The APL Orchard

nathan rogers

1:00 AM

need some help

I'm fighting with @

tio.run/…

@Adám @dzaima @ngn anybody?

iteration works

never mind

nathan rogers

3:57 AM

tio.run/…

I would like some feedback on this solution

This solution answers the question "assuming an infinite chess board, how many unique squares could the king travel to if he made only non-diagonal moves"

Basically a diamond, but solves the problem as directly as I could figure.

Quintec

4:31 AM

@nathanrogers What are you trying to do in input?

nathan rogers

oh, nothing

tio.run/…

sorry

holdover

would also like feedback for solution to problem 11 in project euler
https://tio.run/##ZVTLbtRAELzvV/SNLALiec8ckPgVw5rE0j7CeheColw4IEBZBAeUExe4cEDiAJG4ICTyJ/6RUNW2RCIsa9cz011dVd12fTS/O3tez1cHV1cya@uD/uW7k4nwWuOx330dFh0We/3rT/3uYnq7uheG3YN1O8NBh@PdDxzJv6vffZTuybZeN7KoN@v2eMioEW7FXJ53w/qhri3WcuMaTpvjDc4R8x5571ntBj6Om@VMNoeNMhmTNjVB98yDvX73s7qPoOkUP0PK43Y@H/OaGUPnjTxrN4dibnXy7LAB3erWyG1RH9OCs1f7/e6zvUGPUIMsRrWL7WL0bEXCJ6h8cXne775T2LQ/@4VCp2Pa0Xp1sG66rn3aMLzetKulPFrNt4tlJ/e1B6tlPR85rO8ADyXEX57vA@U6AS9Hq67VfIDOto8213JOxqT@7DfInP75ste/eQGA6R0@oLCYf49/vpnpkEu1a/nvGvSqUFk9lhomQsN2vukmEzmdtMujLYVnQS…

would also like to see how to golf it. couldn't seem to assign values inline when using @, and also in general with a multiline dfn

Bubbler

4:58 AM

How about this?

@nathanrogers

Adám

@nathanrogers To simulate this, it seems ⌺ is well suited.

Downgoat

6:23 AM

Hello, can anyone direct me to a good source to learn APL?

Adám

@Downgoat :-) Look at the lessons and at the chat bot's profile.

ngn

@nathanrogers 34 bytes: {⌈/,⍺×/↑⍵(⍉⍵),(⍉-∘⍳∘≢⌽⊢,≠⍨)¨⍵(⌽⍵)}

31: {⌈/,4×/↑a,(⍉⍳∘≢⌽⊢,≠⍨)¨a←⍵(⍉⌽⍵)}

nathan rogers

@Adám to simulate which?

@ngn could you do up a TIO with comments explaining how you arrived at those reductions?

@ngn seeing the result is one thing, but I don't know how you can see these, so comments on how you got there would be illuminating

ngn

@nathanrogers i can't send the tio link, it says "message too long"

nathan rogers

6:40 AM

Shift enter after @ngn

ngn

@nathanrogers first i tried doing the 4 directions (up-down, left-right, main diag, and alternative diag) separately

then i wondered if i could batch them together somehow. i noticed that padding with 0s is ok, as all numbers are positive, so a 0 product wouldn't affect the result

i noticed that the expressions that handle the 2 diagonals are similar, so i used { }¨ on the matrix and its ⌽

and finally i noticed that if i use the matrix and its rotation (⌽⍉) instead, i could save a few more bytes

tio.run/##LZS9ilxHEIXzforKZhdGqP9/…

@nathanrogers thanks, shift enter worked

Adám

@nathanrogers Flood filling. Every cell spreads to its four diagonal neighbours, ∙ becomes X

H.PWiz

11:48 AM

@ngn You can use ,∘0 instead of ⊢,≠⍨. A single 0 will stop the product having an effect

Also {⌈/∊4×⌿¨a,(⍳∘≢⌽,∘0)¨a←⍵(⍉⌽⍵)}

ngn

1:07 PM

@H.PWiz ah right, nice :)

it could be even shorter

{⌈/∊(4×⌿⊢,⍴⍴0,⊢)¨⍵(⍉⌽⍵)}

H.PWiz

1:25 PM

Very cool

I think it's actually missing a little bit. ⍴⍴0,⊢ throws away the last row

ngn

@H.PWiz oops

@H.PWiz {⌈/∊{4×⌿⍵,↑{0,⍵}\↓⍵}¨⍵(⍉⌽⍵)} ?

⎕←{{⍵,↑{'.',⍵}\↓⍵}¨⍵(⍉⌽⍵)}5 7⍴⎕a

Dyalog APL

1:46 PM

@ngn
┌──────────────────┬────────────────┐
│ABCDEFGABCDEFG    │GNUBIGNUBI      │
│HIJKLMN.HIJKLMN   │FMTAH.FMTAH     │
│OPQRSTU..OPQRSTU  │ELSZG..ELSZG    │
│VWXYZAB...VWXYZAB │DKRYF...DKRYF   │
│CDEFGHI....CDEFGHI│CJQXE....CJQXE  │
│                  │BIPWD.....BIPWD │
│                  │AHOVC......AHOVC│
└──────────────────┴────────────────┘

H.PWiz

{⌈/∊(4×⌿⊢,⍳∘≢⌽,∘0)¨⍵(⍉⌽⍵)} is shorter

Feeds

"Does APL Need a Type System?" At #FnConf18, Aaron Hsu talks about type theory and its applicability in the untyped APL world https://confengine.com/functional-conf-2018/proposal/7712/does-apl-need-a-type-system #FunctionalProgramming #APL #TypeSystem #TypeTheory #StaticallyTyped #DynamicallyTyped #DuckTyped #StronglyTyped #WeaklyTyped

Tweeted by FnConf on November 15, 2018 at 9:54 AM

Aaron Hsu (@IndianaUniv) will be at Functional Conf 2018 @FnConf co-hosting an APL Workshop (with @mkromberg) as well giving two talks. #FnConf18 #APL

Tweeted by dyalogapl on November 15, 2018 at 1:11 PM

nathan rogers

@ngn what do you mean since all numbers are positive 0 wouldn't have an effect?

ngn

2:04 PM

@nathanrogers when a padding 0 is part of a 4-vector of adjacent numbers, the product of that vector will come out as 0, so it won't affect the maximum

nathan rogers

ok what ? f←{⌈/∊(4×⌿⊢,⍴⍴0,⊢)¨⍵(⍉⌽⍵)}

oh yes I see

but what about this guy?

there's nothing for me to ⎕←

ngn

@nathanrogers that one was wrong... see the later messages

nathan rogers

what about this guy? {⌈/∊(4×⌿⊢,⍳∘≢⌽,∘0)¨⍵(⍉⌽⍵)}

ngn

@nathanrogers do you know how trains work?

nathan rogers

yes, but not how infer the behavior from reading them

ngn

2:09 PM

@nathanrogers ok, this might take some time... do you know what ,∘0 does?

nathan rogers

I'm assuming we have something like ,⍳≢⍵⌽⍵,0

binds comma with 0

whatever is to the left , 0

it will append a zero to each row in the matrix

ngn

@nathanrogers something like that. {4×⌿⍵,(⍳≢⍵)⌽⍵,0} would be the equivalent dfn

@nathanrogers yes

nathan rogers

Are trains really worth saving 2 characters?

Quintec

Well, if in code-golf, of course! :P

nathan rogers

I mean, the very existence of trains seems to contradict the goals of APL

Adám

2:23 PM

@nathanrogers Why?

dzaima

2:42 PM

@nathanrogers what's your definition of "the goals of APL"? Trains are just a fancy way to write a dfn without having to spam ⍺ & ⍵ everywhere

Adám

2:56 PM

→ 9 kittens moved to Trash

Quintec

wow

nathan rogers

8:08 PM

@Adám @dzaima eecg.toronto.edu/~jzhu/csc326/readings/iverson.pdf

> By relieving the brain of all unnecessary work, a good notation sets it free to
concentrate on more advanced problems, and in effect increases the mental power
of the race. A. N. Whitehead

It would seem that this notation in particular doesn't relieve any unnecessary work, instead it would seem to add additional cognitive parsing before intuiting the intent of the phrase.

It's like the phrase "The horse ran past the barn fell."

dzaima

@nathanrogers IMO interpreting the ⍵s and parentheses in {(+/⍵)÷≢⍵} are unnecessary work

Adám

@nathanrogers Nice to see the original look, but an easier-to-access version is here.

nathan rogers

It's a poor phrasing and requires you to backtrack reading to understand it

the same with trains. You have to backtrack to understand what kind of train it is

Adám

@nathanrogers Uh, not if it is written properly, so you can read it plainly from left to right:

+⌿ the sum ÷ divided by ≢ the count.

dzaima

@nathanrogers if you have to backtrack, you shouldn't use a train.

nathan rogers

8:12 PM

but then which of the 2 train types is it? and there's talk of additional train forms

dzaima

@nathanrogers f g h is just f g h, there's no question about it

nathan rogers

so by deafult, in longer trains you have to question where the arguments go

which often only save 2 cahracters which are usually parens

dzaima

49 secs ago, by dzaima

@nathanrogers if you have to backtrack, you shouldn't use a train.

nathan rogers

but then the train will typically use parens anyway

Adám

@dzaima Ah, but if you see a very long train, you can't see at a glance if the first function is applied monadically to the rest or is applied to the argument and then used as left argument to the next.

nathan rogers

8:13 PM

yes

dzaima

@Adám I meant that as just (f g h) alone

nathan rogers

but having f g h encourages f g hfgh fgh fhgfhgahf gaurghjltknbj

because its possible now

dzaima

I understand not wanting to read a 10 char train

Adám

@nathanrogers APL does not strive to prevent you from writing bad code.

dzaima

@nathanrogers Dyalog APL allows function arrays, but that doesn't mean I'm gonna ever use them.

nathan rogers

8:14 PM

if they didn't exist, there would bemore finite functional forms

Adám

@nathanrogers What?

dzaima

hell, Dyalog APL also has gotos, but I'm never ever going to be using them..

nathan rogers

if f g h didn't exist, then the forms would be 100% consistent

there is no problem that can't be solved without trains

in otherwords, they aren't necessary and add nothing except novelty

and obfuscate the natural functional forms

dzaima

@nathanrogers So is + unnecessary?

Erik the Outgolfer

brevity is useful sometimes

dzaima

8:15 PM

^

nathan rogers

countermanding the seeming design goals of the language

Adám

@nathanrogers You are free to not use them. But sometimes they add clarity, e.g. when (+,-) is used for traditional mathematics' ±.

nathan rogers

brevity?

> APL is the soul of brevity
W.Shakespeare

without trains

Adám

@nathanrogers Do you actually know what Iverson's design goals were?

nathan rogers

I'm trying to learn what they are, but from what I understand by reading whatever I can find is that consistent notation seemed to be a large priority

creating multiple forms of the same things seemed to be the issue with common math notation

according to the commentary on iverson concept of apl

so why introduce that into the language?

dzaima

8:17 PM

@nathanrogers so is having both > and < pointless?

nathan rogers

seems rather... inconsistent

dzaima

they're literally the same thing but with the arguments swapped.

nathan rogers

no, consistency doesn't mean no redundancy

I'm not talking about redundency

I'm talking about additional forms

there is not some special form of < or > that contradicts its normal use

dzaima

Some things are nicer when written explicitly in dfns. Some are nicer as trains. Just as sometimes using > instead of < leads to better code.

nathan rogers

like j

I disagree that your comparison is valid.

It is more akin to + vs - and > vs <

dzaima

8:20 PM

@nathanrogers do you think {(+/⍵)÷≢⍵} is more readable than (+/÷≢)?

nathan rogers

+ and - is not a different form of the same concept

absolutely

because I never know when 2 functions are joined

Adám

@nathanrogers Eh, he stated them in plain text:

• Ease of expressing constructs arising in problems
• Suggestivity
• Subordination of detail
• Economy
• Amenability to formal proofs

@nathanrogers What do you mean by joined?

dzaima

@nathanrogers well that's where I have to disagree. "Sum of omega, divided by length of omega" is much longer & doesn't explicitly say what it does, whereas "sum divided by length" is clear and to the point

nathan rogers

jot

Adám

@nathanrogers Uh, you never know if jot is there or not?

nathan rogers

8:22 PM

i don't have my keybaord

(+/÷≢)

I never know which functions are composed, glued together, jot

it isn't apparent from the notation

I'm less against trains when there is explicit joining of functions

dzaima

@nathanrogers ok, some times I don't understand ∘ either. IMO trains shouldn't have jot in them as that does just complicate things

nathan rogers

but implicit trains make no sense

Adám

@nathanrogers I don't get it. They are composed if and only if there is a jot between them.

nathan rogers

but here {+/ x %# x}

is the same as +/ % #

but then... %# is comosed clearly

Erik the Outgolfer

that's not APL...

nathan rogers

8:24 PM

i'm pseudocoding I don't have my keyboard

Adám

@dzaima I think the main reason ∘ confuses is because the monadic form is an atop while the dyadic isn't. In other words, ⍺ f∘g ⍵ and f∘g ⍵ are usually not very related.

@nathanrogers Go here to install a browser language bar.

Erik the Outgolfer

@nathanrogers nope, {+/⍵f g⍵} and +/f g aren't the same thing

nathan rogers

i'm not the one suggesting they are

6 mins ago, by dzaima

@nathanrogers do you think {(+/⍵)÷≢⍵} is more readable than (+/÷≢)?

Erik the Outgolfer

no, that's not the same as what you mentioned

dzaima

@nathanrogers the fact that (f g h) applied monadically means monadic dyadic monadic is just a thing one has to learn.

Erik the Outgolfer

8:26 PM

note the parentheses

Adám

@EriktheOutgolfer True, but {+/ ⍺ ÷≢ ⍵} is the same as {+/ ⍺ ÷∘≢ ⍵}. Maybe that's what @nathanrogers means. Nathanrogers, is it?

dzaima

@nathanrogers I don't see anything composition related there..

Adám

@dzaima ^^

dzaima

@nathanrogers uh, did you mean to not include the parentheses around +/⍵?

nathan rogers

no

÷∘≢

{⍺÷≢ ⍵}

that looks like composition

dzaima

8:33 PM

I read the dfn as {(+/⍵) ÷ ≢⍵} - with the division being powered over everything else, not it being composed with the ≢ at all

Erik the Outgolfer

{(+/⍵) ÷ ≢⍵} is the same as {(+/⍵) ÷ (≢⍵)}, APL is read right-to-left

nathan rogers

how about that

those two functions are composed

dzaima

@nathanrogers not in my mind (or my APLs interpreters)

nathan rogers

um

Erik the Outgolfer

what?

dzaima

8:36 PM

since ⍺÷≢⍵ is equal to ⍺÷(≢⍵) IMO thinking of the ÷ and ≢ as being composed together is very strange

nathan rogers

i mean

that was what I said first, isn't it?

dzaima

@nathanrogers you said ÷≢ looks like composition

nathan rogers

16 mins ago, by nathan rogers

because I never know when 2 functions are joined

i literally said that

yeah

dzaima

oh did you mean that as it shouldn't look like that?

nathan rogers

because it looks like composiition

no I mean that
> I NEVER KNOW WHEN 2 FUNCTIONS ARE JOINED

arguments are going on either said of functions that are adjoined

adjoined functions looks like function composition

I never know, when it is, or when it isn't

Erik the Outgolfer

8:38 PM

the compose operator is ∘

nathan rogers

for example.... why doesn't this work? +/÷≢ 3 4 5

but this does
avg←(+/÷≢)

avg 3 4 5

and so does this

avg←+/÷≢

avg 3 4 5

dzaima

@nathanrogers a train happens when the last thing before a ) (or } or ] or \n) is a function. Otherwise, it's regular execution

nathan rogers

that's my point

Erik the Outgolfer

+/÷≢3 4 5 is +/(÷(≢3 4 5))

you need (+/÷≢)3 4 5

nathan rogers

that's dumb. This is what I'm saying about multiple forms of the same thing. the same code meaning different things in different contexts

that's countermanding any kind of notational consistency

Erik the Outgolfer

8:41 PM

execution is right-to-left

dzaima

@nathanrogers blindly inserting a 1 2 in x 3 results in 1 2 3, which is very different from (1 2) 3 too you know

nathan rogers

but those don't look identical but mean different things

the mean different things and look like different things

dzaima

@nathanrogers they look as identical as the function case to me

nathan rogers

but avg←+/÷≢
works and has no parens, doesn't it?

Erik the Outgolfer

neither do +/÷≢3 4 5 and (+/÷≢)3 4 5

dzaima

8:42 PM

@nathanrogers 1 2 3 vs (1 2) 3 is as different as +/÷≢ 5 vs (+/÷≢) 5

nathan rogers

but (1 2) 3
and var←(1 2) 3
do not mean different things

dzaima

@nathanrogers it has a line break (or ⋄) after it though

Erik the Outgolfer

@nathanrogers I don't think variables behave like what you think

dzaima

@nathanrogers neither does +/÷≢ and avg←+/÷≢

nathan rogers

yes it does

we just talked about it

Erik the Outgolfer

8:43 PM

if I type avg ← +/ ÷ ≢, then the whole function gets stored into avg

so avg is defined as the average function

dzaima

@nathanrogers what's the difference here (except the variable setting)?

nathan rogers

+/÷≢ 3 4 5 = .33333333 or something
and avg←+/÷≢
avg 3 4 5 = 4

H.PWiz

@nathanrogers That is a false comparison.

⋄a←1 2⋄⎕←a 3

Dyalog APL

@H.PWiz
┌───┬─┐
│1 2│3│
└───┴─┘

Erik the Outgolfer

consider this analogy: x ← 2 + 3, then x × 5

nathan rogers

8:44 PM

so the exact same phrase in different contexts means different things

Erik the Outgolfer

x is defined to be 2 + 3, so 5

so x × 5 is 5 × 5, not 2 + 3 × 5

nathan rogers

no kidding

but 2 + 3 × 5 is not different from
var←2 + 3 × 5

dzaima

@nathanrogers just as 1 2 in the context of x 3 means a different thing if x is parenthesized or not

nathan rogers

i litearlly pased the exact same phrase, and in different contexts means different things

Erik the Outgolfer

uh

nathan rogers

8:45 PM

no no nowait, in the inline context you need (2 + 3) × 5

Erik the Outgolfer

exactly

nathan rogers

that's literally what I posted

not fucking exactly

that makes no sense

dzaima

@nathanrogers and in the case of +/÷≢ in the inline case you need parentheses too!!

Erik the Outgolfer

please remain calm...

nathan rogers

i'm calm, but then I'm tyring to make a point and being ear boxed

Erik the Outgolfer

8:46 PM

so, in algebra, if you define x = 2 + 3, what will x's value be?

dzaima

@nathanrogers you say "no no nowait" and ask adding "random" parentheses around 2 + 3. Why can't I ask adding "random" parentheses around +/÷≢?

nathan rogers

> Nevertheless, mathematical notation has serious deficiencies. In
particular, it lacks universality, and must be interpreted differently
according to the topic, according to the author, and even according to
the immediate context

Iverson

Argue with him. Is all I'm saying

Erik the Outgolfer

looks like you've misunderstood some things

nathan rogers

trains create multiple forms of the same thing, instead of the universality

dzaima

@nathanrogers APL still needs reading of right to left. It isn't completely 100% context independent you know

nathan rogers

8:50 PM

> and the lack of
ambiguity makes possible precise thought experiments. -Iverson again

dzaima

the case of a←2+3 ⋄ a × 5 vs 2+3 × 5 is a perfect example that you need context in APL too.

nathan rogers

take javascrript for instance (() => ...) function () {} const x = v => ... there are like 7 different ways to create or invoke a function

it's nonsense

APL without trains has a consistant and universal notaiton

H.PWiz

FWIW, I think it is valid to dislike functions behaving differently in different contexts (trains and not trains), but the conversation about parentheses and assignment is not really an argument.

Erik the Outgolfer

a variable is a container, in other words

just like parentheses

nathan rogers

yes, I totally concede to that. I wasn't trying to discuss that, that's what the discussion devolved into

dzaima

8:52 PM

@nathanrogers the difference is that all of those do the same thing in JS, but in APL trains vs dfns allow emphasizing different characteristics of the code

nathan rogers

I made points, and instead of addressing them we went down some olther rabbit hole. the earboxing I was talking about

but the charactiaristics of the code, without trains, is UNIVERSAL

and consistent in all contexts

at least with dfns

I think tradfns are bonkers

dzaima

yeah I agree that the whole parenthesizing argument got a bit out of hand, but you said that trains requiring parentheses is strange and I couldn't just accept you saying that..

H.PWiz

@nathanrogers Ambitious statement. I'm sure you can find more weird things if you look hard enough

dzaima

personally I'd be fine with trains requiring another type of parentheses to indicate their existence, but regular parentheses work just fine

nathan rogers

no, I wasn't saying that trains using parens is strange. I was saying that, inresponse to your complaint about (+/⍵) making thing less readable, I agree, but at the same time, you use parens in trains anyway, so that isn't really a valid comparison

dzaima

8:55 PM

@nathanrogers if you use parentheses in trains, you shouldn't use trains IMO.

The outer parentheses of trains are alike to the {} of dfns

Erik the Outgolfer

the difference is that the train only has the outer parentheses, while the dfn has the outer brackets and parentheses inside too... if the train has too many parentheses, then what dzaima said is true ^^

nathan rogers

46 mins ago, by dzaima

@nathanrogers IMO interpreting the ⍵s and parentheses in {(+/⍵)÷≢⍵} are unnecessary work

so then I said trains use parens anyway

Erik the Outgolfer

but they're the outer parentheses, not parentheses inside the function

dzaima

@nathanrogers the dfn has both {} and (). The train has only (), in place of {}

nathan rogers

that's where things got a little off. I wasn't complaining about parens, but that parens are illegible in regular dfns or in trains, so that didn't seem like a complaint

yeah I hate {}

I'd take any other kind of bracket than those. they make my eyes bleed in most editors with most color schemes

dzaima

8:58 PM

@nathanrogers but there's no way to not have them.

nathan rogers

just design a more human braket

Erik the Outgolfer

like what?

nathan rogers

how about dented parens

Erik the Outgolfer