There is reversed(list) if you want a copy

>>> l=[1,2,3,4,5,6]
>>> for i in reversed(l):
...     print(i)
...
6
5
4
3
2
1

@Adám @Richard suggested APL Quest, but I think others here should decide and it doesn't matter if the name changes

Agreed. And it is time!

:)

@xpqz huh why? i mean is it slow?

Today's quest is Write a dfn to produce a vector of the first n odd numbers.

Does anyone have a solution they'd like to present?

@Adám I do

Go ahead. (We'll discuss alternatives afterwards.)

yes, you first

For ⎕IO←0:

{1+2×⍳⍵}

(Aha!)

Yeah, that's a very obvious solution.

Anyone else?

{1-⍨2×⍳⍵}

That's for ⎕IO←1. Nice.

A more fun one: (⍸1⌽(2↑1)⍴⍨2×⊢)

@xpqz why?

That's ⎕IO agnostic.

{⍸2|⍳2×⍳⍵}

Whoa.

@PyGamer0 (later)

:O

I tried to get silly with it, but in some sense it's "where odd numbers"

@RikedyP Are you sure you want both ⍳s?

@Adám oh yeah

I won't edit

⋄{⍸2|⍳2×⍵}10

Don't put spaces immediately inside the backticks.

@RikedyP 1 3 5 7 9 11 13 15 17 19

@Adám ty

That solution is (non-obviously) ⎕IO agnostic.

@RikedyP nice one

But probably quite wasteful.

@FawnLocke this idea is fun: ⍸1 0⍴⍨2×10

Quite fast too, I imagine

How about {+\2-10↑1}

@Adám +←1

very cool

Totally

hmm

{(2|⍳+⍨⍵)/⍳+⍨⍵}

@Adám trying to understand it

{⍸2|⍳2×⍵}t   → 1.6E¯5 |   0% ⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕
(⍸1 0⍴⍨2×⊢)t → 5.0E¯6 | -69% ⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕
{+\2-⍵↑1}t   → 6.5E¯6 | -60% ⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕

@Richard Let me explain…

@Richard I like to either build up pieces from the right, or take away pieces from the left, until I get it

10↑1 should really be ⍵↑1 which generates a one followed by n-1 zeros.

2- gives us a one followed by a bunch of twos.

Then a cumulative sum gives us odd numbers. ⎕IO isn't involved.

ah ,ok! :)

@BrianBED No need to save and restore ⎕IO as the assignment localises it.

@BrianBED By the way, setting ⎕IO inside a dfn will localise it within that dfn, so ⎕IO in the session will not be affected - perhaps that was the joke, but just so you know

OK, who's up for using the 2×⍳⍵ method but adjusting for ⎕IO?

ill use BQN, no IO worries :P

@Adám oh huh... so even {⎕io←0⋄1+2×⍳⍵} works?

cool

{(¯1*⎕IO)+2×⍳⍵} is what I had

You mean * not ⍣

That's what I had too.

Yes ...

Press UpArrow to edit.

Nice

@Adám {~⎕IO:1+2×⍳⍵⋄○'ouch'}10

⍳+⍳-≢ for a tacit

@Wezl-yizl What?

is changing io inside a function ever useful in practice? i can imagine it might simplify some functions?? :o

@rak1507 Wow, that is cool.

@rak1507 nice

@Wezl-yizl Haha.

@BrianBED Yes. Generally people use a global ⎕IO and flip it locally when neater.

ah cool. i'll keep a look out for when that could be useful

@rak1507 very nice

Anything more we need to say about answering the core quest?

ok i think I got the best one

@rak1507 ditto

∇ArrayResultListVector ← GenerateDyalogAPLOddNumberFactory EndingNumberInteger
    ArrayResultListVector ← ⍬
    :For IterationValueIntegerCandidate :In ⍳ 2 × EndingNumberInteger
    	:If (2 | IterationValueIntegerCandidate) = 1
    		ArrayResultListVector ← ArrayResultListVector , IterationValueIntegerCandidate
        :EndIf
    :EndFor
∇

Hahhhhahah

Should I kick-mute him?

{⍺←1 ⋄ ⎕←¯1+2×⍺ ⋄ ⍺<⍵:(1+⍺)∇⍵}10

how can you dynamically do ]load foo?

@Razetime this is better

@Razetime has written too much java lately

@rak1507 ⎕SE.SALT.Load or just ⎕FIX

thanks

@Razetime But you're using ⍳ instead of an iterator!

@Adám is ⎕SE.Link.Import also ok? I feel like might be nice to push for one way moving forward

oh there are iterators

and in case SALT is ever deprecated

@RikedyP Not sure that works in 18.0

@Adám Don't think so, we got quite a lot out of this I think!

Can you do this with encode/decode?

@Adám Ah fair enough

is there some :To label i can use

@xpqz I wondered for a second, let me think again

@Adám looll

@Razetime No, you'll have to implement it using i+←1 yourself.

oh that

Let's extend this to a general stride as left argument.

{2⊥⍉1,⍨⍉2(⊥⍣¯1)⍳⍵} it seemed like a good idea...

@Adám What did you use for t there. For me the last function is 3-4 times faster than the second

@Garklein {2⊥1⍪⍨2⊥⍣¯1⍳⍵} or 2⊥1⍪⍨2⊥⍣¯1⍳

@Adám and starting value perhaps?

1 1⍉1↓∘.+⍨⍳10 ⎕IO ← 0

@ovs 10000

@Adám ah I knew there had to be a better way

@FawnLocke Yay, O(n²)

:D

@Adám what was O for the other ones?

O(n)

has {⊣/⍵2⍴⍳2×⍵}already been done

@Adám oh... i thought he optimized it lol ¯_(ツ)_/¯

Another goofy ⎕IO ← 0 O(n^2), +⌿0 0 1⍉↑⍤,⍥⊂⍨∘.+⍨⍳10

Look ma, no additions: 2⍟2÷⍨4*⍳

oef...

fun

isodd←'false'≡∘⊃'iseven'⍎⍨'Data'⍎⍨∘HttpCommand.GetJSON 'Get',⍥⊆'https://api.isevenapi.xyz/api/iseven/',⍕
{n←⍵ ⋄ odds←⍬ ⋄ odds⊣{isodd ⍵: 1+odds,←⍵ ⋄ 1+⍵}⍣{n≡≢odds}0}

lol

Outsourcing the problem, eh? I guess that's a good strategy for hard or computationally expensive things like generating odd numbers.

@rak1507 I would contend that'd be the slowest solution presented.

of course, shame there's no is-even library in APL

@Razetime very neat

@Adám that is just magic dude

@xpqz challenge accepted

loooll

Oh slowest huh

@rak1507 implement it with tradfns and gotos for extra points

@Adám what is a general stride?

@rak1507 I'm surprised APLcart doesn't have it.

@Garklein tbh i went for a more enterprise style

@Richard Every nth number.

Question is where to start from. We have to take that as an additional parameter.

@Razetime Well, yes, otherwise a stride of 2 doesn't give odds.

@Adám {1+⍺×⍳⍵} is the classic (starting at 1)

4{1-⍨⍺×⍳⍵}2
starts with 3

Yeah, that's a problem.

With ⎕IO←0 and starting always at 0 you can write just ×∘⍳

theoretically. what's preventing ⎕io being -1 or 2? :)

Sanity.

loolll

@BrianBED ⎕io←3.1415 incoming

:D

How about an operator that takes the stride as left operand, the count as right argument, and an optional starting point (default is 0) as left argument.

⎕IO←0.5 should make both camps equally unhappy.

oo another chance to write enterprise code

@Adám you mean, equally happy

I've got ⋄ _Stride←{⍺←0 ⋄ ⍺+⍺⍺×⎕IO-⍨⍳⍵} ⋄ 10_Stride 4 ⋄ 8(10_Stride)4

@Adám this is the reason why APLers use idioms instead of abstractions 🤦

{2⊥1⍪⍨↑1↓⌽⍵∘⍴¨(-2×i)↑¨1⍴⍨¨i←2*¯1+⍳1+⌈⍵÷2⍟⍵}

@Adám
0 10 20 30
8 18 28 38

tried to construct the argument to base manually

@RikedyP I don't even…

yeah I think it's time for me to stop now...

that was fun

@Adám what is ⍺⍺ doing?

That's the left operand, like , in ,¨

@xpqz I haven't looked at Manacher yet but I will do at some point

@RikedyP I am approaching something reasonable I think

Yeah, this was super fun. I'll try making a video about it in the beginning of next week. See you all on Friday for Making The Grade!

@user18120371 Hi Đoàn Thành. If you want to participate here, please email access@apl.chat

So what is ⍺⍺ then in your example 8(10_Stride)4

Is someone gathering the snippets in a gist or repo?

@Richard 10

i wonder, has anyone ever made an AI that tries to simplify/optimize APL code? if not i'm not gonna be the one to do it, but it would be cool if it was a thing

@xpqz I can do so in connection with publishing the videos.

@Adám what videos?

_stride←{⍺←0 ⋄ +\⍺,(⍵-1)⍴⍺⍺} ⋄ 8(10 _stride)4

I'll try to make a short video for each event (what are calling them?)

Which of the solutions still work with right argument a negative number?

@Razetime Nice!

ah, cool

@Richard You want a negative count of numbers?

Let the stride be negative if you want descending numbers.

well, starting with a negative number

The start is the left argument.

yes ,ok

in my array language, the generalized version is (:stride :start .(* .stride + .start) take: /) used like .(...).:numbers 4 numbers 10 8

hm i don't understand the language but thats cool :o

@PyGamer0 @BrianBED on the Python thing (briefly) -- for larger lists doing the slice is ~20k times slower than iterating over reversed().

@xpqz wahaaat!!!

there are a lot of differences. the main thing is that the anonymous function .(* .stride + .start) is used as a lazy array and taken from

@AlexShroyer you seem familiar, but are you new here? Ah, I know you from the APL Farm!

Welcome to the Orchard.

Heh yeah. I didn't see anyone post exactly my solution. Are we sharing them here?

The event has ended, I think, but yes, go ahead. I might use it in my video.

wait adam do you have a channel?

@xpqz So how do you write m2←⌽m1 in Python?

@Adám oh i can only imagine

{⎕IO←0⋄((1+⍳)+⍳)⍵}

Materialise the iterator: m2 = list(reversed(m1))

@BrianBED I've not yet understood what the difference is between a channel and a normal user, but there's youtube.com/user/abrudz/videos

@xpqz Thanks. I'll stay with ⌽

I prototyped it in J first: ((1+i.)+i.)10

@AlexShroyer You can reverse the arguments of the right-most + for a nice palindromic solution.

@Adám every user has a channel they can post on. theoretcially you're guarenteed to have a channel if you have a google account but usually when someone asks if you have a channel, it usually means if you post vids on your channel

@Wezl-yizl in APL: {stride start ← ⍺ ⋄ ({start+stride×⍵}↑⊢)⍵}. (doesn't work because you can't take from functions in APL)

oh that is cute (⍳+1+⍳)

But of course, this is the same as 1+2×⍳

Shame I couldn't make +⌺(⍪2 2)⊢⍳10 useful

This was fun ;)

@phantomics made that bot from the lispcord template gist.github.com/LdBeth/b59448896a8aff20be5292914e631308

@Adám I think my "user" is Richard Park

@Adám You need 100 subscribers for a /c/ URL

Thank you all for participaing. I have to go, but by all means, feel free to continue the discussion.

Till next week :)

/e waves

when you have 2 elements and you want the second one, is doing ⊃⌽ better than 1⊃? i saw adam use it in his vid so i'm curious

I think 1⊃ is nicer, but it doesn't matter much if you know there's only two elements.

ok cool

A bit late but I haven't seen this one:

¯2-/2*⍨0,⍳

@RikedyP I am reasonably pleased with how this turned out: gist.github.com/xpqz/c19ca3ef97b5437761cad3925ba5388c

I mean, it's still scalar and traddy, but still nice.

imgur.com/GIE4H0v

eval is a workaround. Sure, you can always evaluate a string, but a string isn't a function

It's a hack not a way to implement first class functions within the language

A "kludge" is a more accurate term

hmmm but again about the map example. wikipedia says if a langauge can do map then it has first class functions. so like does APL's / not do the same as map?

or is wikipedia just wrong :p

APLs operators are a single level of higher order functions. Not the same as a first-class function

you can't pass an operator as an argument to an operator for example

It's more vague with APL since it has restricted higher order functions, but having first class functions mean functions are treated no different than any other value, so APL's operators aren't first class

ah hm ok that makes more sense

"For a language to support map, it must support passing a function as an argument." that second part of the sentence doesn't imply that it has first-class functions

oh my bad

thanks for explaining :D

is "traditional" a bad word when it comes to APL :P

)about

@BrianBED To run APL code, write code blocks starting with ⎕← or ⋄ or write a multi-line code block and prepend ⎕← or ⋄ to lines you wish to run. All matching groups / lines will be joined by ⋄ and run via TryAPL, and the output will be posted here. To format a codeblock, write `code`, or for a multi-line code block, use Shift+Enter to type multiple lines and press Ctrl-K, press the 'fixed font' button, or prepend four spaces to each line.

`⋄ create_adder←{⍎(⍕⍵),'←{',(⍕⍺),'+⍵}'} ⍝ ⍵←name, ⍺←int ⋄ 100 create_adder 'test' ⋄ test 20`

@BrianBED
NOT PERMITTED: Illegal token
      `create_adder←{⍎(⍕⍵),'←{',(⍕⍺),'+⍵}'}
     ^

`⋄ create_adder←{⍎(⍕⍵),'←{',(⍕⍺),'+⍵}'} ⍝ ⍵←name, ⍺←int ⋄ 100 create_adder 'test' ⋄ test 20`

:| i give up

⋄ dice ←{(+/((⊣,' = ',⊢)⍥⍕)⊢)?⍺⍴⍵}

@LdBeth Response looks like a 0-by-0 matrix.