Why not use ⊃⌽⍸ to find the last 1?

@rak1507 Sure.

@rak1507 on a boolean vector with all 1s you'll temporarily take up at least 8x the memory that the vector itself took

Oh right

The scan solution is nice, I wouldn't have thought of that

⊢/ is on the tips thread somewhere

@rak1507 <\ is like one of the cases where the ordering of scan is "useful"

Yeah

@dzaima But on a sparse vector, you can save a lot of memory instead. Best would be to have thunks so that 1⍳⍨⌽ would search from the rear and stop early.

is there a way to leak memory in APL?

@Razetime Dyalog allows creating object cycles with namespaces, but it also has a garbage collector, so no

need O(infinity) space complexity

lol

@Adám i'd imagine thunks wouldn't magically make that easier to achieve than with idioms

anyone doing Advent of code this year?

Yeah I'll try

Was inspired by bubbler's repo to try it out

I think I'll give it a go even if I don't get very far, some of the later challenges seem to be a bit tedious I think

@Razetime I might do it

let's see who gets the farthest

How does it work, though? Is there a new challenge each day?

@Razetime i'll probably do some of it, but probably not going for answering speed as it happens in 7:00 in the morning :|

And you only get the next after you solved the previous?

No, you can solve them in any order (right? idk)

@RGS every day has 2 parts. Finishing the first unlocks the second, but you have access to all (past) part 1s

you can use the questions in any way

APL orchard leaderboard y/n?

just simply an endurance kinda thing

how many days consecutively

or longest streak idk

@dzaima Ah I see, ty

every day has a leaderboard of the first (iirc) hundred solvers of each part

there's a private leaderboard of TNB users - 383923-1cc3bff3

@dzaima first?

as in, FGOTW?

@RGS yeah. speedrunning solving challenges :)

the competitive coding kinda thing

@Razetime yeah

I'll solve a couple of them for sure

But won't be competing to place high in the leaderboard

yay

Me neither, especially considering the time

apparently on 2018 i got a reasonable place on one day. Last year i didn't at all compete for time, and i plan to repeat that

woah

dzaima pro coder

@dzaima It'd allow flexibility in phrasing.

@Razetime i know right

very pro considering [redacted information here]

how did you do it without competing for time

them people trying speedrun musta been reeling

I really don't like that everyone has to sit ready at a specific time that is probably very inconvenient for many. Like it is 5 AM here. Also it discriminates against Jews that are prohibited by their religion from using a computer for at least 25 consecutive hours every week.

> If you can't participate at midnight, that's not a problem; many people use private leaderboards to compete with people in their area.

@Adám The "sitting still waiting" is really annoying. Also, is this sort of "accidental" discrimination offensive or just annoying?

@Razetime that was 2018 not 2019. Then i did try to go for time for the first 20 days or so

aww would have been funnier

@Adám That is a very useful rule

@RGS I know full well that no discrimination was intended, and I have no hard feelings at all either. The organisers probably don't know about such religious rules. I don't have an alternative either. Still, it isn't "good". Same issue with SE badges for participation streaks.

@Razetime Indeed. My boss says he wished a deity enforced such a rule on him too. The only way to make people truly take a day off is through religion.

@Adám would it be fair to say that the timeframe when some Jew cannot use computers anywhere on the world is much longer than 25 consecutive hours? I doubt it'd be possible to have a thing happening every day and not discriminate anybody

@Adám I agree it isn't good. I'm just thinking for myself that even a person that is actively looking to avoid these things is in for a lot of researching and whatnot

iirc the organizer said that the time is so that they would have the most time available to moderate/fix things as needed

@dzaima Yes, some Jew cannot participate from sunset on the Aleutian Islands until nightfall in Alaska the next day, so at least 49 hours or so.

@dzaima And that's totally reasonable.

@Adám so 2 full days when nothing can happen. So absolutely 0 chance of a daily advent with 0 discrimination

Of course, and maybe other religions/cultures/whatnot have other restrictions. Who knows. So I don't really like the format as a whole.

understandable

Is there a simpler way to do ⍳⍨∘∪⍨ ?

Not sure how exactly to describe the problem

@rak1507 not in dyalog at least. ∪⍛⍳ in Extended/dzaima

By now, I've started reading f⍨∘g⍨ as an idiomatic phrase.

Right, but is that the best way to get a list of numbers corresponding to each item

Can you give example args?

Or is that monadic?

Monadic

⋄ ⍳⍨∘∪⍨ 'ababcbaacd'

@rak1507 1 2 1 2 3 2 1 1 3 4

Aha, so it is like ⍳⍨ but without skipping any indices.

Yep

I'd probably write ∪⍳⊢ for clarity.

Right, makes sense

^^ +←1; ∪⍛⍳ is only for dyadically calls

@Adám Indeed

@rak1507 How about "Consecutive ids (indices with equal major cells mapping to same index)"?

That makes sense, I should've checked aplcart first

:-D

What would you have searched for?

I'm generally bad at phrasing my questions so probably ∪ and ⍳

@Adám I have some recollection of adding this to APL Cart with you

Whenever I can't think of something I guess the primitives most likely to be involved

(just reminiscing about the past, nvm)

@rak1507 That's actually a pretty cool technique.

APL is easier than english

@rak1507 You damn right it is. And also more international

@Razetime did you end up applying?

I messaged the person in the "ad" because I was curious to know what they were using APL for but they never got back to me

I know of another company seeking someone to do APL in Sweden. Applicant doesn't have to know APL now; education is available there.

CMP: Would it make sense to create an APL Wiki page "Jobs" where the talk page is for companies and individuals to list themselves?

What does CMP stand for?

@dzaima went through all the files i had for AOC18, and there's nothing about day 10, so it was probably all done inside the Dyalog REPL :/

@rak1507 chat mini poll

Ah

@Adám Maybe "Careers"? but I'd vote Y

Well, it isn't just for a full career. It could also be for the odd one-off job.

Yeah true

Sounds like a decent idea whatever name you pick

@RGS yes, I did

@Razetime Cool :D Good luck! And btw, do you have any job description you can share? "I messaged the person in the "ad" because I was curious to know what they were using APL for but they never got back to me"

@RGS messaged them 18th and they replied today

they might have responded to you

No job description given yet, they asked for my resume

@Razetime You can include your mini-project for Dyalog!

Should add that

ok, done!

@Razetime messaged them on the 17th :/

Oh well

I asked how I could apply

so maybe that worked idk

Hey there @xificurC, the upcoming Advent of Code is going to be a great way to exercise your APL skills.

Hi @RGS! I am looking forward for AoC, but my heart is set on using forth. Unless someone persuades me otherwise :) I visited today because I expect a lot of churn here when AoC comes on, is that correct? :)

Interesting, I've heard about Forth but never looked at it.

I'm confident at least a couple of problems will be particularly simple in APL

You could use APL for those!

@RGS it's another mind-bending experience! But while APL is a very high-level language forth is very low-level. You can think of it like an interactive concatenative (arguments passed through a stack) assembler

@xificurC That does sound interesting :) maybe next year I'll attempt AoC in forth?

@RGS Yeah I'm keeping that in mind too :) Or at least look at and discuss other people's solutions

@RGS I'd be willing to reciprocate and teach you the basics ;)

@RGS will you be solving the AoC puzzles? Last 2 years I only got till 7-8 solutions (I don't have time during the weekend and then I just lose breath)

I will give them a go once in a while, if I remember them. But I don't expect to put much effort into them, as I don't feel I have the time.

forth could be fun if it had nice docs somewhere

@Razetime supposedly, the link xificurC sent me says there's a large documentation available

@Razetime there's many forth dialects, since it's so easy to write one. If you're interested in the beefy ANS Forth there's e.g. a long gforth manual. FreeForth which I linked is pretty well documented as well. I agree it's an exception though.

@Razetime I'm going to try and be "competitive" this year, and I'll be streaming my attempts to do so

I'm actually planning on starting Thursday night (Friday 05:00 UTC) with some previous year challenges to warm up, and I'll try streaming daily through Dec 24th

We'll see if that actually lasts :)

Good luck for that ! I hope you have fun :)

@RGS Thanks! It should be a good time, I had a ton of fun in 2019 though the "intcode" challenges were mostly just tedious

@voidhawk I wouldn't know, never solved an AoC challenge! What was an "intcode" challenge?

Basically they came up with a machine code spec (opcode decoding, relative addressing etc.) and then give you an integer vector representing code/memory. Lots of the challenges were running the "program" against your implementation, blocking for input/output and whatnot. Later there was some stuff about running multiple independent instances and doing networking

Every other challenge was an intcode problem, often with an extension to the spec

I see what you are talking about

Yep, I have a walkthrough of one of the problems here: youtube.com/watch?v=HYsaSae5Bbo

ngn has very impressive solutions for all 2019 problems in ngn/k here

@voidhawk not sure if all, some seem to be missing

Hrmmm, "most" then

Is there a free version of NARS2000 online? I wanted to try out its rational numbers feature but apparently it's not on TIO.

@user not that I know of (and i doubt there could be without significant work)

Oh, that's too bad.

@user I guess the only reasonable alternative is for you to make your own APL implementation with rationals (certainly missing MANY links)

Oh cool, are those all by Stack Exchange users?

first is dzaima's, second is ngn's, 3rd (in early stages) is mine, 4th is by Elias Mortensen (..son?); fifth is by "phantomics" but I never saw them around here, only on the IRC chat

so many are definitely by SE users

Ah, I see

@user if swapping dzaima/APL and ngn/apl, they are ordered by time since being in this room. mine and RGS's are the only ones made after joining, and all have messages here from the author

<phantomics> Hi @user, my APL compiler supports rational numbers: github.com/phantomics/april

<phantomics> ÷⍳5 => 1 1r2 1r3 1r4 1r5

<phantomics> It also supports rational parts of complex numbers: 3r4J9r5×⍳4 => 3r4J9r5 3r2J18r5 9r4J27r5 3J36r5

@phantomics April sounds really cool, I'll make sure to try it out

ngn/apl, dzaima/APL, and April have a page on APL wiki, along with many other implementations

(Zachary's RAD and Dennis's Jelly are there too, listed in "derivatives")