The APL Orchard

xpqz

12:06 PM

Is there a name people use for split-transpose-mix ↓⍉↑?

Adám

@xpqz I informally call it (nested) transpose (of a vector of vectors).

RGS

@xpqz I call it remix, but that's just for myself hehe

xpqz

@RGS remix! that's genius.

Adám

I've now added that synonym to APLcart

@xpqz "Zip"

RGS

@Adám ah that's a good one

The zip function in Python kind of does what ↓⍉↑ does

xpqz

12:18 PM

Not quite py-zip -- isn't that more like

{(⊂⍋∊⍳∘≢¨⍺ ⍵)⌷⍺⍪⍵}

Adám

@xpqz I.e. if lengths don't match, then unmatched elements become trailing.

xpqz

@Adám Yeah, had to remind myself. Py's zip will only line up elemets to the shortest vector:

>>> list(zip([1,2,3,4,5],['a','b','c']))
[(1, 'a'), (2, 'b'), (3, 'c')]

Adám

CMC: ^

RGS

@xpqz I didn't say it was like py-zip, I said it is "kind of", as in "there's some resemblance"

xpqz

There is indeed a striking resemblance.

Adám

12:31 PM

I've got 16 bytes.

RGS

I have 15

but maybe I misunderstood the problem..?

Adám

      ⎕SE.Dyalog.Utils.repObj 'abc'YourFn⍳5
('a' 1) ('b' 2) ('c' 3)

Well, actually Python's zip() takes an arbitrary number of lists.

RGS

18 bytes for an arbitrary number of lists as right argument

but I wouldn't say it is golfed

Adám

14 for that.

RGS

      ⎕Se.Dyalog.Utils.repObj myFn ('abc') (⍳5) (⍳10) ('bananas')
('a' 1 1 'b') ('b' 2 2 'a') ('c' 3 3 'n')

Adám

12:37 PM

Yup. 14 for both my dfn and my train solutions.

xpqz

Python wins this CMC :) list(zip(*l)) -- 13

RGS

I wrote the 18 outright as a train but I'm a monkey, so that train can be golfed 4 sure

@Adám I want to have a look at these

Here is mine

Adám

@xpqz Yeah, but there isn't much of a difference between the length of APL's code (which says what happens with uneven lengths) and using Python's built-ins. And furthermore, APL's code can be customised to do specify other behaviours than cutting short. Try erroring or filling on unevenlengths in Python!

xpqz

fo sho.

Adám

@RGS Mine

xpqz

12:43 PM

CMC: same, but for unzip. I have 10

RGS

Ah, turns out mine only has 15 bytes, I had extra ∘ and () that I could remove if I were to assign the train to a name instead of using it inline

dzaima

@Adám i have 11

Adám

@dzaima Ooh? Care to share with APLcart?

@xpqz Uh, isn't that just ↓⍉↑?

dzaima

@dzaima (a train, no ⊢)

Adám

@xpqz I seem to remember it being called "the clothes dryer" or "the washing machine" or something like that, but I can't find a reference.

CMC: ↓⍉↑ but error on length mismatch.

I've got 11 on that.

dzaima

12:57 PM

i've got a stupid 13

down to 11, but still stupid ("stupid" as in it doesn't work in dzaima/APL and possibly other non-conforming APLs)

Adám

1:17 PM

@dzaima Yeah, mine is definitely a hack.

rak1507

{↓⍉↑(⊢↑¨⍨(⌊/≢¨))⍵}

Hmm how can you get 11

dzaima

@rak1507 1) all train, 2) get rid of one of the ∘s required in ↑∘⍉∘↓ by moving things around (and save another byte in that process)

@rak1507 also that ⍨ there is pointless

rak1507

true

dzaima

@dzaima (been sitting looking at the language bar trying to think of more primitives that error "helpfully" for a while, but that's only resulted in a 12-byter)

rak1507

1:33 PM

Oops I just rearranged my language bar, how do I reset it

{↓⍉↑(⊃1⍨¨¨⍵)∘/¨⍵}

Here's one that errors

lots of ¨s

dzaima

@Adám got a 10 by trying to make an erroring one :D (definitely way more obscure)

@dzaima it's 6 in BQN

ngn

@Adám i've heard "washing machine". but to me it looks more like a screw with instructions for how to unscrew it - something ikea would draw.

Dyalog APL

<phantomics> Are those ¨¨s in sequence valid? There's no function between the ⍵ and ⍨

dzaima

@dzaima on and it also discloses the items, so doesn't work on e.g. 'abc' (,¨1 2 3 4) :/

phantomics: yeah, Dyalog extended A⍨ for array A to be equivalent-ish to {A}

Dyalog APL

<phantomics> Makes sense

<phantomics> Also had a question about your APL, is it usable within Java the way April is usable within Common Lisp? Can you evaluate an APL string on an array or scalar value within a Java program?

dzaima

1:50 PM

phantomics: there'd have to be an interface for doing that well, as I have custom types around all types as native Java arrays can't hold shape/rank information. But with some intermediate conversion, it's definitely feasible

@dzaima (luckily, i (can't remember if publicly) already hated the idiom i've used there that makes it do that)

@dzaima (and yes, the 6-byter in BQN also gives a wrong result for ⟨1‿2‿3‿4‿5,⥊¨"abc"⟩)

(if i haven't criticized the idiom, i definitely have the thing that makes it do the funky disclosing)

(/rant over)

Adám

@rak1507 On Windows: Right click it and select "Revert to defaults"

@dzaima Are you going to reveal your solutions?

dzaima

@Adám it's probably been long enough

my 11-byter for zip

Adám

@dzaima I have 9 as full program.

Dyalog APL

<phantomics> I see, Lisp has native n-dimensional arrays so it's a lot easier to pass them back and forth

Adám

@dzaima That's really good. I'll APLcart that.

dzaima

1:59 PM

@Adám so that's another solution

(my 10)

@dzaima (and i still hate that 1,2 silently works very very much)

Adám

@dzaima Ugh, that's horribly wonderful.

RGS

2:18 PM

When localising variables in tradfn headers is it bad practice to write spaces after the semi-colons? smth like
∇ r ← Tradfn arg; var1; var2; var3...

I often see everything mushed together, as in ∇ r ← Tradfn arg;var1;var2;var3...

ngn

@RGS writing tradfns is bad enough practice

RGS

@ngn Let us assume I have to write tradfns :P

ngn

@RGS spaces don't matter. functions are stored in tokenized form.

so next time you open the function, the interpreter will format the list of locals according to its own taste

RGS

I don't think that is what happens if I have the tradfn in a text file, or is it?

ngn

if you're writing source code in files (as you should), i think personally i'd prefer no spaces

dzaima

2:25 PM

@ngn you'd prefer no spaces anywhere though :p

RGS

Well, I didn't have to ask here to know you would prefer no spaces

dzaima

(personally i'd prefer with spaces, but i've used it like, once?)

ngn

@dzaima why insert meaningless bytes everywhere :)

dzaima

@ngn because they might or might not [citation needed] improve readability to the weird mess that is a human brain

ngn

"readability"..

Adám

2:29 PM

@RGS No.

RGS

@ngn "readability is in the eye of the beholder"

@Adám ok, thanks. and is it good practice?

Adám

No. :-) Can't really be due to it traditionally being auto-removed.

RGS

makes sense

Razetime

CMC: Cut an array into n equal parts.

Adám

@Razetime Anywhere?

RGS

2:31 PM

@Razetime n is known to be smaller than or equal to the length of the array?

Razetime

yep

but n may not be divisible by the length

Adám

@Razetime ⊂⍨∘,

RGS

@Razetime hence Adám's question

dzaima

@Razetime so error if impossible?

Adám

⋄ 'Hello world'⊂⍨∘,4

Dyalog APL

2:32 PM

@Adám Hello world

ngn

@RGS be careful who you choose as your judge of "good". apl culture is generally tolerant to bloat.

Marshall

@Razetime Obvious BQN solution.

RGS

⋄ 3 {⍵⊂⍨⌽0=⍺|⍳≢⍵} '123456789' @Razetime smth like this..?

Dyalog APL

@RGS 123 456 789

dzaima

@ngn unless you're forcefully filling up every line to the width of your screen, spaces will only "bloat" source code size (who cares about that?), and, at worst, probably [citation needed] won't affect "readability", and, at best, improve it significantly

Razetime

2:35 PM

I made a really really long answer to that

like, really long

so I was wondering if there was something shorter

RGS

@Adám this was a "joke", right? because the 4 parts don't have equal lengths..?

Razetime

@Adám but this is insane

{0~⍨¨(x⍴1,0/⍨⌊⍺÷⍨⍴⍵)⊂⍵↑⍨x←(⍴⍵)+⍺|⍴⍵}

RGS

@Razetime look closer, I think that was the answer before you specified the parts had to have equal sizes

Razetime

More accurately, this question

@RGS bad wording

@Marshall should try using BQN sometime

maybe we should put up an entry for it on LoTM

or is it already there

Adám

@RGS Yeah, the challenge was underspecified.

Marshall

2:38 PM

@Razetime Don't think I know that acronym?

Razetime

Language of the month

dzaima

@Razetime so what is the correct wording - as the challenge?

RGS

@Razetime if you don't mind I'll post my dfn

Razetime

yes, just read the challenge, pretty much

@RGS Oh sure go on

Adám

16

Q: Nominations for Language of the Month, 2020 edition

We've decided to revive the Language of the Month event. We've also decided to start a new post for nominations. (You can see the old nominations post here.) So let's get to nominating! Procedure A language can be nominated for Language of the Month by posting an answer to this thread. We all vot...

Razetime

2:39 PM

but Adám's is the correct one

shortest, sorry

Adám

@Razetime Wasn't this a competition phase I question?

RGS

@Razetime Adám's doesn't solve the challenge :)

Razetime

@RGS lol it just splits empty space

didn't realize

Marshall

By the way, Partition representations is the guide to always knowing how to split up data. BQN's group (⊔) uses target indices like Dyalog's ⊆, but it actually uses the numbers as indices instead of just looking at where they change, and you can rearrange them to group distant cells together.

Adám

@Razetime ⋄ 5{⍵⊆⍨⌈⍺÷l÷⍳l←≢⍵}'Hello world'

Dyalog APL

2:42 PM

@Adám He ll o wo rld

Razetime

⋄4{⍵⊂⍨⌽0=⍺|⍳≢⍵}'Hello World'

Dyalog APL

@Razetime llo Wo rld

RGS

@Razetime lol mine split things in sublists of length ⍺, not into ⍺ sublists

also it is sensible to ⎕IO

Razetime

loool

@Marshall Thanks for the link

Adám

@Razetime Try it with 4.

Razetime

2:44 PM

⋄4{⍵⊂⍨⌽0=⍺|⍳≢⍵}'Hello World'

Dyalog APL

@Razetime lo W orld

RGS

4 {⍵⊂⍨⎕IO=⍺|⍳≢⍵} 'Hello world' tries to split in pieces with size 4, not in 4 pieces :P

Razetime

⋄5{⍵⊆⍨⌈⍺÷l÷⍳l←≢⍵}'Goodbye, cruel world!'

Dyalog APL

@Razetime Good bye, cru el w orld!

RGS

⋄ 4 {⍵⊂⍨⎕IO=⍺|⍳≢⍵} 'Hello world'

Dyalog APL

2:46 PM

@RGS Hell o wo rld

Razetime

owo

Adám

@Razetime Extended, 12 bytes ⊢⊆⍨1+∘∧|∘⍳∘≢

Razetime

@Adám kewl, what exactly is it using from extended?

Adám

@Razetime monadic ∧ for sort ascending

Razetime

It works almost prefectly

question states that the last element must be the longest

Well, now I know that cutting builtins are cheating

Adám

2:54 PM

@Razetime Oh, I missed the link to the full spec.

Marshall

@Razetime Yeah, seems to be kind of a contest of which language's partition method happens to do what the spec wants out of a large number of reasonable options.

Razetime

12

Q: N-chotomize a list

Given a list of integers L, and an integer N, output L splitted in N sublists of equal lenghts. Non-divisible lengths If N does not divide the length of L, then it is not possible that all sublists have equal length. In any case, the last sublist of the output is the one that adapts its length...

code-golf array-manipulation

Marshall

Following the partition representations page, division endpoints is the representation that fits the task, and others have to convert.

Razetime

oh nice

@RGS still posting?

codegolf.stackexchange.com/questions/76355/… This is a more interesting challenge.

RGS

3:12 PM

@Razetime I was solving the wrong challenge

and I've got a lot on my plate so I didn't bother fixing it :P

go ahead

Marshall

3:24 PM

Not too awful BQN solution to N-chotomize. The parens and ˜ aren't needed to define the function, so it's 19.

Basically the Pyth solution using Interval Index (⍋) to convert from division endpoints to target indices.

Marshall

3:40 PM

Also reasonable at 20. Get division endpoints from beginnings by shifting in the total length at the end, lengths by subtracting beginning from end, and indices with Indices/Where (/).

Just for completeness the third way through the diagram is 22.

TessellatingHeckler

6:43 PM

@ngn becarefulwhoyouchooseasyourjudgeof"good".aplcultureisgenerallytoleranttobloat. But why don't you write like this?

ngn

@TessellatingHeckler because space acts as a separator

but in local name lists there already is a different separator: ;

TessellatingHeckler

I went looking for how to outer-product in J ( aplwiki.com/wiki/Outer_Product ) but instead found how to run J Server via COM; turns out it's also easy to use from PowerShell after running the jreg.cmd to register it, $j = New-Object -ComObject "jexeserver"

Then you can $j.Do("q=.i.9"); $x=$null; $j.Get("q", [ref]$x)

Marshall

@TessellatingHeckler Adverb / in case you're still wondering.

I've been thinking about that wiki page. Writing it will take forever though.

TessellatingHeckler

@Marshall Thanks; I found it in "J for the APL Programmer" article in the end

@ngn that's a good point

Marshall

I guess a nested APL outer product is a little more like &.>/, since J's table works on cells and not elements.

Adám

6:58 PM

@ngn Do you put a space after "." and "," in English?

ngn

@Adám usually yes

Adám

@ngn but there already is a separator…

ngn

english is usually written in a variable-width font, so . and , are too narrow

dzaima

@ngn so that's.. too little spacing?

ngn

and english has a lot of history.. programming languages are supposed to be driven by practical considerations

@dzaima yes, unlike in a mono font where ";" and " " are the same width

dzaima

7:02 PM

@ngn that's the more understandable answer. (one could also say the space can be a disambiguator)

@dzaima but in most cases it's not disambiguating, like in APL, but just as it's pretty helpful for the sentences to be very visibly separated (when not looking directly at the separator), it can be beneficial in code

ngn

@Adám a better example is bash - usually i put a space after ; there

because bash requires spaces in most other places, so "a;b" would look like a single token

Adám

8:08 PM

@RGS I've lately begun to put the locals on a separate line, especially if there are a few of them. And on multiple lines by name class or by usage if there are a lot of them.

TessellatingHeckler

I have wondered if there are any English words with a space in them. It doesn't seem like it, but then English has so many broken "rules" that it would be surprising if "space separates words" is an unbroken rule.

dzaima

@TessellatingHeckler i'm pretty certain a word by definition doesn't have spaces

Adám

@TessellatingHeckler Well, kind-of. There are compound nouns that in related languages would be written without space. E.g. full moon which in Danish is fuldmåne.

dzaima

@Adám that's still very much two separate words in english

Adám

@dzaima Is it now? It is read as a single word with no pause.

dzaima

8:16 PM

@Adám i don't think i'd read it with a smaller pause than other words, but i'm not native english so what do i know

Adám

@dzaima Compare: "Tilt the board and the egg rolls down." with "Let's eat egg rolls today."

RGS

@Adám how does one do that..?

Adám

@RGS APLcart…

dzaima

@dzaima (fwiw "full moon" is a single word in latvian too - "pilnmēness")

@Adám still very much two words. Single concept, maybe, but definitely two words

TessellatingHeckler

@dzaima sure, generally, but where does that definition come from? There are current or ancient writing systems with no spaces between words. English has no "Académie Française" to officially make rules about the language. English words have "foreign" letters and accents sometimes, hyphens, apostrophes, compounding, many loan words ... it wouldn't surprise me if there was some argument of a single word with a space in it.

@Adám hmm, those are good suggestions

dzaima

8:21 PM

@TessellatingHeckler i mean, sure, you could define "single word" to be a single word, but that wouldn't be very helpful. hyphens, apostrophes, etc are a more confusing topic, but i'm pretty sure a space can't be a part of a word

TessellatingHeckler

@dzaima is "to be" a single word?

dzaima

(also fwiw wikipedia mentions "ice cream, … get up each are generally considered to consist of more than one word")

Adám

See you in the new year, after new year.

dzaima

@TessellatingHeckler no, at least as far as i'm considered

TessellatingHeckler

@dzaima that page includes a mention of German "Similarly, some have separable affixes; in the German sentence "Ich komme gut zu Hause an", the verb ankommen is separated."

dzaima

8:34 PM

@TessellatingHeckler i'm mainly talking about english here, but even there space isn't part of the "word" (wouldn't " gut zu Hause " then count as "part of the word"?)

but, as with most of linguistics, agreement is definitely unattainable :)

TessellatingHeckler

@dzaima I was also talking of English originally, just noticed that. I'm not trying to argue it one way or another; it occurred to me that I would be surprised if there is a word with a space in it, and surprised if there was an unbroken rule in English, and presumably both cannot be true, so I wondered which way it resolved. It probably is not attainable agreement. :)

Adám

@dzaima No, they serve distinct purposes, where as komme and an are inseparable. Similarly in French "Je ne peux pas."

@TessellatingHeckler You'd be surprised… ;-)

dzaima

@TessellatingHeckler i mean, the nature of the question is very dependent on the exact definition of what a "word" is, and that's an extremely complicated question no linguist knows the answer to

@Adám assuming that's a response of the parenthesized part, that was an intentionally stupid question

Adám

@dzaima True, and "word" may not even have the same definition in a spoken language and its written counterpart. Look how spaces are used in IPA…

RGS

8:55 PM

@Adám xD

ngn

9:06 PM

@TessellatingHeckler the word "workspaces" has a space in it :P

"no one", by analogy with "nobody", etc, should have been (or is?) a single word, but it's written with a space to prevent the two "o"-s from merging into a single vowel

dzaima

@ngn seems i don't care :)

ngn

@dzaima what does the line over "e" mean?

Adám

@ngn no-one or noöne

dzaima

@ngn just another letter - it's a "long" form of it

ngn

10:12 PM

whoneedsspacesanyway

Adám

@ngn It took me way too long to read that.

RGS

10:38 PM

@ngn spacesAreSuperfluousIfYouUseCamelCase,IAgree

ngn

@RGS the romans didn't even use lowercase :)

RGS

I don't trust the capitalisation conventions of a culture that didn't know 0 :P

ngn

but arabic and the indian scripts (afaik) don't have case distinction either

Dyalog APL

<moon-child> no, but they have specialized letters for the beginnings and ends of words, so you can still distinguish

ngn

specialized letters - but only some of them?

Dyalog APL

10:47 PM

<moon-child> most of them afaik. It's not like hebrew where there are only 4 or 5 letters with a final form

Adám

The five "Hebrew" final forms are actually Aramaic and of newer origin. Original Hebrew has no final forms, and was apparently always written with spaces or inter-puncts.

@RGS IsGIFWithAGOrAJ?

:55761771 My point exactly.

RGS

@Adám still better than "isgifwithagoraj"

Adám

@RGS Yes, but isn't "Is GIF with a G or a J" best?

RGS

@Adám much better tbh

I overlooked these problems yeah

ngn

11:04 PM

@DyalogAPL right, nicely laid out in a table here

Transcript for