The APL Orchard

6:53 AM

@JosephAdams OK, incorporated. Thanks a bunch!

Any takers for such important operators as Scan, Inner product, and Outer product?

7:49 AM

I'm not sure if this is the appropriate place to ask, but I get "VALUE ERROR: Undefined name: ⎕SE.UCMD" whenever I type a ] command in Dyalog. What can I do about it?

RikedyP

@ArpitSaxena Sounds like you don't have a session file loaded - let me see how you get that back

@ArpitSaxena How did you start Dyalog?

RikedyP

Also how did you obtain Dyalog and how do you start it and what platform are you on?

:-)

RikedyP

Okay @Adam is here lol

7:52 AM

I installed this aur.archlinux.org/packages/dyalog-bin

Independent of everything, )load buildse and then executing BUILD_SESSION 'UK' should fix the problem.

Yeah that worked. Thank you :D

Also is there a way to make an array of functions? f f f just makes a train (not really familiar just read the aplwiki page on tacit programming)

@ArpitSaxena Kind of: aplcart.info?q=array of functions

@Adám Oo thanks. That's a nice website

1 hour later…

9:05 AM

@Adám any updates on this?

@rak1507 Yes. The devs at Kattis are too busy, but they are attempting to hire a contractor to do the work, so it should be happening.

oh cool

do you know how they're planning on getting around the lack of scripting support? multiline dfns/large inputs?

@rak1507 Simples: 18.1

so it'll only be once 18.1 is out or are they getting early access

@Adám why are you writing Portuguese now?

9:08 AM

@rak1507 Early (as in already so) access.

@RGS Por que não?

has it been tested if ⎕ and ⍞ both work with large inputs? I assume it would, but TIO doesn't so it might not do I guess

@rak1507 Good question. I'll forward that.

Working for Dyalog, I write/speak/listen to English most of the day, not Pt. The other day, I went out for dinner and as I was ordering, I had to reference the menu because I had forgotten the name of the thing I wanted.

@JeffZeitlin We'll be publishing some details of this very soon.

As I was reading the name of the dish, the menu had an English expression there, so I read it, and just completed the order in English

Without realising the switch I had made :P

I'm being brainwashed.

9:13 AM

@synthmeat Hey, welcome back. It has been a while!

wow 954 days

@rak1507 For things like echo input | hello.apl and cat input.txt | hello.apl there'll be no limit.

Only interactive input will retain the limit.

great, was thinking if it just pipes the apl to the interpreter somehow it might get a limit error like TIO does

@Adám why is there a difference?

@rak1507 TIO's "script support" is a total hack. Not the real deal.

Is the real deal having a limit on interactive input?

9:22 AM

@rak1507 Not sure. I think it has something to do with having to reserve an input buffer. Should I ask for the technical details, or for John to come here?

@rak1507 No, I mean that it isn't really using script support (as it didn't exist in 17.1).

@Adám fair enough

xpqz

9:53 AM

This was quite interesting, and mentions APL and friends: canonical.org/~kragen/memory-models

1 hour later…

LdBeth

11:12 AM

funny enough, I think I find the similar "replace one function to a different one still passed all tests" situation in the dyalogcompetation problem

that probably means there's a theorem I can use to simplify my current answer

Joseph Adams

11:53 AM

@LdBeth note that I've found that the test cases aren't actually exhaustive for the set tasks...as in I've submitted solutions that were obviously not up to spec but still passed all tests

I suppose they're just guidance to reassure you that you're a least in the right ballpark

LdBeth

12:20 PM

@JosephAdams I'm now almost sure that it's the same problem we're taking about ;)

yeah, I find a solution that I can construct a counter example passed all the test

1:11 PM

It is actually easy to write solutions to the problems that pass the few test cases in there, but that aren't actually correct.

So it is advisable that you spend some time figuring out hard test cases/edge cases and then throwing them at your solutions.

damn so you mean my submission {⍵≡testcase1: answer1 ⋄ ⍵≡testcase2: answer2 ...} won't get full marks? :(

@rak1507 That should fail anyway, as the system has randomised tests.

not for phase 2

Phase 2 doesn't run any automated tests.

I know

1:27 PM

Lots of array programming goodness in the previews for the upcoming ARRAY 2021 conference.

the neural network one mentions APL but says it wasn't performant enough, shame

2:14 PM

@rak1507 I'd give it half the marks

I suppose now you're an employee you might be marking them!

@rak1507 If I'm marking like this, probably I won't be let anywhere near the solutions :P

2:35 PM

Announcement: Webinar on error trapping techniques in 25 mins.

1 hour later…

Q: In APL how to turn a string into a one cell vector

3:52 PM

tryapl on HN news.ycombinator.com/item?id=27460887

user

4:46 PM

in The Nineteenth Byte, 7 mins ago, by DJMcMayhem

@cairdcoinheringaahing What do you call a group of APL superheros? The Dyaleague

5 hours later…

Feeds

9:25 PM

So, I have a function f that takes a string as input. I want to create a function g that maps f to a vector of strings. I.e. g 'Hello' 'world' should yield (f 'Hello')(f 'world') Here's what I did: g ← {f¨⍵} And this works just fine for the example above. However, it doesn't work when the right...

apl

@Feeds Yet another need for the Depth operator: f⍥1 Try it online!

nice

⊆ is fine though

@rak1507 Nope, see my second comment.

oh, good point, doable with ⊆⍣¯1 though (which is a fun inverse I just found out about)

it's really f¨ under ⊆

@rak1507 Won't work if f changes the structure.

@rak1507 Yes, but again an ad-hoc solution. ⍥1 is the generalised form.

9:37 PM

it'd be nice to know what their specific f function was because otherwise finding a 'proper' solution seems pretty hard

is depth planned or is it a far future concept?

oh well looks like ⊆ was what they needed after all

@rak1507 I hope it is in the fairly near-future pipeline. I think I've convinced the necessary people. J has it, after all, and the symbol to be used is obvious.

why is ⍥ obvious?

Because it is very analogous to ⍤k and f⍤g is already related to f⍥g, so…

depth being similar to rank and the symbols already being used for related things doesn't necessarily make it obvious ⍥ is the right symbol for depth to me, but it makes sense a bit

@rak1507 {1=≡⍵:f⍵⋄∇¨⍵} or {⍵⍵≥|≡⍵:⍺⍺⍵⋄∇¨⍵} for a general Depth operator (limited to monadic application and positive ⍵⍵).

2 hours later…

11:39 PM

Huh, I just noticed that John Scholes mused that k¨ could mean k⍨¨.

seems a bit strange but slightly reasonable

Wezl

@Adám just spent half a minute trying to find that because of the ambiguity of ⍨ :/

@rak1507 My original proposal was for all of ⍨ and ¨ and ⍤k and ⍥k to do that.

it doesn't seem like a bad idea but it also doesn't seem particularly useful either, so I'm on the fence

… Just like @ does.

BQN does it in general too: Any array operand acts like the corresponding constant function, and any function operand is applied to the argument(s) to calculate the corresponding array operand.

11:46 PM

I quite like k's thing of having arrays basically act like indexing functions

@rak1507 Quite useful for @ and thus for ⍢ (e.g. 'x'⍢⊃ is like 'x'@⎕IO or 'x'@(⍴↑1⍨)) and if done for ⍢ then it should apply to ⍥ too, and if to ⍥ then to ⍤ and if ⍤ then ¨. QED.