak1507

I think you should reconsider

nathan rogers

right because the array language community is robust enough to fill a single channel on an obscure site

Wezl

* Wezl suggests IRC as an alternative to the web

Adám

@rak1507 K is prominent enough to have its own room. You can just use that.

nathan rogers

right... there's like 3 people that go there besides me. "Prominent"

Adám

21:56

@ngn Enjoy the increased traffic! ;-)

rak1507

@Adám I agree. I disagree on making language design off topic. I don't think you have a problem with the language design, but with the pointless long conversations. There are other ways to cut down on that, that don't mean language design is now off topic.

ngn

@rak1507 anyone can make a new room. i started "the k tree" only because adam was moving conversations about k elsewhere, to "keep it clean" - that was annoying

nathan rogers

i'm pretty sure @Adám has engaged plenty in language design discussion

Adám

@rak1507 APL language design is still on-topic.

rak1507

Making a new room be it on SE or TopAnswers is pointless and stupid when it is so obviously on topic here

nathan rogers

21:57

And even morten has been meaning to discuss language design himself

Adám

@rak1507 It isn't any more.

rak1507

@Adám BQN is a language in the APL language family

Adám

Yes, but not an APL.

rak1507

it's as good as (pretty much translates 1 to 1)

nathan rogers

"AN APL FOR YOUR FLYING SAUCER"

I'm pretty sure the comparison goes thusly
APL -> BACON -> J -> K -> LISP

ngn

21:58

@Adám i don't benefit from increased traffic

rak1507

apart from anything, topanswers isn't allowed on my school internet, whereas the apl orchard is. what am I meant to do in economics now?!

Wezl

@rak1507 making a new discussion site on www is ewe ewe ewe

nathan rogers

bqn is as close as it gets to an APL without being literally APL

ngn

who uses topanswers anyway..

Adám

Relevant: What is APL?

Transcript for