The APL Orchard

5:51 AM

Hello!

I made a one-time pad function:

⋄{⎕←⎕A[x←?⍨⍴⍵]⋄⎕A[26|(⎕A⍳⍵)+x]}'HELLO'

@Razetime Illegal code

6:03 AM

@Razetime ⋄ {⎕A[⍉x,⍪26|(⎕A⍳⍵)+x←?⍨⍴⍵]}'HELLO'

@Adám
DECBA
LJONP

@Razetime ⋄ {⎕A[↑x(26|(⎕A⍳⍵)+x←?⍨⍴⍵)]}'HELLO' (alternative using nested arrays)

@Adám
EABCD
MFNOS

6:15 AM

@Razetime Unicode version: ⋄enc←{p←?(2*16)⍨¨⍵⋄(⎕ucs p)(⎕ucs p+⎕ucs⍵)}, and then can ⋄⊢(p c)←enc 'helló'. Dec is pretty simple: ⋄dec←{⎕ucs(⎕ucs⍵)-⎕ucs⍺}. then ⋄p dec c

@Moonchild
Illegal code
VALUE ERROR
Illegal code
VALUE ERROR

d'oh. ⎕ucs, I guess?

@Moonchild No, ⎕UCS is permitted.

@DyalogAPL I don't quite see what the issue is. It works for me using DSE.

@Adám it seems to be a problem with word splitting. Dyalog interprets ⎕ucs⍵ as two separate symbols, but DSE interprets it as only one, which isn't on the valid tokens list

⋄{⎕ucs⍵}65

6:20 AM

@Adám Illegal code

⋄{⎕ucs ⍵}65

@Adám A

@Adám nice, it removes the ⎕←

@Moonchild yes, but the question only asks for encryption of letters

so ucs would make it longer

@Moonchild Yes, that appears to be it, but strangely, it does work on the new Try APL based on DSE.

@Adám odd. Maybe another unix oddity? I can replicate it locally, but don't have a windows box handy

@Razetime Fair enough. What question is that?

6:23 AM

Q: Implement a One-Time Pad

Background A one-time pad is a form of encryption that has been proven impossible to crack if used properly. Encryption is performed by taking a plaintext (comprised of only letters A-Z) and generating a random string on the same length (also only letters). This string acts as the key. Each cha...

@Moonchild No, it runs on Linux, and the language should be the same anyway.

I'll have to step through it, but first kids need to be preped for school…

1 hour later…

7:25 AM

Hey I'm back

I got the encryption and decryption functions down.

But the full porgram I tried is giving me an error.

tio.run/##SyzI0U2pTMzJT///qKO9oCj9v8ajruZHvfM0bdVd1a2qH/…

@Razetime you can't use : outside of dfns

7:42 AM

CMP: In the new array notation, should x ← 10 ⋄ c ← ( a: x ← 20 ) ⋄ x give 20 (like in JavaScript) or should it give 10, that is, the expression for a member of an object is done in a temporary scope.

It works! Try it online!

Thanks a lot @dzaima

@Razetime No need to brace inner expressions: Try it online!

thats amazing

it's at 53 bytes now

maybe you can combine some duplicated code?

7:55 AM

hmm

I tried assigning ⎕A

but that adds a byte

⎕A[⍉x,⍪26|(⎕A⍳w)+x←26?⍨⍴w←⍞] is it possible to shorten this with ⊢?

(⎕A[⍉x,⍪26|(⎕A⍳⊢)+x←26?⍨⍴⊢])⍞

@Razetime Remember that ⊢ is a 100% normal function, equivalent to {⍵}

You can certainly move ⎕A[…] outside.

ah instead of assigning it

-4

I get INDEX ERROR sometimes.

8:05 AM

probably missing ⎕IO←0

Try it online!

this should be fine

This was fun, I'll post it now

@Razetime Wait, I think I have an idea.

@Adám sure go on

@Razetime Move out 26| too.

Try it online! yep it works

never thought of doing it after tabling.

@Razetime the "encrypt" returns key and enrypted message, is that it?

8:14 AM

yes

and decrypt takes message and key and returns the original

cool

2 hours later…

10:23 AM

@xpqz is it you that usually uses RIDE?

@RGS that would be me, yes

Finally managed to complete AoC 2018 (now 2015-2018 completed): nbviewer.jupyter.org/github/xpqz/AoCDyalog/blob/master/…

Some hard nuts to crack there.

@xpqz let me ask you a question: do you usually save functions in files?

I was taking RIDE for a spin and editing functions and fixing them seems similar to the windows terp. but then saving a dfn/tradfn to a file... I don't think there's a menu item for that :P

rak1507

bin←{(32⍴2)⊤⍵} you would be probably be better off using ⍣¯1 here

@RGS I've used ]Link to 'save' stuff to files. But the experience is somewhat lacking if used to the model presented by other editors.

@xpqz alright, I see

@xpqz +← 1

10:33 AM

Most of the time I write APL in a jupyter notebook. If I need to use the debugger, I fire up RIDE and cut & paste what I need into it.

I see, thanks

I'd love to see a VS Code plugin for Dyalog that simply allowed me to save my work in a single file and allow me to run the debugger through its interface,

@xpqz maybe the plugin mentioned here might evolve into that?

I guess that may eventually be possible as Dyalog's moving closer to allowing files to be run as scripts.

btw @JamesHeslip do you have any idea about whether that plugin supports 18.0 out of the box? Assuming I have 18.0 on my machine?

10:39 AM

I tried it ages ago, but couldn't figure how to run my code on a Mac due to its strange installation.

@JamesHeslip did you guys build it?

1 hour later…

James Heslip

11:45 AM

@RGS I've not tried it with 18.0. I last used it when it was being developed (a long time ago...) It was worked on primarily by Gil Athoraya (github.com/OptimaSystems/vscode-apl-language-client) but I don't think it's being actively worked on.

hm ok, just wondering :)

12:42 PM

@xpqz That should be simple from 19.0 when we have proper script support.

Yeah. I can see that's the direction of travel which makes me happy.

2 hours later…

2:54 PM

<jibanes> o

3:34 PM

jibanes: Uh, hi?

1 hour later…

TessellatingHeckler

4:37 PM

In the JavaScript BQN old interpreter (tenuous link to this channel), what is v in this line? github.com/mlochbaum/BQN/blob/…

In the JS console, get1([[10,20,30], 1]) returns 20, which matches the i[0][i[1]]; part of the linked line. Where does the anonymous function involving v come into it

@TessellatingHeckler (v=>(assert(v!==null),v))(…) is an immediately called lambda; All it does is check that the … isn't null, and in that case, returns it

TessellatingHeckler

@dzaima ohhh, so simple!

thanks

TessellatingHeckler

5:22 PM

PS C:\> run (&$arr 0,0,25 3) (&$arr @(1) 1) @(@(&$arr 0,1,0,0  4))
1

that's the first bit of BQN compiled bytecode which has gone through it and produced the right answer!

\o/

5:39 PM

@Razetime are you not allowed to inline assignments in Dyalog Extended? I tried doing it with one of your answers to the ellipse circumference

@RGS Extended messes with it - 4-h← is expanded to 4 Minus h← which then breaks for semi-obvious reasons

2 hours later…