:62759939 no, but very curious how you managed in 8. I'was very proud of 11

  0(≥⌈⍥(+/)≤)t

@Richard ⌈/×+.×⌸×

and had a fight with +/ giving an error until i put it between ()

@Adám incredible! How can you ever think of something like that.

@Richard Well, using ⌸ was pretty obvious, since we need to count size of groups of elements that share a trait (negativeness, positiveness; i.e. the sign). Then we just need to sum them up and adjust the sign, which we can do using the sign too.

chapeau

@Adám Beautiful! I saw Key but failed to notice the inner product. Thanks for lighting up those neurons again.

How do I assign the result of a niladic ]usrcmd to a variable?

@xpqz ]r←usrcmd

I should add this to APLcart.

@Adám I'm sure I used to know this... I even rtfm'ed, but couldn't see it.

@xpqz try entering ]

:D

APL Quest?

no, my try on a leetcode problem Adam mentioned

@mathcat no, I meant to ask "What is an APL Quest?"

See apl.wiki/APL_Quest

ah, thanks

> Write a function that will reshape a given array into the smallest square matrix that will contain all the elements of the argument, padding with additional elements if necessary. The pad element should be 0 if the array is numeric and space ' ' if the array is character.

{s s⍴⍵↑⍨2*⍨s←⌈0.5*⍨≢⍵}

with three selfies :)

⍨

Basically the same approach: {n⍴⍵↑⍨×/n←2⍴⌈*∘0.5≢⍵}⍥,

And the obligatory tacitification: (⊢(⊢⍴⊣↑⍨×/⍤⊢)2⍴≢⌈⍤*0.5⍨)⍥,

Yeah, I don't think there are so many ways to attack this.

is the ⍥, necessary?

No.

for the ⍬ not is it?

Does seem superfluous: ⎕←≢0

I had the exact same thing as Richard. Also tried to tacitify it: ,⍨⍤f⍴⊢↑⍨2*⍨f←.5⌈⍤*⍨≢

@rabbitgrowth at least 4 ⍨'s

I think I like the dfn better though

Inline assignments in tacit functions are indeed a bit hairy.

@B.Wilson While I admire your ⍨s, ⊣↑⍨×/⍤⊢ is just ×/⍤⊢↑⊣

@Adám Hah! Can't believe I didn't see that.

Still dissatisfied with all the ⊣⊢ going on in that inner tacit function.

is (⊢⍴⊣) not the same as (⍴⍨)?

Yes, but that ⊣ is part of the expression on the right.

ah ok!

Not sure if this is better: ⊢(,⍨⍤⊢⍴↑⍨∘(×⍨)).5⌈⍤*⍨≢

In 20.0 you'll be able to write ↑⍨∘(×⍨) as ×⍨⍛↑⍨

Nice

Now I see why it's called reverse compose

Speaking of new operators, the overtake length computation has a nice Under expression, IIUC: {⌈⍢(*∘0.5)≢⍵}.

{↑s↑⍵⊂⍨l⍴1↑⍨s←⌈2*∘÷⍨l←≢⍵}

@B.Wilson Indeed, although it'd look even better as ⌈⍢√≢⍵

@rabbitgrowth Good, then this session wasn't in vain.

I love these sessions, I almost always learn something

⍢

With a bunch of potential language extensions: {⍵@(⍳≢⍵)⍢,⍵⍴∘⍷⍨⌈⍢√≢⍵}

⍢ is Under, √ is (square)root, ⍷ is type.

@Adám Hah! That is pretty. J's monadic defaults (%: in this case) do come in handy from time to time.

Somewhat curious what Aaron would have to say about the HCI of Root. I suspect there's pushback to introducing "frivilous" primitives and such?

I think I'd like {n⍴⍵↑⍨×/n←2⍴⌈√≢⍵} best. Or maybe {(⊢⍴×/↑⍵⍨),⍨⌈√≢⍵}

@B.Wilson HCI?

Human Computer Interface

I myself am ambivalent about it.

If we add it, we might also want to add its inverse, which is *⍨

Anyway, that's a discussion for another day.

Little sidestep. Why not introduce a circle with a dot in it ⊙ for outer product? And of course keep the ∘. It is the only function which uses two characters. Probably stupid question as newcomer / non programmer

@Richard … or some other symbol. I suggested it, but Roger Hui countered that we already have a regular way of writing outer product, using Rank.

@Adám but much less readable, at least for me

Yeah.

Now we can celebrate the half-point of the existing phase 1 problems. See you next week for 2018-1: Oh Say Can You See?

\○/

yes ,2018-1 is a nice one

\⍢/

Btw @Adám, do you have a solution for the LeetCode problem that's fewer than 8 characters? I'm really curious

@rabbitgrowth No, but @dzaima said he had.

If only / was not ambivalent, I wouldn't need the parens in ≢⍉⊢⌸(/⍨×)

Yes indeed

≢⍤⍉(/⌸×) seems to work, but isn't fewer than 8 characters

Sweet