⎕←'Good ',', APL!',⍨'morning' 'afternoon' 'evening' 'night'⊃⍨1+⌊4×1|(÷¯6)+1⎕DT'I'

@B.Wilson Good morning, APL!

Now I've started typing )off into bash shells by accident instead of exit :D

⎕←⎕NL¯2

@B.Wilson Response looks like a 3-by-0 matrix.

Ha! smtutor workspace is under the default WSPATH. Learning actual ⎕SM sounds fun.

This tutor workspace has a really intense color scheme :P

@TryAPL Looking for APLVersion. I vaguely remember seeing some function that would let you peek at/use definitions from workspaces without needing to load the workspace.

Interesting. So )ed opens up differently decorated windows (terminology?) depending on the object's type. In the terminal, at least, both the window border color and DISPLAY-like border tags seem to change depending on item type.

@B.Wilson hahaha

@B.Wilson in this case just a containing tradfn. in general, the outermost function when nesting functions.

@Adám Ah, okay. Simple enough. I am aware that the mix of dynamic and lexical scope has some funny edges to it. Does the concept of capsule have something to do with this?

Not really. Dfns calling dfns use lexical scope. Anything else is dynamic,.

Why is the term capsule even a thing, though? It's not a standard concept as far I as I am aware.

@B.Wilson It is relevant for how far out ⎕SIGNAL will signal, and when tracing into nested functions, there's an option to exit the capsule because if you make changes to code that's on the stack, the current capsule isn't updated.

So, if you're deep in some stack when a ⎕SIGNAL fired, there's some command to wind back the stack down past the function that happens to also form the highest lexical boundary?

(I notice that I am confused)

@B.Wilson No, the two things are unrelated. ⎕SIGNAL will exit the current capsule and then throw. Unrelatedly, if you hit an error or have traced into a (possibly recursive) state, you can cut out of the capsule by using the <CA> shortcut.

@Adám What does it mean to "exit the current capsule"? My (anemic) mental ontology here is just a call stack and associated state.

@B.Wilson Let's say f calls g which calls h and you're in h. Popping one level of stack (with a signal, or manually) takes you to g. Let's say h is defined in g but g is not defined in f. Exiting the capsule will take you to f. Now lets say g is defined in f. Exiting the capsule will take you all the way out.

Okay, I just defined f←{g⍬} ⋄ g←{h←{∘∘∘} ⋄ h⍬} (but multi-line, so tracing works, and running f⍬ we track into h with ⎕SI≡,¨'hf'.

So the reason g is missing is because ⎕SIGNAL exits the capsule before trapping?

And nesting the definitions of h in g in f, leaves just ⎕SI≡,⊂,'h'.

I didn't realize one would lose call stack information like that.

> ... if you make changes to code that's on the stack

So the above is the main reason for partially unwinding like this?

I guess it makes sense that trying to edit the definition of a function that's already on the stack might be problematic.

Ah, looks like <CA> isn't bound by default.

@B.Wilson Well, there's no ⎕SIGNAL and ⎕SIGNAL doesn't behave like normal errors. The latter throw where they happen, while ⎕SIGNAL cuts out of the capsule first. What you might be seeing is that one-liner dfns cannot be suspended.

CMP: Which core language features would you want Dyalog to prioritise after adding ⊇ Select, ⍛ Behind, and ⍥ Depth?

Structural under gets my vote. I found a use for 'behind' in the wild the other day.

@Adám The actual dfns definitions I ran were multi-line.

@Adám Hrm. I want to say Under, but that's likely just availability heuristics operating. What are some options currently thrown on the table?

Ah! Remembered this: github.com/abrudz/dyalog-apl-extended

@B.Wilson I have a list of 50 items… Maybe I'll write it up as short points rather than long explanations for each.

Am all ears!

Apart from new glyphs, I'd like to see some QoL things, like (somehow) allowing for one free dimension in reshape (like in a recent Dyalog comp, or like k).

There would seem to be available design space for this.

A way for stencil to not do its overhangs would be nice.

This was a nice application of behind, to take cells from the front and back: ↑,-⍛↑

hey y'all's do you know what the best career for someone with a degree in mathematics and physics which involves array programming? asking for a friend (i.e. myself).

@xpqz A third row in the right operand specifying first top left offset (default: ⌊2÷⍨1-⍨⊣⌿⍵⍵-1)

@11Kilobytes I don't know of anything in particular involving physics, though I do know of jobs involving chemistry and medicine. Mathematics is always good. You could look at this list or start with an internship at Dyalog to meet our customers, many of which are looking for young people to pick up the mantle.

I am trying to get a handle on removing parenthesis and rewriting with commute but I am having a hard time with it. Could use some tips and things to look for.

@Jeremygee (F G H x) K y → y K⍨ F G H x

@Jeremygee Noice! What about an approach like this: 10ׯ1+2*0.1?

@B.Wilson Yeah, I think it was just the principle. Otherwise just compute the value.

@B.Wilson oh! yes that makes sense. Now I can build a function that computes it in sets of 10. I like it.

actually I need to edit it a little. but thanks for the idea!

@Adám Sure. Was trying to show ¯1+ and X× patterns for replacing - and ÷. At least, those are little tools that I remember discovering when trying to avoid parentheses.

@B.Wilson This is great. Saving that as well. Always get into trouble with subtraction

@gwa000 )load smtutor

@Adám i get "ws not found". i'm at version 18.2.45405 and it is installed on my D drive. i don't see a file by that name in the install folder.

Which OS? D: drive, that'd be Windows. I believe it isn't shipped on Windows.

@Adám yes, i am on Windows so it looks like your are right.

I just tried running it under Windows, and while it "runs" things don't look right.