lynn

Earlier I thought you were implying x+.×y is allowed too, because it's a tradfn body. I guess the actual meta is you don't have to count the tradfn header if it's only a name f, but you do have to count it if it's like x f y.

And the examples marked ⍝ Invalid, … in this post being invalid is very sensible to me :+1:

It sounds like in a sense, Python doesn't have implicit output by default (but the REPL is an exception), whereas Dyalog does have implicit output by default (but shell scripts are an exception)

I did just find this: codegolf.meta.stackexchange.com/a/14232/3852

That's interesting, I don't have Dyalog installed right now so I can't play with it, but I wish ATO supported that

VALUE ERROR: Undefined name: f
 f
 ∧

If I put that in ATO, I get

I tried to guess at what you mean by "tradfn body".

Like Ruben said, if there's a flag to make Dyalog do implicit output, that's ok but it should probably be in the post

In that sense it seemed analogous to that last Python program, to me

It makes sense to write up a meta consensus for APL that doesn't invalidate hundreds of APL answers of course. But I was surprised when I ran Ruben's code on ATO and it read two lines of input but didn't output anything.

I tried to make a fair comparison, but of course Python is not APL :) Anyway, in Python, the first four are allowed, the next two are snippets (invalid), and the last is a full program that doesn't output anything except in a REPL, so we don't normally allow this either, except I guess you could submit it as "Python REPL, 15 bytes"?

print(int(input())**2)   # ⎕←⎕*2
def f(x):return x**2     # ∇ r←f x ⋄ r←x*2 ⋄ ∇
lambda x:x**2            # {⍵*2}
2 .__rpow__              # *∘2

return x**2              # r←x*2
x**2                     # x*2

int(input())**2          # ⎕*2

What's the syntax for giving the body a name?

I don't know if APL has special conventions on the site, but we don't allow function bodies return x+1 or REPL programs for x in range(9):x*x in Python, right?

@RubenVerg Yep, that's it

I think this ⍸ behavior might originally be a K thing, where it's called &. J's I. does it too. Jelly used to have it, and it was called O, but it got replaced by T which doesn't do the duplicating, and O is ord now

Yep, that's right :) Ooh, you can then make it a tacit function to get something really short.

OK, gotcha! I see a nice byte save. Monadic ⍸ has a behavior I wish Jelly or Vyxal had, but I don't think they have it: ⍸3 5 is 1 1 1 2 2 2 2 2.

I think it needs either ⎕← to be a full program or {} to be a function. It looks kinda like they had it in {} and copied only the } and then also didn't count it to their byte total.

codegolf.stackexchange.com/a/263321/3852 I think this is a snippet, right?

then it's described as "Returns the first truthy argument, otherwise the first falsy argument." but that is also not what it does

it's documented as a or b but implemented as process_element("rhs or lhs", 2),

so what is up with 2 3∨ being 3?

I submitted an answer with a backslash before the 1 at first

this makes me realize I don't really understand why comparing "f0" and 1 (for example) is fine

so I thought "wow too bad, I bet there isn't a 1-byte max(key) builtin or something" and looked for key in elements.md and found the arrow :)

TypeError: '>' not supported between instances of 'str' and 'Zero'

@lyxal It's funny: I wanted to use G (and zip the other way around), but that was giving me str/int comparison errors between "R" and 0

removing that took me to 17, but the 16 is pretty different

@Steffan I had a 3-element lambda with an unnecessary : in it

I'm getting the hang of this Vyxal stuff (I submitted a 16)

I'm just summing lists

ok I got a 15 using slightly better stack manipulation

annoyingly, on ATO I don't have to tuck the [] and I can save 2 bytes, maybe because I'm putting the input only on STDIN there, whereas weekgolf also supplies it as a command line argument or something?

I start at [] [a] and do something to it n times

not sure what that means

so I am missing a byte somewhere else

I see, ⌈⌊ṫ and Ė combined with wf (to do that thing I need) comes out to the same length

oh huh, well you try replacing the whole thing with Ė :)

I'm looking for a single-byte command that'll turn 12 into [12] and also [12] into [12]. is that a thing?

(I must admit I don't totally understand how it's helpful on desktop, either)

cool, thanks!

I created about half of these with a search program that generates and runs random 3-byte Jelly programs, it's pretty wild to watch the outputs

I just realized there are two nice solutions to [9, 0, 0, 9]. Mine isn't, um, tacky

@hyper-neutrino That's a great output, I should include it in a future puzzle set!

I think it will only need td { display: inline-block; }

I can't scroll to see the rest. but enabling word wrap seems better anyway

@hyper-neutrino it would be nice if I could use the keyboard on mobile (on your TIO page) :)