The APL Orchard

Adám

12:10 AM

@rak1507 Sorry, I misplaced their SE chat id. Can you have them join the room again, and then click their avatar, and then give me the link to "user profile" (not "user profile on…")?

rak1507

stackoverflow.com/users/13748303/nebb ?

Adám

@rak1507 Yeah, that's "the user profile on …". We need their chat "user profile".

rak1507

Ah

Adám

@rak1507 Never mind. Found it.

@ThomasWard chat.stackexchange.com/users/491670/user13748303

rak1507

Thanks

Thomas Ward

12:13 AM

done

salts @Adám in the interim, then goes to consume foods before spinning his RPi 400 up for the first time

Adám

I deserve no less.

Today I found a method of getting the chat user id, only given the "user12345678" name: Go here and filter by "user12345678". Voilà.

rak1507

Ah nice, I was trying to find it from that

Thomas Ward

@Adám fortunately pouring Salt on people is not technically against any rules :P

just me being weird :)

(but I'm an IT Security guy, so I mean...)

Bubbler

security is salty

Adám

@ThomasWard Quite the opposite. I'm the guy in charge of SALT at Dyalog.

Thomas Ward

12:20 AM

hah lol

if you want I could replace it with the proverbial firebombs of bugs but i'm not that evil right now :P

yawns and consumes all the coca-cola within a mile

Adám

@ThomasWard I'm not joking: Look at the change log. And this one.

Thomas Ward

i believe you :)

sips coffee but doesn't share it, because hoarding

at least you aren't working with SolarWinds on Orion xD

Adám

@ThomasWard No, because COVID!

Thomas Ward

points at the coffee maker

... they're single cups. :P

but they're all mine 'cause hoards

no but still at least it's not SolarWinds Orion xD

or you'd still be backlashed from the chaos of that recently :)

Adám

Ouch.

I did plug a security hole recently, though at worst, people could have trashed a Docker container, taking down a server that isn't essential.

Razetime

2:58 AM

⋄LA4 ⎕FMT ↑'APL' 'BQN'

Dyalog APL

@Razetime VALUE ERROR

Razetime

⋄'LA4' ⎕FMT ↑'APL' 'BQN'

Dyalog APL

@Razetime FORMAT ERROR

Razetime

why's this giving me a format error?

Josh D

5:26 AM

I never really use ⎕FMT but it looks like 'LA4' isn't a valid combo in the format specification table here (https://help.dyalog.com/18.0/#Language/System%20Functions/Format%20Dyadic.htm)

Looks like you're doing a 'rAw' Alphanumeric format. 'L' is a qualifier which is abbreviated as 'q' in the specification.
And the Alphanumeric format seems to not accept qualifiers.

⋄'A4' ⎕FMT ↑'APL' 'BQN'

Dyalog APL

@Josh D
┌→───────────┐
↓   A   P   L│
│   B   Q   N│
└────────────┘

Josh D

is valid

I guess you could do a simple workaround like

⋄ 4{(⍺-1)⌽('A',⍕⍺) ⎕FMT ⍵}↑'APL' 'BQN'

Dyalog APL

@Josh D
┌→───────────┐
↓A   P   L   │
│B   Q   N   │
└────────────┘

Josh D

To left justify with ⍺ spaces. That's my guess

Razetime

5:44 AM

It's funny, because it says to use the L specifier just about the A example

Thanks for the help!

kimmolinna

@ngn I'm building interface between Dyalog and Q and It's mainly so that someone has to be able to import 64-bit integers from kdb+ to Dyalog. And because IPC use serialized and typed data I'm able to select correct FR before importing data.

ngn

@kimmolinna ah, i thought it was in the opposite direction

@kimmolinna is dyalog faster or any better at analyzing data than kdb+?

Josh D

6:05 AM

@Razetime Hmm yeah that is a strange example. Either the example is misleading and needs to be fixed, or I'm wrong and there is some way to use the left justify qualifier with that. Maybe Adam can confirm.

Razetime

6:59 AM

Day 24 AOC answers

Bubbler

@Razetime You might have been better off using ⎕S and fetching the offset inside right operand {...}

Also, if you skew the hexagon in a different way, you can get away with chopping first and last elements instead of doing [1 2 4 5 6 8 9]

⍵[⊂2 2] → ⍵[2;2]

You don't need parens for ({...}⌺3 3)⍣100

Razetime

8:45 AM

@Bubbler mor eof a yle choice than anything

@Bubbler I should've known this

@Bubbler not sure how that's work

@Bubbler ah the other way

then 1↓¯1↓

ngn

9:17 AM

@Bubbler do you publish your solutions?

Bubbler

Not solving in APL this year, so no

maybe I'll return to it later, since I have a place to publish my APL solutions

ngn

9:57 AM

my day24: k, apl

Adám

@ngn Do you have a collection of your APL solutions?

ngn

10:22 AM

@Adám same place

i couldn't translate all into apl

nebb

@Adám Thanks for letting me in! It took a while but I've got the RIDE up and running, looking forward to getting better at APL

Adám

@user13748303 Sure. Sorry for the delay. Is there anything I can be of help with?

ngn

@Adám also bear in mind there are a lot of k-isms like flattened 2+ dimensional arrays, and workarounds for dicts like k v←↓⍉{⍺⍵}⌸..

Adám

@ngn What about previous years?

Kamila Szewczyk

One byte saved:

      (#+.⍎⍕)#×.⍎⍕ 1234
6
      {+/⍎¨⍕×/⍎¨⍕⍵} 1234
6

ngn

10:26 AM

@Adám i have some of those but only in k

nebb

@Adám No, but thanks for asking

Adám

@ngn OK. I've added yours to APL Wiki's listing.

@KamilaSzewczyk OK, you're basically there. Want another hint?

ngn

@Adám can i delete them after 25 Dec?

Kamila Szewczyk

i thought that bind will help here, because i've seen it in ngn's hint

i tried randomly placing it in places that made sense after enabling boxing

Adám

@ngn You're free to take down your solutions, but I'd appreciate a heads-up so I can remove from the wiki too.

@KamilaSzewczyk OK, hint: Think of f∘g as f with its right argument preprocessed by g.

Kamila Szewczyk

10:29 AM

i don't really understand how stuff binds together, but I know what bind does

#+.⍎∘⍕#×.⍎⍕ this works

Adám

That was my 11er.

Kamila Szewczyk

@user suggested a golf to my answer as a whole, though, so I don't think i'll use this golfed snippet

{⍎¨⍕×/⍎¨⍕+/⍵}⍣≡⍳

ngn

@Adám i'll remove them in january. btw, these are not all my solutions. some are modifications of rak1507's, xpqz's, razetime's, and maybe others.

Adám

@ngn Why are you removing them, if I may ask?

ngn

@Adám if i can't make them perfect, i'd rather not have them hang on the internet forever. also, of course i'd like to encourage people to learn k6, not so much apl.

Adám

10:35 AM

@ngn Uh, once published on the internet, things tend to get archived and thus hang on forever.

ngn

ok, but not in my little perfect garden :)

Razetime

How do I get non-overlapping non-square windows of a matrix?

Adám

@Razetime ⌺ with a matrix right operand.

Razetime

yeah but what do I make the right operand as

for 2x2 It's just 2 2⍴2

what do I do for 16 4?

Adám

2 2⍴16 4 :-)

Just watch out for shards.

Razetime

10:38 AM

My matrix is 16x16

Adám

Oh, so you just want to cut it into groups of 4 columns?

Razetime

yes

Adám

Then (16⍴4↑1)⊂ is easier.

ngn

@Adám i forgot to mention: another reason the link can go stale is if sr.ht becomes paid, which is a real possibility

Adám

Wait, paid to read or paid write?

Or do you mean that if it becomes paid, you'll leave?

Kamila Szewczyk

10:45 AM

@ngn why would sourcehut become paid

ngn

@KamilaSzewczyk it says so when you register, and also on the billing tab

it doesn't say when

"sr.ht is currently in alpha, and the quality of the service may reflect that. As such, payment is currently optional, and only encouraged for users who want to support the ongoing development of the site."

Adám

@Razetime Should I add "cut matrix into N columns" or "cut matrix into columns of width N" to APLcart?

ngn

@Adám paid to write. or more precisely: paid to own a repository (or other similar thing)

Razetime

@Adám sure, it's a good idea

I never inferred that matrices could be cut using the same bitmask

Adám

@ngn OK, but then your existing solutions could stay.

@Razetime Inferred from where?

ngn

10:54 AM

@Adám i'm not sure, it doesn't say that

Razetime

from the dropdown that comes up when you hover on ⊂

Adám

@Razetime cough, array language, cough

ngn

@Adám in any case, sr.ht is so lightweight and pleasant to use, i'm planning to move all of my projects there, at least until the owner starts demanding $

Razetime

@Adám cough, slow learner, cough

Kamila Szewczyk

@ngn you can probably host it yourself

just like gitlab or gitea or whatever

Adám

10:58 AM

That said, I'm impatiently waiting for dyadic ∪ and ∩ and ~ to become proper array functions.

Razetime

@Adám how would those function with matrices?

Adám

@Razetime ∪, ∩, ~

ngn

@KamilaSzewczyk maybe. i never learned how to do these things.

Razetime

⋄(8⍴4↑1)⊂8 8⍴⍳8

Dyalog APL

@Razetime
┌→────────────────────┐
│ ┌→──────┐ ┌→──────┐ │
│ ↓1 2 3 4│ ↓5 6 7 8│ │
│ │1 2 3 4│ │5 6 7 8│ │
│ │1 2 3 4│ │5 6 7 8│ │
│ │1 2 3 4│ │5 6 7 8│ │
│ │1 2 3 4│ │5 6 7 8│ │
│ │1 2 3 4│ │5 6 7 8│ │
│ │1 2 3 4│ │5 6 7 8│ │
│ │1 2 3 4│ │5 6 7 8│ │
│ └~──────┘ └~──────┘ │
└∊────────────────────┘

Razetime

11:07 AM

(16⍴4↑1)⊂[1]⊂8 8⍴⍳8

⋄(16⍴4↑1)⊂[1]⊂8 8⍴⍳8

Dyalog APL

@Razetime RANK ERROR

kimmolinna

@ngn I don’t think so because both are APLs but I have used Dyalog APL much longer time. Of course an interface has to able for two way operations. And due to layoffs I have now time to do interface. 😉

Razetime

⋄(16⍴4↑1)⊂[1]8 8⍴⍳8

Dyalog APL

@Razetime RANK ERROR

Adám

@Razetime What are you trying to do? You can't partition a scalar.

Razetime

11:11 AM

how do I get groups of rows

⋄(16⍴4↑1)⊂[1]8 8⍴⍳8

Dyalog APL

@Razetime LENGTH ERROR

Razetime

owait

Adám

@Razetime You have 8 rows, but are specifying 16!

Razetime

⋄(16⍴4↑1)⊂[1]8 8⍴⍳8

Dyalog APL

@Razetime LENGTH ERROR

Razetime

11:13 AM

⋄(8⍴4↑1)⊂[1]8 8⍴⍳8

Dyalog APL

@Razetime
┌→────────────────────────────────────┐
│ ┌→──────────────┐ ┌→──────────────┐ │
│ ↓1 2 3 4 5 6 7 8│ ↓1 2 3 4 5 6 7 8│ │
│ │1 2 3 4 5 6 7 8│ │1 2 3 4 5 6 7 8│ │
│ │1 2 3 4 5 6 7 8│ │1 2 3 4 5 6 7 8│ │
│ │1 2 3 4 5 6 7 8│ │1 2 3 4 5 6 7 8│ │
│ └~──────────────┘ └~──────────────┘ │
└∊────────────────────────────────────┘

Razetime

there

Razetime

11:25 AM

another problem

⋄(⊖@1 3)3 3⍴⍳9

Dyalog APL

@Razetime
┌→────┐
↓7 8 9│
│4 5 6│
│1 2 3│
└~────┘

Razetime

⋄(⌽@1 3)3 3⍴⍳9

Adám

@Razetime What?

Dyalog APL

@Razetime
┌→────┐
↓3 2 1│
│4 5 6│
│9 8 7│
└~────┘

Razetime

in a vector like (⌽@1 3)(2 3)(3 4)(4 5)

⋄ (⌽@1 3)(2 3)(3 4)(4 5)

Dyalog APL

11:28 AM

@Razetime
┌→──────────────────┐
│ ┌→──┐ ┌→──┐ ┌→──┐ │
│ │4 5│ │3 4│ │2 3│ │
│ └~──┘ └~──┘ └~──┘ │
└∊──────────────────┘

Razetime

I need it to give me 3 2 3 4 5 4

ngn

@Razetime try ⌽¨ instead of ⌽

Adám

@ngn Can it not be done with a clever scan?

⋄ 1↓⌽⍤⊢\0,(2 3)(3 4)(4 5)

Dyalog APL

@Adám
┌→──────────────────┐
│ ┌→──┐ ┌→──┐ ┌→──┐ │
│ │3 2│ │3 4│ │5 4│ │
│ └~──┘ └~──┘ └~──┘ │
└∊──────────────────┘

Adám

:-D Look at me, I'm writing APL like ngn!

Razetime

11:31 AM

I sense shade

ngn

lol

@Adám could also be ⌽¨⌽⍤⊢\ without the 0,

Adám

@ngn Yes, and that'd be a proper function.

kimmolinna

@ngn I’d like to use kdb+ more but I haven’t got enough time and a proper project to learn it properly.

ngn

@kimmolinna there's a newer one now: shakti, aka k9 (by arthur, but not kx). it's work in progress.

@Adám @Razetime shorter: ⌽/,⍨⌸ (with ⎕io←1)

Razetime

wow, all three of these look like war crimes

Romilly Cocking

11:46 AM

@ngn Is there an ETA for k9? It looked very interesting when Arthur talked about it at Iverson college a few years ago.

Adám

@ngn But that's only for these specific values.

@RomillyCocking It is already available, but is still changing.

ngn

@RomillyCocking ETA - not that i know of

Razetime

@ngn It stumps me how you found out that ⌸ was perfect for this oddly specific task

Adám

Another fun one: ⋄ (⊃⌽⌽)¨(2 3)(3 4)(4 5)

Dyalog APL

 @Adám
┌→──────────────────┐
│ ┌→──┐ ┌→──┐ ┌→──┐ │
│ │3 2│ │3 4│ │5 4│ │
│ └~──┘ └~──┘ └~──┘ │
└∊──────────────────┘

Romilly Cocking

11:48 AM

@Adám I will seek it out. ISTR that it was expected to run on resource-constrained hardware, which would be interesting for robotics.

Adám

@RomillyCocking shakti.sh

@Razetime Quite obvious actually. There's only one of each, so the indices they occur at are 1 2 3. Only the first and last need reversal, but since they are 2-element vectors, an odd rotation amount is the same as a reversal. 1 and 3 are odd, 2 is even. Voilà.

Romilly Cocking

@Adám Thanks. The executable is pleasantly small, and the documentation, while fine, doesn't take up much space either. It looks as if arthur is not open sourcing it atm, which he had planned to do :(

ngn

@RomillyCocking there's a shakti google group

Romilly Cocking

@ngn Thanks. I;ll take a look

Razetime

nice. there's a k version named power

ngn

11:54 AM

@Razetime shakti means power?

Adám

@RomillyCocking And a k chat room:

the k tree

app.element.io/#/room/#ngnk:matrix.org

Razetime

@ngn yes it does

shakti means strength; power, so on

Adám

Shakti

Shakti (Devanagari: शक्ति, IAST: Śakti; lit. "Energy, ability, strength, effort, power, capability") is the primordial cosmic energy and represents the dynamic forces that are thought to move through the entire universe in Hinduism, and especially Shaktism, a major tradition of Hinduism. Shakti is the personification of the Energy that is creative, sustaining, as well as destructive, sometimes referred to as auspicious source energy. As the Shakti or Creatrix, She is known as "Adi Shakti" or "Adi Para Shakti" (i.e., Primordial Inconceivable Energy). On every plane of creation, Energy manifests...

ngn

@Razetime i see, some kind of "divine" power. you probably wouldn't say "the square of x is x to the shakti of 2" or "could you unplug the shakti cable of my laptop, please"

Razetime

@ngn you could, and people would stare at you in a concerned manner

ngn

11:57 AM

@Adám yeah, i knew about the goddess, of course

Romilly Cocking

@ngn or even "shakti to the people"

ngn

haha

Adám

The logo of Shakti Software appears to be a Sri Yantra, which is a combo of Shiva and Shakti.

ngn

absolute shakti corrupts absolutely

Romilly Cocking

lol

Adám

12:00 PM

IBM ShaktiPC

kimmolinna

@ngn Thanks for info. I will look at that.

ngn

@Razetime what language is this, btw? sanskrit?

Adám

Probably common to Sanskrit/Hindi/Urdu/etc.

Razetime

@ngn common to many, many languages

@dzaima how do I convert the output of APLImg.mat to an rgb format?

it gives a lot of unsigned ints

dzaima

@Razetime as in, outside of APLP5? Each unsigned int is a single pixel specified in ARGB

Razetime

12:13 PM

in APLP5

also how doe P5.Img work?

does it need an APLImg object?

dzaima

@Razetime so you want a separate matrix per red/green/blue channel?

Razetime

either that or a matrix of rgb triplets

@dzaima First I wanna know how to display an APLImg

dzaima

@Razetime P5.G.img imageObject

Razetime

now allowance for coordinates?

dzaima

and P5.img colorMatrix to create an image object, or alternatively P5.img "filenameToLoad" to get an image object for a file

@Razetime the left arg of P5.G.Img can be x y specifying where to draw, x y sc to scale the image by sc, or x1 y1 x2 y2 to draw in the bounds of those

Razetime

12:19 PM

oh cool

great

now I can view pictures of cats through dzaima/APL

chrispsn

12:40 PM

@RomillyCocking have you seen b? a system language similar to k, released before k7 (the iteration before k9) kparc.com/b

If you're interested in reading Arthur's implementation code (albeit not for k), that's probably the closest that exists in public

Adám

@chrispsn Isn't B just a C-like implementation language for K?

chrispsn

@Adám syntax of k with semantics of c, might be a good way of putting it

Kamila Szewczyk

1:01 PM

@chrispsn B is an useless piece of bugged C code

from non-whitney standpoint

chrispsn

Sure, I only mention it as a potentially interesting example of relatively recent implementation code

Kamila Szewczyk

it doesn't even work

and the claims are most probably plain false

Adám

1:36 PM

How about a totally pixel design?

Romilly Cocking

@chrispsn I've seen Arthur's implementation code for the first prototype of J. Really useful if you want to shut C programmes up :)

Adám

Nah, I like the fancier logo better.

nebb

2:02 PM

https://github.com/n3bb/dyalog-unborked/
hacked together an install script (only for ubuntu atm though)

Adám

@nebb What is this supposed to do?

Romilly Cocking

@nebb what do you think is borked in dyalog at the moment? I've not had a problem installing it on Linux Mint for Intel or on the Raspberry Pi?

Adám

@RomillyCocking On thing it appears to fix is our stealing the Super key.

Romilly Cocking

@Adám OK - thnks

Adám

It streamlines installation of both Dyalog and RIDE (like we do on macOS).

Romilly Cocking

2:15 PM

@Adám Just realised - I have always had a problem with APL keys in RIDE on Intel Linux, but the problem has been around for so long I have got used to it. If this fixes the problem I will do a slomo dance of celebration.

Adám

@nebb You can extract the latest versions of Dyalog from the HTML here.

@RomillyCocking Video or it didn't happen.

Romilly Cocking

@Adám Be careful what you ask for. Some things cannot be unseen :)

Adám

@nebb Using APL (I'm sure you can translate to bash+curl+grep or something):

      ]load HttpCommand
#.HttpCommand
      r←HttpCommand.Get'https://www.dyalog.com/download-zone.htm?p=download'
      ↑2↑'"([^"]*(deb|rpm))"'⎕S'\1'⊢r.Data
/uploads/php/download.dyalog.com/download.php?file=18.0/linux_64_18.0.39712_unicode.x86_64.deb
/uploads/php/download.dyalog.com/download.php?file=18.0/linux_64_18.0.39712_unicode.x86_64.rpm

And for RIDE:

      ]load HttpCommand
#.HttpCommand
      r←HttpCommand.Get'https://www.dyalog.com/download-zone.htm?p=download'
      ↑'"([^"]*64\.(deb|rpm))"'⎕S'\1'⊢r.Data
/Dyalog/ride/releases/download/v4.3.3463/ride-4.3.3463-1.x86_64.rpm
/Dyalog/ride/releases/download/v4.3.3463/ride-4.3.3463-1_amd64.deb

nebb

2:45 PM

see what adam said @RomillyCocking
although there are obvious problems with the script which i will fix at some point, as Adam pointed out, getting the latest version of the .deb/rpm files, and also the lack of error checking, etc

Adám

@nebb Very worthy endeavour, though. Once you get those things sorted, we should add it to APL Wiki.

nebb

alright, I'll do so and post when i'm done - probably sometime between boxing day and new years'

Adám

Cool.

user

5:54 PM

@Adám Do you know why ⍭ uses negative arguments for getting primes less than ⍵, getting the ⍵-th prime, and getting the previous prime? For context, Razetime suggested using it instead of pco on an answer, but it turned out the same number of bytes because I had to use ¯. Would it be possible to also be able to use positive arguments for those? Razetime did say in a later comment that that would mess up older answers, but what about single digit numbers like 3 or 7 that haven't been used yet?

user

6:46 PM

In Extended, I mean

Razetime

Thill pribably save a byte in many answers

Adám

7:02 PM

@user Extended was actually never intended as a golfing language. I changed the arguments of ⍭ because I thought I could make it make more sense than pco. And even if I change it, TIO is frozen.

rak1507

9:34 PM

⋄ ⎕FR←1287 ⋄ ⎕PP←34 ⋄ *○.5*⍨163 ⋄ ⌊*○.5*⍨163

Dyalog APL

@rak1507
2.625374126E17
2.625374126E17

rak1507

oh well that doesn't work but for some reason it ceils? is it because it's so close to an integer?

Adám

9:47 PM

@rak1507 I don't understand what the problem is here. The √163≈13. 13π≈40. e⁴⁰≈2.7⁴⁰ and is in the order of 10¹⁷.

rak1507

Flooring it does the same thing as ceiling it

Adám

Ah, that's what you're asking. Yes, because it is tolerantly integer:

      ⎕FR←1287 ⋄ ⎕PP←34 ⋄ *○.5*⍨163 ⋄ ⌈*○.5*⍨163 ⋄ ⌊*○.5*⍨163
262537412640768743.9999999999992474
262537412640768744
262537412640768744
      ⎕DCT←0 ⋄ *○.5*⍨163 ⋄ ⌈*○.5*⍨163 ⋄ ⌊*○.5*⍨163
262537412640768743.9999999999992474
262537412640768744
262537412640768743

rak1507

Ah, right

Thanks

user

10:07 PM

@Adám Oh, I see. May I ask what you made Extended for? Was it just a pet project or for a practical use?

Adám

@user It was as a sandbox to try out various ideas I had for extending Dyalog APL.

user

Oh cool

Adám

10:20 PM

@user If you look at the APL Wiki article about me, you'll see that three operators pioneered in Extended, were eventually adopted in Dyalog proper.

user

11:16 PM

@Adám Nice! And thank you for ⍤!

the k tree

Transcript for