@Adám I thought about using 0⍴ to get the prototype but I assumed I'd have to apply that on each element so I didn't bother -- Didn't occur to me that you can just do it on the whole array and "dig the results out from that shape 0 array" Thanks for sharing that. It is almost ⎕ML independent if it weren't for the ⊃. At least this version lets you work in both ⎕ML levels without having to explicitly set it (Just swap between ↑ or ⊃). I think this is the winning solution :)

And do you have more detail on what you would plan for monadic ⍷? Are you just proposing defining monadic ⍷ as ⊃0⍴⊂ ?

I've decided to move from the GNU APL behaviour of ⊂ to ⊆ like Dyalog. That means I need to implement the Dyalog-style ⊂. Is there anyone willing to write some examples that I can use as test cases to ensure I implement it correctly? Like last time, something that captures the edge cass would really be helpful.

@JoshD ⎕ML and ⎕IO agnostic: ⊃⍬⍴0⍴⊂

@xpqz Version 18.1 session transcript:

      data←'racecar' 'racing car' 'car driver'  'version'
      ]get github.com/abrudz/dyalog-apl-extended/blob/master/…
#.DelDiaeresis
      0 DelDiaeresis (0 0∘⍉) ∘.{(¯3↑⍺)≡(3↑⍵)} ⍨data
0 0 1 0
0 0 1 0
0 0 0 1
0 0 0 0

@Adám do you know if the post-apl seeds conversations were recorded?

@rak1507 They were. I've suggested adding them to YouTube/dyalog.tv but it wasn't thought of as useful. If you want it done, I can probably get it done.

maybe it could be uploaded as an unlisted video and then you could share the link, if people think it doesn't warrant a 'proper' upload

I think it'd be useful as Aaron Hsu talked quite a bit about his compiler and stuff

I don't see how it would hurt to add to the AS21 list, which is very short.

well that would be great

@Adám I haven't yet had an opportunity to study under sufficiently.

@xpqz It is basically the generalisation of @.

@Adám did you get a chance to read the new paragraph I added to the dyablog?

dyablog, lol

@rak1507 I'm here all week.

@xpqz Yes, and after that, the whole thing did a few rounds of review. It is now ready to be published, just waiting for the designated time (2pm UK, next week, I think).

At that point, we'll also add the notebook to our collection on GH.

@Adám Always happy to make more tweaks if there are any more comments.

Nope, we're good. We've fixed a few commas and made the English ever so slightly more formal.

@Adám Having a good copy editor is essential for us non-English speakers.

(You probably intended to reply to another message, no?)

@Adám Umm. Yes. But on the topic, is Under making it to 18.1?

No, definitely not. A new primitive will only go into a .0 release ― but I don't even hope for it in 19.0.

20.0?

Maybe.

@RikedyP Hi.

@Adám Hello

Can you add it to dyalog.tv?

Did you upload the video?

Upload to? I put it on our network drive under Meetings.

Ah ok - it needs to be uploaded to youtube before Dyalog TV

are you sure everyone who participated is aware / consenting that the recording be uploaded?

Zoom notifies them when they join.

hmmmm

That's the only viable way, as people were coming and going all the time, so I'd have had to ask for permission again and again. FWIW, BAA also publishes their meetings without asking everyone every time.

Alright just trying to err on the side of caution

I think it's fine

'BAA also publishes their meetings' well...

@rak1507 Was literally about to comment that

lol

Could someone check this for me -- it's either a Dyalog bug, or a wetware bug. Here's a snippet from a Rosalind problem:

      'CA' 'AC' 'TA' 'AT' ∘.{1∊⍺⍷⍵} 'TAGACCA' 'GATTACA' 'ATACA'
1 1 1
1 1 1
1 1 1
0 1 1

So far, so good. Now let's add another substring to check:

'ATA' 'CA' 'AC' 'TA' 'AT' ∘.{1∊⍺⍷⍵} 'TAGACCA' 'GATTACA' 'ATACA'

Dyalog hangs, eating all CPU, only recourse is kill -9.

weird, it works for me

Btw, {1∊⍺⍷⍵} is (1∊⍷).

Works for me too.

works fine for me too (18.0.39712 linux)

reminds me of when ,\ didn't work for me

probably worth making sure you're on the latest version just in case

Works on TryAPL's Linux 18.0.39712.0 too.

So maybe a Mac/pre-pre-pre-18.1 thing. Let me try in vanilla 18.0

@xpqz I'd email support about it.

I promised @MortenKromberg to not email support on this version, as he'd get into trouble...

Yes, works fine in 18.0

But we'd better not have broken something on Mac-18.1

I'll email MK about it.

@MortenKromberg Sure thing -- if you're aware of it, I'll leave it be.

@xpqz You will receive an e-mail when ready, so you can replace that unofficial build with something that support will be comfortable with.

@xpqz NONCE ERROR for me.

jk, it works perfectly :P

I break things, clearly.

CMC: ⍷ (without using the built-in, of course).

Not a bad competition problem, btw.

{(≢⍵)↑(⍺∘≡)¨(≢⍺),/⍵}?

{(≢⍵)↑⍺∘≡¨(≢⍺),/⍵} but that doesn't work for 2d

wow

identical

lol

{(⍴⍵)↑⍺{⍺⍺≡⍵}⌺(⍴⍺)⊢⍵}

I think this should work

what's b?

oh, ⍵, oops

@rak1507 ⍺{⍺⍺≡⍵} → ⍺∘≡ but can fails if ⍺ has prototypical elements.

what on earth is a prototypical element

0 for numbers and spaces for chars

ugh

apl.wiki/Prototype

help.dyalog.com/18.0/#Language/Object%20Oriented%20Programming/… means I can't use dfns as class methods, correct?

@RGS Only as private methods, but you can wrap them in tradfns.

*as public class methods

Wrapping them in tradfns is too much work for what I wanted :P

Is (1∊⍷) special-case fast? Is it faster on ints than on chars?

don't think it's special cased, he was just pointing it out I think

@Adám are the Dyalog test cases public? Specifically I'd like to reuse whatever Dyalog uses for testing ⊂

@EliasMårtenson They are not, although I guess it can't hurt asking if you can have them. I was planning on answering your request, I just haven't had time.

@Adám Thanks. I was asking about the test cases in general, since I think it's likely I'm going to need other tests as well, and I feel somewhat embarrassed to ask for help all the time :-)

I once asked a closely related question on TopAnswers, and IIRC Marshall gave a bunch of links to public test cases

@Bubbler For APL implementations?

Yes, though there's no guarantee that they agree with Dyalog

or easy to use or whatever

There's a lot of freedom beyond the ISO APL

And even then, nobody is forcing you to comply with that spec either. I think GNU APL is actually the most compliant implementation.

APL2 and APLX intentionally ignore certain parts that GNU and Dyalog follow.

@Adám If I understand it, GNU APL was originally implemented very much based on the ISO spec. I doubt any other implementation did that.