is Dyalog implicitly multi-threading that? It's using more than one CPU core

but not much more, 30%-31% of a 4-core chip

@TessellatingHeckler there's no implicit multi-threading afaik

is there anyway to interrupt a ton of printing when ctrl+break doesn't work and strong interrupt from the taskbar notification area doesn't work?

@TessellatingHeckler ^question for @Adám

@TessellatingHeckler I sometimes have success with Threads→Show Threads and double-clicking on the offending (usually only) thread. If the result has already been computed, and is just printing, you can also try to make the windows small, as that speeds up printing.

@Adám Cheers; Threads and double-clicking didn't do much, threads -> show stack, clicking on stack, fatal error'd Dyalog so non-issue now

@TessellatingHeckler If you can reproduce that, then we'd be grateful for your report.

@Adám it's repeatable on this computer, let me try another in case it's my hardware

yep

if not exactly the same crash, a reproducible one; email sent to support@ to report it.

@TessellatingHeckler Check 1111⌶⍬

@Adám It says 4

@TessellatingHeckler Then it'll implicitly use up to 4 (virtual) processor threads.

@Adám ok

@ngn I'm making a netpbm image for a chess board and so far I've gotten this:

'P3'
2⍴8
1
8 8⍴ (3⍴0)(3⍴1)

But the thing is i need to shift every other row

how do i do that?

@Razetime ⌽ can take a vector of amounts to rotate with, one per row.

oh perfect

tio.run/##SyzI0U2pTMzJT///qG9qZv6jtgkG////…

whats with ppm files being so blurry?

Probably just being scaled.

@Razetime (3⍴0)(3⍴1) is simply 3⍴¨⍳2

@Razetime A mathematical approach: 3⍴¨2|∘.+⍨⍳8

@Adám with IO←0 right

Yes.

Squares should have 50px * 50px except for vectorial images eeeeee

I don't thin ppm is a vector lol

need to make an 8x8 board of 150 x 150 matrices of 0's and 1

@Razetime You can just scale it with 50/50⌿

'@Adám my array is getting too big to copy paste directly. How do i send it to a file?

oh 50/50⌿ makes it a lot simpler

@Adám ok so theres an aplcart snippet for writing to a file, but it says

:55461115 RANK ERROR: Left argument should be content, optional encoding and optional line
 ending
      (50/50⌿((⍳8)⌽8 8⍴3⍴¨⍳2))⎕NPUT'chessboard.txt'

just realized I could use P1 instead and remove the rest of the stuff

@Razetime Native files contain text (simple vector or vector of vectors). You've got a matrix of numeric vectors! Try to ,⍕ it.

@Razetime you want to ,⍕ the left argument of ⎕nput, not it's result (why would formatting its result help its execution anyways?)

oh, gotcha

My googling has failed me when was dyalog v 18 released?

nvm found it

@rak1507 In general, try APL Wiki for info about APL.

@user94387 Hi there jimfan, welcome to the APL Orchard!

in BQN, is there an equivalent of roll or deal? ('?' in APL)

looks like its destined for a library, psng next i guess

@cannadayr dzaima/BQN has •RAND, but there's nothing in the spec yet.

I think a library function makes a lot more sense than a primitive. In particular, you can have a function that takes a random seed and returns a closure that generates a new value whenever it's called.

To get a seed, you'd have some other library that interfaces with the OS.

@Marshall thanks, do you have any familiarity w/ PRNG algorithms? i was gonna pick a simple "good enough" and use it for now

ill upload my library today, im starting on porting my math fns over

Do any programming languages come with a true RNG built in?

youve gotta talk to the HW for a true RNG

so via the kernel

@cannadayr Yes; I think xoshiro256** is a pretty good choice.

@Adám Yeah, easy to specify and not so easy to implement. Not to mention impossible to verify an implementation.

@cannadayr Sure, but modern computers tend to have plenty of true randomness sources available, like noise on the serial port, microphone, and camera.

@Adám So glad I finally have enough reputation to chat. Thanks for your greetings.

@Adám /dev/random is as truly random as the kernel can make it

cmc: in an n×n×n cube, vertices of the same parity form a regular tetrahedron. generate a boolean cube marking its surface, e.g. for n=5:

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

(^using ⊂[⎕io+1 2]result for better visualization)

@Marshall btw uploaded my very inefficient, likely incorrect implementation of xoshiro256ss. is garbage, but also is start.

github.com/cannadayr/bqn-ws/blob/master/src/math.bqn#L54

@ngn Arithmetic, geometric.

@Marshall nice! it works. now i should try to understand how.

Much better geometric.

@cannadayr Those ^= lines depend on each other, so you need to do at least the first pair and the second pair separately. I could have missed something but it looks like you're doing them all in parallel.

@Marshall does "table" mean outer product?

@ngn Yeah.

ok, makes sense

To be honest I'm not entirely sure how this one works

@Marshall but it's really good :)

when translated to apl, it's shorter than my own solution

@ngn I don't think it works with APL monadic transpose though. Make sure to test it.

@Marshall i used ⍉⍤2

Try it online!

@ngn Okay that is yet another solution where I don't know why it works.

No, it makes sense now. But explaining why is beyond me.

let's leave it as an exercise to the reader :)

@ngn What was your original solution then?

@Marshall i'm ashamed to post it now :)

here, this is my punishment: {∨/⍵=1+(↑⍳3⍴⍵)+.ׯ1*=/↑⍳3 4}

(⎕io←0 as usual)

@ngn The extended identity matrix is pretty cool though.

thanks

@Marshall in bqn did you give up on the "uninterrupted circle" principle for choosing "modifier" symbols?

@ngn I don't think I did.

.. or maybe i'm talking nonsense - in the language bar there are many yellow circular things, but some among them are not circular

no! my bad. the others are purple