@rak1507 I would love that in my uni, but there isn't many people interested in programming in general (I study physics) :c

Btw, is there any way to attend Dyalog '23 virtually?

@AmadoC No, but all talks are recorded and published on YouTube later in the year.

@Adám oooh, can I do this as well? My 1.5 (≤,⍥(+/)>){⍺(⍵⍵⍨ ⍺⍺ ⍵⍵)⍵}(⌊⍥≢⍴⊢) is 0r9bably enough

Hey, sorry to just barge in here randomly. I stumbled across an APL related question in Unix&Linux review queue, and I'm not really sure what to do with it. The question is about setting up RIDE. It needs some work, and for all I can tell it's on the wrong site. But I'm not really familiar with APL myself, so I don't honestly know how to direct them. It's a couple days old and they are actively replying in comments, so if anyone is feeling generous, could you please give it a look?

@BryKKan That's fine. Thank you. I'll have a look.

@RubenVerg Indeed. That was overly complex :-) but 20.5 overall, P7 was perfect. Phase 2 was generally of excellent code quality, but your vin failed many cases and your sortVersions could have been written better (and failed if a package name had a . in it).

@Adám Oh I see :c hopefully I'll attend someday

Ooh, this palindrome checker uses all the non-train combinators: ≡∘⌽⍨⍤∩⍥⎕C∘⎕A

@Adám I Just liked that more than a fork that repeated one of the functions but with left instead of right,

P7 was also an (accidental) nick from APLcart :)

what was an optimal 1.5 sol?

@RubenVerg A couple of people submitted +/0 1∘.=1 1⍉∘.> and similar, which imo is elegant, though it is somewhat wasteful, while +/1 0∘.=(⌊⍥≢↑⊣)≤⌊⍥≢↑⊢ is very clear.

ah, those are nice

kinda like (the expanded version of) mine more, tho

and everyone did the same thing in the one with find last, right?

unless someone wrote an explicit under operator which would be funny af

oh I'm really curious about 1.9 though, what did people do/what was the intended solution?

gotta say that last axis instead of first was really annoying but maybe it allowed for something I didn't think of

@RubenVerg No. Never underestimate people's ability to come with with… interesting solutions. The longest solution was 40 chars (of which 12 spaces). Quite a few people used normal ⍷ and rotated or shifted it appropriately.

@RubenVerg Nobody did, but someone did add a comment to the effect. (Not that comments are considered.)

@Adám I was about to comment "← under semantics dyalog gods please give us said operator"

removed it before submitting :)

@Adám "normal" meaning without reversing the array first?

@Adám hmm, can we get longest sol stats for all P1?

@RubenVerg There's never a single intended solution. {⊃,/⌽¨⍵⊂⍨2|⍳≢⍉⍵} is clean but nested. {⍵⌷⍤1⍨⊂⌽⍒⌈2÷⍨⍳⊃⌽⍴⍵} is a nice flat solution. {⍉⍵⌷⍨⊂{⍵⌊(⍳⍵)+⍵⍴1 ¯1}⊃⍴⍵}∘⍉ manages without even a grade so probably O(n), just like yours. Btw, why did you use ⌈⌿ instead of ⊢⌿?

@RubenVerg Yes.

@RubenVerg 34 40 45 155 81 139 58 148 89 238

@homeCorn Hi. I've seen both your RIDe repo issue, SE post, and this. I'll try to get someone who understands to look into it. Best thing might be to email support@dyalog.com so we can get in direct touch with you if necessary. Meanwhile, I'm pretty sure you can use the dark mode even in the browser, simply by chosing a dark scheme from the colour preferences.

@homeCorn Can you try RIDE_INIT=POLL:ip-of-host-machine:4502 instead of RIDE_INIT=SERVE:*:4502, and then in RIDE select a connection type of "Listen" (on port 4502), to see if you can make a connection the other way?

@homeCorn Also, can you confirm that some of your attempts have been made using an official release of RIDE, as opposed to ones that you built yourself?

@MortenKromberg I tried the reverse connection, and it seems that the connection is happening, but RIDE just stays blue. When I click on some menu option like Preferences, it give an error alert saying Unknown command: PRF

Finally I am finding the same issue wrt the docker image. I dont think this is an issue from the Interpreter side. Highly likely I am missing some dependency on the Ride side.

@homeCorn OK, thanks. We'll have a think and see whether we can suggest anything or reproduce the experiment. Can't promise anything about when we'll be able to do the latter.

@B.Wilson Your tasks 4 (from both phase 2 problems) failed on some cases.

@Adám Ohhh. I think I know what happened on Phase 1. I ended up reading the FASTA format specs in an attempt to get a broader solution. The stop codon is generally encoded as '*', but tho contest's Rosalind example uses a bespoke 'Stop' word instead. Blah...

Not sure what happened with Phase 2, though. I went way out of my way to support negatives in the starts and movements, since the problem spec was silent on the matter. The submitted solution has a TEST function that demonstrates everything working properly in the checked cases.

Was particularly proud of my really general partition and decently fast makeChange. Sad I didn't get far enough for anyone to even see them.

Anyway, thank you for going out of your way to check.

Ew... What?! Did my new solution not copy over??

       cmpx 'euro makeChange 200'
 5.5E¯2

Definitely not in the ms range though!! That's cool.

This was the total time for running a large collection of tests cases.

Yeah, 25ms is really nice.

       ⍴10∘⊥⍣¯1⊢0
 0

That doesn't look right, does it?

Why not? You need exactly zero digits to encode 0 (in any base).

      0≡10⊥10⊥⍣¯1⊢0
1

Cf.

      ⍬+.×⍬
0

⊥ is really just a fancy +.× which replaces the left argument with the reversed cumulative product (shifted so the rightmost element is 1, and the leftmost is lost).

Oh, it's an artifact of the identity on +. How odd. I definitely don't think of the empty string as encoding 0.

I think of b⊥⍣¯1⊢n as "digits of n in base b".

Guess that only works on the natural numbers.

@Adám This doesn't work with negatives, right?

       10 10 10+.ׯ999
 ¯29970
       100 10 1⊥¯999
 ¯110889
       10⊥⍣¯1⊢¯999
 0 0 1

Uh... the +.× left argument should be 100 10 1 just like the ⊥ example.