The APL Orchard

I don't think it's superior, it just has my preferred environment out of the box. Nice interface, up arrow gets the last executed command, runs in my browser (bonus for me). RIDE just feels "claustrophobic"

Adám

Ah you want that REPL single-line input box?

Fawn Locke

Yes

Adám

You can have RIDE be a browser tab too, btw.

Fawn Locke

I'm aware, I should make that my default, actually.

Adám

12:55 AM

@FawnLocke Please log an issue to that effect.

Fawn Locke

Will do! That would be a nice option :)

Adám

While I don't think setting <BK> to UpArrow will work, you can make it Shift+Up or Alt+Up etc.

Fawn Locke

Btw, is it possible to have a second pane open in RIDE? For sending output there (instead of having one window for input & ouput). It'd possibly be nice to have an arbitrary amount of panes open on different threads too.

Bubbler

I can confirm that <BK>=Up doesn't work

Adám

@FawnLocke I think github.com/abrudz/out works with RIDE.

Fawn Locke

12:59 AM

I'll try it out later, thanks

Adám

Doesn't do per-thread, though. (Do you really wrote multi-threaded APL?)

Fawn Locke

Actually yes, but that's not why I want it. It'd be nice to have an arbitrary amount of (possibly) related problems in view at once - at least for me. I usually define functions in projects that don't rely on but extend from the ideas of others

Just a thought

Adám

@FawnLocke Heh, it decided to restart itself while I was looking at it.

Fawn Locke

:)

Adám

At least I got a stack indicating where things went sour.

Maciej Goszczycki

1:09 AM

Hello, I'm probably missing something very obvious, but is there no way to clear RIDE history without deleting a history file?

Adám

@MaciejGoszczycki The feature is missing. Please log an issue asking for this.

You're not alone in wanting this. Though I'm curious why people want to get rid of the log. I never do.

You can disable the log file, which means you get a clean log every time you start. For that, set LOG_FILE_INUSE=0

Maciej Goszczycki

I don't mind it too much, but for example today I messed up an outer product and got loads of output in the log. It's not exactly unusable but I can feel my CPU struggling a bit a more every time I hit return on a new expression.

Adám

Ah, that's an issue.

Maciej Goszczycki

I guess that would go in dyalog.dcfg?

Adám

Either there, or (assuming you're not using RIDE 4.3 on macOS) in the environment variable area of the start screen, or as an normal system environment variable.

Maciej Goszczycki

1:16 AM

haha, I do happen to be using RIDE 4.3 on macOS, so dcfg is it. Thanks

Fawn Locke

1:34 AM

Is there a simple way to convert `0 0 0 0 1 1 0 0 0 0 0 0 0 0 1 1 0 1 0 0 0 0 1 0 0 0 0 0` into `1 1 1 1 2 2 3 3 3 3 3 3 3 3 4 4 5 6 7 7 7 7 8 9 9 9 9 9` I can't figure it out.

Looked up "partition mask" and "partition" in the aplcart

Adám

So increment on every change?

Fawn Locke

Yeah

Adám

+\1,2≠/v

Fawn Locke

Ah, that makes sense...

Thanks!

Fmbalbuena

1:53 AM

@Adám i have a question

André Leria

@Fmbalbuena Why don't you just ask it?

Adám

@Fmbalbuena Next time, just ask me. I'll respond when I can.

Fmbalbuena

@Adám help

Adám

Do you want help with anything in particular, or just in general?

Fmbalbuena

a command for me?

Adám

1:56 AM

Was that your question?

Anyway, I'm having a hard time remembering what you've learned. Did you learn ? (question mark)?

Adám

2:12 AM

@FawnLocke I should go to bed. Now you made me implement a native solution for this under Windows:

∇OutPane
 :If 0=⎕NC'⎕SE.outPane.t'
     :With '⎕SE.outPane'⎕WC'Form' 'Output'('Dockable' 'Always')
         Dockable←'Always'
         :With 't'⎕WC'Edit' ''('Style' 'Multi')('Coord' 'Prop')('Size' 100 100)('VScroll' ¯1)('HScroll' ¯1)('Border' 0)
             P←{Text,←{0=⎕NC'⎕SE.Dyalog.Out.Filter':↓⎕FMT ⍵ ⋄ ⍵ ⎕SE.Dyalog.Out.Filter ⍬ 1}⍺ ⋄ SelText←2⍴⊂1,⍨≢Text}
         :EndWith
     :EndWith
 :EndIf
 ⎕SE.onSessionPrint←'⎕SE.outPane.t.P'
∇

Fawn Locke

Heh, I can't say I'm sorry, that's awesome!

Adám

Fawn Locke

Woohoo!

Adám

The Output pane can be docked anywhere, or undocked. I can put it on a different screen, etc.

Can also change colours independently of the input session.

Fawn Locke

Awesome, I'll try and get this working on wine.

Adám

2:20 AM

I doubt the interpreter will run under Wine.

Fawn Locke

I've got it working before

Didn't try the IDE

Adám

Oops. Tiny bug in my code.

∇OutPane
 :If 0=⎕NC'⎕SE.outPane.t'
     :With '⎕SE.outPane'⎕WC'Form' 'Output'('Dockable' 'Always')
         Dockable←'Always'
         :With 't'⎕WC'Edit' ''('Style' 'Multi')('Coord' 'Prop')('Size' 100 100)('VScroll' ¯1)('HScroll' ¯1)('Border' 0)
             P←{SelText⊢←2⍴⊂1,⍨≢Text⊣Text,←{0=⎕NC'⎕SE.Dyalog.Out.Filter':↓⎕FMT ⍵ ⋄ ⍵ ⎕SE.Dyalog.Out.Filter ⍬ 1}⍺}
         :EndWith
     :EndWith
 :EndIf
 ⎕SE.onSessionPrint←'⎕SE.outPane.t.P'
∇

Fmbalbuena

@Adám random. i guess

Adám

Setting custom Output pane colours ^.

Fmbalbuena

@Adám is that "?"?

Razetime

2:26 AM

now i shall stack alternating dark theme and light theme muhahahahaha

Fawn Locke

Noooo!

Adám

It'd be trivial to make it change colour randomly every time you print something…

Fawn Locke

oh dear

Fmbalbuena

@Razetime Ctrl+C Ctrl+V Enter and now everything changed to normal.

Razetime

@Adám Dyalog APL but it's a 2000s internet website

Adám

2:28 AM

@Fmbalbuena No no, no connection. ? does indeed generate random numbers. ?n generates numbers from ⍳n like dice.

?0 generates a random non-integer between 0 and 1.

And of course, it works on arrays. So to roll 3 dice: ⋄ ?6 6 6

TryAPL

@Adám 6 5 2

Adám

@Fmbalbuena Want a task?

⋄+⌿?6 6 6 6 6

@Fmbalbuena 19

@Adám yes

2:31 AM

Well, you almost solved it already: Write a function that takes a number, rolls that many 6-sided dice, and returns their sum.

⋄{+⌿?⍵⍴6}10

@Fmbalbuena 38

@Adám done

Very nice.

another?

2:33 AM

OK, so ?10 10 10 gives you three random numbers from 1–10, but they might be the same: ⋄ ?10 10 10

@Adám 5 6 6

why?

See? We got two 6s.

and how to fix?

Because each element of the argument to ? is a separate die.

Fmbalbuena

2:34 AM

random seed?

Adám

Sometimes it will happen, sometimes not.

⋄?10 10 10

@Fmbalbuena 6 8 8

⋄?10 10 10 ⍝

@Fmbalbuena 10 3 9

2:35 AM

⋄?10 10 10

TryAPL

@Fmbalbuena 1 2 1

Fmbalbuena

hmmm i don't understand.

Adám

If you want to select a subset of the indices, you use dyadic ? with the count of numbers you want as left argument: ⋄ 3?10

TryAPL

@Adám 3 8 2

Adám

@Fmbalbuena The numbers are random, and not connected in any way.

Fmbalbuena

2:37 AM

then

Adám

(I could keep editing my message; we'll never get any duplicates.)

Fmbalbuena

⋄f←?⍨6
⎕←f 7

TryAPL

@Fmbalbuena
┌───────────┬─┐
│2 4 6 5 1 3│7│
└───────────┴─┘

Fmbalbuena

Hmmmm

Adám

What's the problem?

Fmbalbuena

2:39 AM

i want ?⍨6 become {⍵?6}

Adám

Oh, you're using the wrong operator.

f←?∘6 is what you want.

Fmbalbuena

hmmm

@Adám ⍨ is swap, but why ∘?

Adám

∘ binds an argument to a function.

So ?∘6 is like {⍵?6}

Fmbalbuena

@Adám what does this mean?

Adám

And 2∘? is like {2?⍵}

Fmbalbuena

2:42 AM

why ?⍨6 won't work?

Adám

It does work, but it does something else. It is the same as 6?6

Fmbalbuena

oh, I mean ⍨?6?

Adám

That doesn't make sense.

⍨ is an operator. It needs an operand on its left.

Fmbalbuena

⋄8⍨?6

TryAPL

@Fmbalbuena 8

Adám

2:44 AM

This is similar to {8}?6

Fmbalbuena

@Adám ERROR

Adám

No, that's not an error.

Fmbalbuena

⋄{8}

TryAPL

@Fmbalbuena (1 trailing line)
{8}

␄

Adám

?6 generates a random number. Then {8} takes that random number and ignores it, instead returning 8.

Fmbalbuena

2:48 AM

@Adám CMC: visualize 4d array but

Adám

but…?

Fmbalbuena

gives

Adám

Oh, you want the 4D array to be shown as a 2D array of 2D arrays?

Fmbalbuena

yes

for making game.

Adám

⋄ my4Darray←2 2 2 2⍴⍳16 ⋄ ⊂⍤2⊢my4Darray

TryAPL

2:51 AM

@Adám
┌─────┬─────┐
│1 2  │5 6  │
│3 4  │7 8  │
├─────┼─────┤
│ 9 10│13 14│
│11 12│15 16│
└─────┴─────┘

Fmbalbuena

No

don't use

┌─────┬─────┐
│1 2  │5 6  │
│3 4  │7 8  │
├─────┼─────┤
│ 9 10│13 14│
│11 12│15 16│
└─────┴─────┘

use

1 2  5 6
3 4  7 8

9 10  13 14
11 12  15 16

Adám

Oh, I see.

⋄ my4Darray←2 2 2 2⍴⍳16 ⋄ ⍕⍤1⊂⍤2⊢my4Darray

TryAPL

@Adám
 1 2  5 6
 3 4  7 8

  9 10  13 14
 11 12  15 16

Adám

Like this?

Fmbalbuena

⋄ my4Darray←2 3 2 3⍴⍳16 ⋄ ⍕⍤1⊂⍤2⊢my4Darray

TryAPL

2:54 AM

@Fmbalbuena
 1 2 3   7  8  9  13 14 15
 4 5 6  10 11 12  16  1  2

 3 4 5   9 10 11  15 16 1
 6 7 8  12 13 14   2  3 4

Fmbalbuena

seems to work fine

and

⋄ my4Darray←2 3 2 3⍴'ab' 'cd' ⋄ ⍕⍤1⊂⍤2⊢my4Darray

TryAPL

@Fmbalbuena
  ab  cd  ab    ab  cd  ab    ab  cd  ab
  cd  ab  cd    cd  ab  cd    cd  ab  cd

  ab  cd  ab    ab  cd  ab    ab  cd  ab
  cd  ab  cd    cd  ab  cd    cd  ab  cd

Fmbalbuena

ok, i like

useful

now, bye

Fmbalbuena

3:45 AM

@Adám does TryAPL support graphical output?

Fawn Locke

No

Fmbalbuena

why?

Adám

Because we haven't gotten around to implementing it yet. It is planned, however.

Fmbalbuena

Who made TryAPL?

Adám

Look at the bottom of the HELP tab.

Fawn Locke

3:47 AM

Adám since you're here, is there a nice function that turns '**' '****' '***' into 2 (2 2) (2 1) fine if not, I'll just use cmat

Adám

I don't get the transformation.

Why are you copying that in here?

Fmbalbuena

Because i like

A Sudoku Solver in APL

►

Adám

Well, please don't clutter this room. Use the sandbox room for that.

Fawn Locke

I'm attempting to partition n amount of asterisks into as many pairs of * as will fit. (For parsing * from markdown)

Fmbalbuena

Hmmm

Adám

3:50 AM

@FawnLocke Ah, but shouldn't the first element be (,2) then?

Fawn Locke

Yes

André Leria

Is there a way of displaying large numbers (as in, over 10 digits) without scientific notation in the REPL?

Fmbalbuena

@Adám lang bar?

André Leria

(running it locally)

Adám

@AndréLeria Set ⎕PP to as many significant digits as you want.

Fawn Locke

3:51 AM

Yes, you can increase ⎕FR & ⎕PP to increase precision and floating point type.

André Leria

ty

Fawn Locke

(Other way around, sorry)

@Fmbalbuena What?

APL bar

What about it?

3:52 AM

abrudz.github.io/lb/apl

Adám

Is there a problem?

Fmbalbuena

@Adám no, only i need

Adám

@FawnLocke {2 1/⍨⌊2(÷⍨,|)≢⍵}¨

@Fmbalbuena What do you need?

Fmbalbuena

APL bar

Adám

wdym?

Fmbalbuena

3:54 AM

@Adám APL keyboard

Fawn Locke

Much appreciated

Adám

@Fmbalbuena Try to state what you need using a full sentence.

Fmbalbuena

@Adám ⋄{⍵⌿ ⍵/ ⍵ ⍵⍴ ⍳ ⍵*2}3 seems interesting to me, Can you explain?

TryAPL

@Fmbalbuena
1 1 1 2 2 2 3 3 3
1 1 1 2 2 2 3 3 3
1 1 1 2 2 2 3 3 3
4 4 4 5 5 5 6 6 6
4 4 4 5 5 5 6 6 6
4 4 4 5 5 5 6 6 6
7 7 7 8 8 8 9 9 9
7 7 7 8 8 8 9 9 9
7 7 7 8 8 8 9 9 9

Fmbalbuena

@Adám I need APL bar, I need to type glyphs.

Adám

3:56 AM

@FawnLocke Shorter, less efficient: (≢¨⊢⊂⍨1 0⍴⍨≢)¨

@Fmbalbuena OK, and the language bar you linked to doesn't work for you?

Fmbalbuena

@Adám This works, But I need link to APL bar.

Adám

But you already have the link!

Fmbalbuena

@Adám Can you explain {⍵⌿ ⍵/ ⍵ ⍵⍴ ⍳ ⍵*2}?

Adám

I can try.

⍵ ⍵⍴ ⍳ ⍵*2 you should already understand. Correct?

Fmbalbuena

@Adám ⍵ ⍵⍴ ⍳ ⍵*2 where ⍵ is 3 gives 3 3⍴9

Adám

4:00 AM

@FawnLocke Even shorter: (≢¨≠\⍤=⍨⊂⊢)¨

@Fmbalbuena Yes, very good.

And do you remember /?

Fmbalbuena

no

Adám

@Fmbalbuena Hold on, you're missing a ⍳. That was a typo, right?

Fawn Locke

Mmm I feel like these should've been more obvious to me, maybe I'm tired. Thanks for your time regardless

Fmbalbuena

^ Move to trash for some reason

1 2 3
4 5 6
7 8 9

@Adám "/"?

Adám

We used it to "filter" things: ⋄ 1 0 1 1 0/'hello'

TryAPL

4:04 AM

@Adám hll

Fmbalbuena

ok

hmm

⍵⌿⍵?

Adám

But actually, the numbers on the left are repetition counts: ⋄ 1 1 2 1/'helo'

TryAPL

@Adám hello

Fmbalbuena

i know

Adám

And a single element distributes: ⋄ 3/'abc'

TryAPL

4:05 AM

@Adám aaabbbccc

Adám

So 3/⍵ ⍵⍴ ⍳ ⍵*2 is that same 3 3⍴⍳9 but every column is triplicated.

Fmbalbuena

ok, not now.

André Leria

Is ⎕IO ← 0 a default of sorts? I realize it's convenient for using ⊃ for conditional lookups, but should I expect it everywhere, or only when noted?

Adám

No, quite the opposite. ⎕IO←1 is the default.

André Leria

Ok, good to know

Adám

4:12 AM

However, many prefer ⎕IO←0 and most use it at times (it is especially useful when dealing with number bases). If you don't know exactly what code will call what, it is wisest to localise ⎕IO (to whichever value) unless you're sure your code doesn't need it.

André Leria

It does make me more comfortable. As someone who's done some Lua, I can understand 1-indexing as "ok" instead of "blasphemy"

Adám

What makes you more comfortable, 0 or 1?

André Leria

Whatever the language is designed around

Fawn Locke

Heh...

Adám

I'd say it handles both equally well. Both settings have upsides and downsides.

André Leria

4:15 AM

I mean, humans tend to read stuff by index 1, so it's like a breath of fresh air sometimes. Saying "the zeroeth element" of something feels weird

And as I said, the time I toyed around with Lua taught me to adapt

Adám

Yes, and the last element has its index match the length.

André Leria

(I did trip around indexing in that language in the beginning, however)

Adám

However, in some situations, the best setting really depends on what exactly you're doing.

André Leria

I see

Adám

E.g. if you want to take up until and including the first occurrence of some element then ⎕IO←1 is best: ⋄ 'abracadabra'{(⍺⍳⍵)↑⍺}'c'

TryAPL

4:18 AM

@Adám abrac

André Leria

Hmmmmmmmmmmm I didn't realize this kinda behaviour could change

Adám

But if you want to trim then ⎕IO←0 is best: ⋄ ⎕IO←0 ⋄ 'abracadabra'{(⍺⍳⍵)↓⍺}'c'

TryAPL

@Adám cadabra

André Leria

Ok, then I do need to pay more attention than I initially thought

Adám

Anything that takes or returns indices is affected,

but anything that deals in offsets or lengths isn't.

If you use one index to do some other indexing, your code is likely ⎕IO-agnostic. This applies e.g. to the DNA/RNA translation code we had earlier.

André Leria

4:22 AM

I thought everything would be offset, so the results would never differ

Adám

But if you combine offsetting/lengths with indices, then ⎕IO matters.

André Leria

Guess I just hadn't seen such case

Fawn Locke

There's {∇⎕ML ⎕IO⊢←⎕IO-⍨?4 2}&⍬ if you ever wanna write ⎕ML & ⎕IO independent code

Adám

Haha.

Fawn Locke

:)

Adám

4:23 AM

Throw in ⎕WX and ⎕DIV too.

And why not ⎕CT/⎕DCT while you're at it.

Fawn Locke

Oh dear

André Leria

@FawnLocke Still haven't reached the trains chapter of the APL books, sorry

But I can kinda realize it's a joke

Fawn Locke

This isn't related to trains. On that note though, I'm really loving writing tacitly!

Maybe abusing it...

            ┌──────────┴───────────┐
            @                      ⍨
     ┌──────┴──────┐             ┌─┘
     ¨           ┌─┴──┐          ∘
   ┌─┘           ∘    ⍤      ┌───┴────┐
┌──┼──────┐     ┌┴┐ ┌─┴──┐   ⍨   ┌────┴─────┐
bi ∘  ┌───┼───┐ 2 | @    ¨ ┌─┘ ┌─┴─┐        ∘
  ┌┴┐ 2 ┌─┼─┐ ≢    ┌┴┐ ┌─┘ ⊆   \ ┌─┼───┐   ┌┴┐
  ⍨ ⌊   ⍨ , |      ⍤ ⊢ ∘     ┌─┘ 1 , ┌─┼─┐ * ⍷
┌─┘   ┌─┘         ┌┴┐ ┌┴┐    +       2 / ⊢
/     ÷           ⍳ / * ∊            ┌─┘
                  ┌─┘                ≠

André Leria

h o l y

Fawn Locke

These trains might be a tad too long

Adám

4:26 AM

@FawnLocke That's the flying spaghetti monster, right?

Fawn Locke

Yes, my favorite pokemon

André Leria

Kinda looks like a Cthulu implementation

Adám

@FawnLocke Well,, if you build them up piece by piece using names, only that the result is like that, then it is fine.

Fawn Locke

Yeah.

I have been working on this one-liner that handles markdown italics/bolding for a few hours now

emboldening? Words are tough

Adám

Handles? What does it do? Render to HTML?

Fawn Locke

4:29 AM

Yeah

Adám

I suppose it is for your enjoyment? There's already a MarkAPL library.

Fawn Locke

Yes, rewriting my static site generator without using markdown-it

Adám

Isn't it easiest with regex?

⋄ MD←'\b__(.+?)__\b' '\b_(.+?)_\b'⎕R'<b>\1</b>' '<i>\1</i>'⍣2
⋄ MD 'this _is_ __bold__ and ___both___ and _this __overlaps___'

TryAPL

@Adám this is bold and both and this overlaps

Fawn Locke

Probably, my APL skills are weak though, more non-trivial projects the better

Adám

4:37 AM

⋄ MD←'(`+)(.+?)\1' '<code>.+?</code>' '\b__(.+?)__\b' '\b_(.+?)_\b'⎕R'<code>\2</code>' '&' '<b>\1</b>' '<i>\1</i>'⍣2
⋄ MD 'this _is_ __bold__ and ___both___ and _this __overlaps___ but `this _is_ code`'

TryAPL

@Adám this is bold and both and this overlaps but <code>this _is_ code</code>

Adám

Etc. for each syntax.

Fawn Locke

5:05 AM

Wish we had hooks in a glyph

Adám

Uh, we do: ∘

Well, or "hook" — the dyadic hook. For the monadic hook, you need ∘ with ⍨

Fawn Locke

I should've been more specific

Didn't know beside was a dyadic hook, I normally think of the monadic form of hooks so that's probably why

Adám

You've not heard the latest Array Cast‽

Well, I propose f⍛g be backhook even when monadic.

Fawn Locke

I rarely listen to podcasts, I've only listened to 1 episode, sadly

Adám

I rather like that you have to write monadic hooks as f∘g⍨

I prefer that the duplication of the argument is explicit.

Fawn Locke

5:10 AM

That's an interesting way to look at it. I tend to avoid over-hookifying things because it gets ugly quickly

Adám

@FawnLocke Thank you so much for your report about TryAPL being slow. I managed to narrow it down to a bug in a member of dfns.dws and I've now added code to TryAPL so it patches that.

Fawn Locke

Yay!

Adám

Did you by any chance experiment with Under?

Fawn Locke

I've been using Under, yes

Adám

And you hit cases where trying to display your definition just hung, right?

Fawn Locke

5:19 AM

Yes, I have

Adám

Try again now.

Fawn Locke

Hmm, intended?

            U ← {⍺←{⍵ ⋄ ⍺⍺} ⋄ ⍵⍵⍣¯1⊢(⍵⍵ ⍺)⍺⍺(⍵⍵ ⍵)}
      U
SYNTAX ERROR
      U
             ∧

Adám

Yes.

Fawn Locke

Nice, may I ask what the issue was?

Adám

(Other than the ∧ being in the wrong position, but that's a separate issue.)

Fawn Locke

5:23 AM

Gotcha :)

Adám

The code was trying to identify the naming of a dfn, so it looked for ←{ but caught the inner ⍺←{. I amended it to look no further than after the first {

Fawn Locke

Ah, I see.

Adám

Fun thing that since I can't change the dfns.dws that's included in the off-the-shelf docker container, I instead "hot-patch" the code by getting the code representation, running a regex replace on that, and re-fixing it.

Fawn Locke

Heh, that's cool

milia

8:38 AM

@FawnLocke that's a nice tree you got there :)

Adám

8:49 AM

@FawnLocke Dynamic favicons now fixed

@FawnLocke And I've fixed the worst offenders.

Alex B

11:20 AM

the problem is not so much in 0-based systems (which computer scientists like) or 1-based systems (which mathematicians like). The real problem is with ⎕IO setting a mode. Your code behaves differently depending on the mode your system is in at the time, unless you reset the mode just before running your code. This aspect of APL is thus not purely functional, because there are contexts and side effects.
I suppose that at this point ⎕IO is legacy and we have to live with it.

Adám

@AlexB Absolutely. In hindsight, everyone agrees that a single suboptimal index origin would have been much better than giving the choice.

Alex B

@Adám ah OK, so apparently this issue has been debated at length in the APL community. Being new, I am still trying to figure things out.

Kamila Szewczyk

12:07 PM

@AlexB are mathematicians really that much into 1-indexing

Adám

I think it depends on their domain, but mathematically inclined languages tend to use 1, e.g. Fortran, MatLab, Mathmematica, Julia. Wikipedia says:

> In recent years this trait has also been observed among many pure mathematicians, where many constructions are defined to be numbered from 0.

which sounds like it is a recent development, possibly influenced by computing.

Jeff Zeitlin

12:46 PM

I suspect that zero-origin became more common when languages that implemented array indexing as exactly equivalent to pointer math came into wider use. When I first got involved in programming, C was not yet the "hot" language, and 1-origin was essentially assumed. I was exposed to APL before I was exposed to C, and ⎕IO was a novelty, one which I thought was interesting, but wasn't sure I saw the use for. I generally (and still do) code with ⎕IO being assumed as 1.

Alex B

1:20 PM

@KamilaSzewczyk I don't know (because I am an engineer, not a mathematician) but that is my perception. Search for the word "zero" on this page: en.wikipedia.org/wiki/Natural_number

Algol and Pascal also have 1-based arrays, though some Pascal dialects make that configurable. See also: en.wikipedia.org/wiki/…

According to that Wikipedia page, Edsger Dijkstra is responsible for pushing languages to be zero-based.
https://www.cs.utexas.edu/users/EWD/transcriptions/EWD08xx/EWD831.html

Adám

While I'm nobody, I'm not convinced by Dijkstra on that.

Fawn Locke

@Adám Awesome, works great!

Alex B

@Adám I agree that Dijkstra's argument is not compelling. By my personal experience, C had a big influence on the zero-based approach. C was meant as a very efficient high-level (by the 70's standards) system programming language. Computers had to be very efficient in using memory, so arrays were just pointers and the PDP architecture had offsets and increments built into the hardware.

Adám

@FawnLocke Actually, I enlisted @Leo to fix all of them :-)

Alex B

So the natural choice was to start from zero, so the initial element of an array would be at the same address as the array base address.

Fawn Locke

1:34 PM

Heh

I agree with Alex. I don't think there are any strong arguments one way or another, nowadays it's just convenient to implement IO←0. ⎕DIV is much worse imo despite being less "harmful"

Alex B

@Adám how will the logo be selected? Are votes being solicited?

Adám

@AlexB I'm going to set up a website where people can rank them, and then we'll find a few finalists using Ranked Choice Voting. If there's a clear non-apple winner, we'll use that. If there's a clear apple-based winner, we'll use that until we get sued by Apple, bask in the publicity, then fall back to the next winning non-apple logo. Depending on outcome, we may need a second round to decide on variants of the winner.

Alex B

@FawnLocke actually, I wasn't advocating one way or another for APL. It seems that ⎕IO←1 is more idiomatic, from what I understand.

Adám

I wouldn't even claim that. Very relevant reading.

Fawn Locke

Yeah, I simply meant I agree with why ⎕IO←0 is the defacto and stuck on my own opinion :)

Adám

1:42 PM

For a lot of everyday practical problems, ⎕IO←1 is clearly superior, but for many things, especially those that involve (implicit) modulus arithmetic, ⎕IO←0 is a must.

Alex B

@Adám my 2 cents on "apple" in the logo: I doubt that Apple Corp can do anything about this because APL's usage of the apple in the logo is probably very old. If anything, I don't know if this is good for APL, because people will associate the language with a company and/or get confused.

Adám

E.g. it is convenient to get prefixes as (⍳≢a)↑¨⊂a using ⎕IO←1

Fawn Locke

Yeah, I prefer ⎕IO←1

Adám

@AlexB I tried finding the earliest reference to an APL-apple pun and it is unclear what came first, APL puns or Apple's logo.

Fawn Locke

I could definitely see Apple taking action

Adám

1:44 PM

Yeah, Apple is definitely in the programming language business, so…

Alex B

@Adám a trademark has to be in continuous use.

It may be more prudent to use a logo without an apple in it. It may also be better long-term for APL.

Adám

Apple has been using an apple motif since 1976.

Yeah, while widespread, the apple pun isn't even very good. APL isn't pronounced like that, and it has no connection to any characteristic of APL.

Fawn Locke

Trains output is broken on TryAPL again, btw

Only for certain trains though.... hmm

Oh, it's only when a train has an array tine?

Adám

Example?

Fawn Locke

((Tag⍨/¨)@⍸⍤Ca⊢(2,⍥⊂/⊢)).s
⊣ ∊ ⊣ ⊣@ }∘= ⍸⍤ ~∘ +\ -⌿ {} ∘.= ⊢ ⌷⍤ ¯1 99 ⊢ ,∘ {} , /> ,⍨ ¯1 ∘↓ ⍨ <{}> ∊ ⊣ ⊣@ }∘= ⍸⍤ ~∘ +\ -⌿ {} ∘.= ⊢ ⌷⍤ ¯1 99 ⊢ ⍨ ⊢ ⍥⊂ ⍨ / ¨ @⍸ ⍤ *∘∊ ¨ ⊢ 2 ,⍥⊂ / ⊢ . <h2> *** About me * help <h2/>

Adám

1:53 PM

What is Tag and Ca and s?

Fawn Locke

I ← ⌷⍤¯1 99
U ← {⍺←{⍵ ⋄ ⍺⍺} ⋄ ⍵⍵⍣¯1⊢(⍵⍵ ⍺)⍺⍺(⍵⍵ ⍵)}
Title ← ('h',⍕∘≢)U⊃@(<\≠)' '∘(≤\⍤≠⊆⊢)
Inset ← ∊⊣⊣@('}'∘=)⍸⍤(~∘(+\-⌿)'{}'∘.=⊢)I⊢
Tag ← (⊣Inset(,∘('{}','/>',⍨¯1∘↓)⍨'<{}>'Inset⍨⊢))⍥⊂
Ca ← '*'∘∊¨

Accidentally included the . before the s and that was the output, not sure if it's expected behaviour

Adám

I'll try it.

@FawnLocke But what was s?

Fawn Locke

1 second

Adám

((Tag⍨/¨)@⍸⍤Ca⊢(2,⍥⊂/⊢)) displays just "fine" (it is a giant monster) for me.

Fawn Locke

 P_Asterisk ← ⍸⍤(2|⍸@⊢∨~)⍤Ca I⊢)⍤((+\1,2≠/⊢)⍤Ca ⊆⊢
s ← P_Asterisk ⊃Tag⍨/ Title '## ***About me*** *help*'

Oops P_Asterisk ← (⍸⍤(2|⍸@⊢∨~)⍤Ca I⊢)⍤((+\1,2≠/⊢)⍤Ca ⊆⊢) You probably caught that already

Richard

2:03 PM

Nice @Adám!
Should also something mentioned about identation (using tabs or spaces) and the position of { and } on seperate lines or not. And the use of whitelines? Or is it a custom around APL programmers to leave it to the programmer?

Adám

@FawnLocke OK, I can repro it now with a much smaller input: +.'a' 'de' 'g' '' — I'm not very concerned though, as that is an invalid construct.

Fawn Locke

Yeah, I was curious why that didn't error

Adám

I'm actually surprised it doesn't. I was very certain it would.

Who would have thunk: ⋄ +.'a'/'b'

TryAPL

@Adám b

Fawn Locke

Indeed

Transcript for