Back after a 3 week hiatus of family visits, R&R, and lots of interpreter work :P

@Adám Just checked this out. Looks like a fun hangout session! I am definitely of the side that wants to commute the arguments for - and ÷.

Anyway, how's everyone's Friday?

Okay, so I have a depth vector dp, but it's not normalized. I.e. children can have depths arbitrarily greater than their parent. For example, 0 3 2 3 normalizes to 0 1 1 2. Notice how right-siblings can be lower than left-siblings.

Anyway, I want to turn dp into a parent vector. This is what I have |(-p)⌈⌈/∘⍸⍤1⊢(∘.>⍨p←⍳≢dp)∧∘.>⍨dp

In words, what I'm trying to say is that a node's parent is the rightmost node of lesser depth, otherwise it's self (root nodes self-loop).

I think the code above pretty much reads like that, but I suspect it has room for improvement. Any suggestions welcome!

@Adám which problem are we on today?

problems.tryapl.org/psets/2019.html?goto=P6_Telephone_Names

> Write an APL function that takes a character vector right argument that consists of digits and uppercase letters and returns an integer vector of the corresponding digits on the keypad:

  	ABC	DEF
 1	 2	 3
GHI	JKL	MNO
 4	 5	 6
PQRS	TUV	XYZ
 7	 8	 9

 *	 0	 #

I'm actually interested in seeing a solution with the hinted-at ∊ since I didn't see a need for it at all.

@Adám It;s called Telephoner numbers btw

{⍵∊⎕D:⍎⍵ ⋄ 1+⌈3÷⍨i-19 26⍸i←⎕A⍳⍵}¨

nice

ah :)

⍎⍕{n,⍵/⍨~⍴n←1+⍸⍵∊¨'ABC' 'DEF' 'GHI' 'JKL' 'MNO' 'PQRS' 'TUV' 'WXYZ'}¨

@rabbitgrowth Whoa, what?

I think we're done :)

Works, too.

@rabbitgrowth does this work? 7 is a group of 4 letters. Or please explain me

@xpqz I had 10|(1↓⎕D,'@@ADGJMPTW')∘⍸

Not a ∊ in sight.

Right, not even an uppercase ∊

I understand @Adám's version. Not @rabbitgrowth's, at a glance.

@Richard That's what the ⍸ is for - whether or not you need to adjust once you've gone beyond O

@Richard Ah, I get it now. Right that ↑

I didn't know ⍸ worked with letters :(

it doesn't you've done ⎕a⍳ first

⍸⍒⍋ use Total Array Ordering.

@Silas Sure it does. I used that in my solution.

oh sorry

was just focused on fact you've done the dyadic ⍳

@Silas right :) I used ⍸ just like that but didn't succeed because of 7(PQRS). This solves it nice!

@Richard The i-19 26⍸i part adjusts for the extra letters S and Z, which are the 19th and 26th letters. I'm basically subtracting 1 from the indices of STUVWXY and 2 from the index of Z.

@rabbitgrowth yes, nice!

My solution also gives 10 for invalid characters, which is kind of consistent with the behaviour of ⍳ I guess?

Yup.

We should be able to make one using just ⍳ and arithmetic.

And no guard too?

by dividing by almost 3 and rounding up?

@rabbitgrowth Yes.

@Richard I was thinking of using the Iverson Bracket.

almost 3?

dividing by something like 2.8 which will round up one higher at the PQRS group

ah, was confused by how that'd affect the ABC group

I'm almost there.

👀

Got it! Now for a bit of cleanup…

@@

{9⌊(⌊(i-27<i)÷1+2×9<i)-2×10<i←¯1+⍵⍳⍨⎕D,' ',⎕A}

We can probably train this.

@Richard That's easy to fix: ¯2↓⎕A

not quite another instance of a last primative being nice

{8⌊⌈3.17÷⍨⎕A⍳⍵}¨

Yes, or that. Btw, no need for ¨

@Richard This is amazing: {(¯1+⎕D⍳⍵)⌊1+8⌊⌈3.17÷⍨⎕A⍳⍵}

yes , was working on that. Combining the solutions of everybody

Oof:

      n←(⎕D,⎕A)[?1e6⍴36]
      RS←{(¯1+⎕D⍳⍵)⌊1+8⌊⌈3.17÷⍨⎕A⍳⍵}
      Arith←{9⌊(⌊(i-27<i)÷1+2×9<i)-2×10<i←¯1+⍵⍳⍨⎕D,'  ',⎕A}
      AB←10|(1↓⎕D,'@@ADGJMPTW')∘⍸
      cmpx 'RS n' 'Arith n' 'AB n'
  RS n    → 1.8E¯1 |   0% ⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕
  Arith n → 1.7E¯1 |  -6% ⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕
  AB n    → 1.0E¯2 | -95% ⎕⎕

Congrats at @xpqz

Is the cat out of the bag?

stackoverflow.com/users/4432671/xpqz?tab=profile#main-content

Are we done with the quest?

was a really nice one, thanks!

Wouldn't have been enjoyable without everyone's contributions. Next week will be 2019-6: In the Center of It All.

{2 2 2 3 3 3 4 4 4 5 5 5 6 6 6 7 7 7 7 8 8 8 9 9 9 9 0 1 2 3 4 5 6 7 8 9[⍵⍳⍨⎕A,⎕D]}

:)

Oh, that's probably going to be the fastest.

  AB n → 1.1E¯2 |   0% ⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕
  RG n → 4.4E¯3 | -58% ⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕⎕

Arguably also the most clear (imho)

I'd put ⎕D before ⎕A, but yes.

I want '012ABC3DEF4GHI5JKL6MNO7PQRS8TUV9WXYZ'∘⍳⊇0 1 2 3 4 5 6 7 8 9 2 2 2 3 3 3 4 4 4 5 5 5 6 6 6 7 7 7 7 8 8 8 9 9 9 9⍨

ASCII order?

No, keypad order with digits before letters.

I see

@xpqz seems slightly faster too

Slightly‽

58%

oh, sorry was looking at another modification. Was trying to see how much the ⍨ would take

Not significant, since it only happens once.

@Richard Thank you. One week in. It'll be a blast.

@Adám what's the ⍨ for there?

The rightmost tine of a fork must be a function, but we want an array, so we make the array into a constant function.

@Adám in a class with a keyed property (e.g. xpqz.github.io/cultivations/AdvancedOO.html), can I call the keyed setter inside a method somehow using the obj[...]←... syntax?

@xpqz yes: :Property default Keyed K

Of course. Note to self: the property is keyed. It's not overloading [] on the object itself.

so sad nobody did the dumb yet fun solution

{A B C D E F G H I J K L M N O P Q R S T U V W X Y Z←2 2 2 3 3 3 4 4 4 5 5 5 6 6 6 7 7 7 7 8 8 8 9 9 9 9 ⋄ ∊⍎'⍬',∊'⍬',⍪⍵}

Hmm, given your ⎕A←... can't you just then do {⍎¨⍵}?