hey guys quick question, is there something equivalent to J's amend (}) in APL?

i.e. a non-destructive update

@ElectricCoffee Yeah, check out @

E.g. ⋄ 'X'@3⊢'Hello'

@Adám HeXlo

And ⋄ -@(3<⊢)3 1 4 1 5

@Adám 3 1 ¯4 1 ¯5

negate anything bigger than 3?

Yes.

You can mix-and-match either giving values as operand, or a monadic function, or give a left argument and a dyadic left operand.

The right operand is either an array of (leading) indices, or a function which is applied to the right argument, and returns a full mask.

interesting, can } do that too? Use functions and bit masks like that I mean

J isn't my strong side. I think you can do something with gerunds.

There's an obsolete form of } which uses a verb operand, but since } is an adverb, not a conjunction, it is of course more limited.

I see

thanks, a bunch! I gotta run

Take care!

Hi, @Adám. I am just curious. What is your favorite APL dialect? Dyalog

Depends on what is included in "dialect", but full disclosure: I work for Dyalog.

I like the core language syntax (and some glyph choices) of BQN more.

APL+ also has various features that I miss at Dyalog.

Especially when it comes to keywords/control structures, and the tracer.

I'd like to look more into the left-to-right APL derivatives.

Anything a k/J user might be missing out on?

Other than the beautiful glyphs? ;-)

Btw, what technology are you using for typing APL?

K has a very different array model. J less so.

@Adám which glyph choices?

@Adám array model?

so each array languages have different array models?

Yes, Dyalog APL, BQN, and J each have a unique array model.

Nested, Based, Flat: See apl.wiki/Array_model

And K's "not really an array model, just lists" is described there too.

@PyGamer0 ⥊∾≍⋈⊏⊑⊐⊒⍷⊔ and I find the 1-modifiers fiddly; need large font size to read. Not sure I quite like the 2-modifier scheme either. I had an alternative proposal where glyphs had 1 dot if they were 1-modifiers (e.g. ∔≐⩪) and 2 dots if they were 2-modifiers (e.g. ⍥⨬⸚).

@Adám I use J on a Macbook Air.

@Adám maybe different features? I assume the syntax is very similar.

@M4X_ So you have RIDE start when you want to use APL. Entry of APL glyphs should be fairly easy, assuming you have a backtick key, but you can choose another key as prefix key.

@M4X_ The J and APL syntaxes are very similar indeed. J has some more powerful features when it comes to function application. Dyalog has nicer built-in utils for "business use" like XML, JSON, CSV, CEF, PCRE, etc.

@Adám I didn't even know the difference between these. Thank you so much @Adám

"Array models"

Does J have dicts?

No. And its namespaces ("locales") can only really be used to store code, not as name-value pair collections.

@user17925981 Hej Brian BED. If you want to participate here, please email access@apl.chat

@Adám FWIW, I like Dyalog APL’s glyphs. Haven’t used J, but I read some of R. Hui and K. Iverson’s papers about J, and I find it difficult to associate in my mind functions to pairs of ASCII chars. OTH, I perceive APL’s unique glyphs as strongly associated with their respective functions.

Agreed.

ASCII-only, scriptable today, cross-platform native GUI, less syntactic anomalies.

also many more features useful for maths and stuff like the hypergeometric function

Ah, yes, forgot about that.

I'd consider open source a plus too

Another plus is not being splintered into many implementations like APL and K are. There's basically only one J.

Practical question. How can I make some kind of bookmark of some the comments, so I can find them back more easy?

@Richard Hover over the left side of a message, then click the downwards triangle, click permalink, bookmark that.

👍

Say I have a rank 3 numeric array. Each major cell is square. I'd like to set the main diagonal of each major cell to 0. Assigning through a rank 2 dyadic transpose doesn't seem to work:

      0 0⍉⍤2⊢d
┌→──────────┐
↓¯4 ¯4 ¯4 ¯4│
│¯1 ¯1 ¯1 ¯1│
│ 1  1  1  1│
│ 2  2  2  2│
└~──────────┘
      (0 0⍉⍤2⊢d)←0

Syntax error.

@xpqz Yeah ⍤ isn't allowed in selective assignment (not sure why, other than not having gotten around to implement it) but (0 1 1⍉d)←0

Didn't know selective assignment existed :o

Gotta know your dyadic transposes, my friend!

((0 0∘⍉⍤2)d)←0 gives a length error

@Razetime Is your d rank 3?

      d←3 3 3⍴3
      ((0 0∘⍉⍤2)d)←0
LENGTH ERROR
      ((0 0∘⍉⍤2)d)←0
       ∧

@Adám oh man, why didn't I think of (0 1 1⍉d)←0...

oh, just wrong ⎕IO

Wait, no, I don't think so.

@Adám That's actually quite a neat demo.

The syntax checker gets fooled if you parenthesise the selection function (this is how I implement my structural under model) but that probably means errors become odd.

i see.

Is there somewhere I can read about what functions dyalog can figure out the inverse of, and what the inverse is? So in J, i can always do u b. _1 where u is a verb, and it will show me the inverse (or the obverse if defined manually)

@pmikkelsen No, and the inverse doesn't always have an APL equivalent.

Oh yeah of course. But okay, it would just have been nice to see a table of some sort for the primitives and their inverse

In general, if there's a reasonable to compute inverse, it'll work. Although sometimes it doesn't. Moving some things around can make it work.

Okay thank you

Disappointingly, we don't find the inverse of ∘.f⍨ even though it is simply (‽) the function f⍨⍣¯1⊢⍉⍨⍴∘⍴⍴∘⍳2÷⍨⍴∘⍴

Simply :D

I should add that to APLcart…

One thing I always find amusing in J is to look at the inverse of 10&#. which i use often to decode a number into i'ts base 10 digits. The inverse is: ($&10@>:@(10&(<.@^.))@(1&>.)@(>./)@:|@, #: ]) which I can't even begin to understand

Ah yes, that's quite complicated. You need the logarithm, but negative numbers throw a spanner.

@user1136288 Hi taronish. If you want to participate here, please email access@apl.chat

@xpqz @xpqz The only way I've found to save as source files for git is:

@bwanab Does ]import work?

@Adám Yes, ]import works fine.

How about ]export?

@Adám I haven't tried ]export. I couldn't get it to work, but maybe I'm doing it wrong. Here's what I tried: ]LINK.Export #.wordle '/Users/<username>/src/test_export/' using my username. I get Command Execution Failed: VALUE ERROR

I've seen that reported before. New Link 3.0 doesn't have that issue, but as you know, upgrading on macOS is less than elegant.

@Adám I'm fine using the current scheme I've got since it works. I'll wait for good times to roll when they're ready.

@Adám working on it again... but too complicated for me. I understand the base 5 encoding but looking up what ⍠ does and ⎕NS doesn't make sense to me. 1200⌶ is for formatting date/time I figured out but 1200⌶⌶?

@Richard ⍠ is kind of a hack. It takes a function as left operand and a one or more name-value pairs as right operand, where the name-value pairs set options for the function. So this is providing the datetime formatter with a dictionary to replace the built-in English weekday names.

⎕NS creates NameSpaces (objects); I use it to build the dictionary.