(Similar to _←resultIsIrrelevant to indicate that _ doesn't matter!)

@dzaima I went through the definitions in order and tested them, so most of the errors were only a level or two down. Then just walk through the function by running lines manually to debug.

@Marshall Good to know, thanks! Happens to be what I wanted in this case, but very good to know.

I believe a found a bug:

 f←{
     _←{⎕←'Done'}⍬
 }

Works just fine! However, the following doesn't:

 f←{
     _←{⎕←'Done'
     }⍬
 }

@AviF.S. i believe that's just how dfns work unfortunately

This is an issue if you have a multiline expression that you want to write in the Stroustrup style:

 f←{
     _←{
         ⎕←'Done'
     }⍬
 }

@AviF.S. it indeed is an issue, and i've ran in it before

More generally, this occurs whenever you have a nested dfn (one or more lines), which doesn't return anything. And which you want to write with the Stroustrup brackets

it's the thing with dfns continuing to the next line if the previous ended with assignment. Here the last line is empty, and dyalog tells you so

@dzaima But it is, because I tested dfns written the same way with a trailing } and it does work as long the dfn returns something... anything!

 f←{
     _←{
         'Done'
     }⍬
     ⎕←'Ok, now I''m really done!'
 }

@AviF.S. but you are returning "something" in that last (empty) "line". It just happens to be "nothing" that you're returning, and you can't assign that "nothing" to "_"

(also a simpler way to test it is with a ⋄ - {_←{⎕←'hi' ⋄}⍬}⍬ vs removing the ⋄)

@dzaima Ah, so you're saying the same is true of the second function as well, but it never gets that far because the non-assignment line 'Done' terminates execution before it can get stuck on }?

@AviF.S. yeah

the behavior is equal to just {}⍬ - you can execute it in the REPL, but _←{}⍬ fails (this though also happens in dzaima/APL)

@dzaima Not sure it's working for me the way it should...

I should initialize `a` first?

@AviF.S. "a" is outside of the code-block :p ("{_←{⎕←'hi' ⋄}⍬}⍬" is the actual code-block)

(my dark theme for chat colors code blocks in a significantly different color)

Thanks! What did the a ⋄ - denote then, though, just out of curiosity?

@dzaima WDYM, that's a thing you can do?!?!

@AviF.S. "you can use a diamond instead of newlines, like this - …"

@dzaima I'm afraid I'm terribly confused... ⍨

@AviF.S. PPCGDark though i have many local modifications, chat is probably the only thing that works there

^ ^ ^ ^ ^

@dzaima My goodness!! So cool, am definitely downloading it now

@AviF.S. "a" was the english work, and "-" was an em dash

@AviF.S. (requires the browser extension Stylus or equivalent (stylus was the original extension, but added logging or something similar that was bad for privacy))

@dzaima Still isn't clicking but I suppose it isn't all that important. Just curious!

@dzaima *english word ಠ_ಠ

I imagine ⋄ being a newline is in your extension? And why an em-dash preceding the function

Also, just to be nitpicky, that's actually a hyphen :p (The one in APL) En & em get longer: - – —

@AviF.S. i specifically said em dash to signify that it's a part of the sentence, not APL code

@dzaima Haha, all clear :)

@dzaima (i didn't actually write an em dash in the sentence originally for obvious reasons)

@dzaima Of course! Just in the middle of trying to setup Stylus & friends :)

@dzaima "Stylish" was the original (now bad) extension, "Stylus" is the okay one, god i'm making horrible typos aren't i

@Marshall I thought that was Avi's request, wasn't it?

My goodness! It's a whole world at our doorstep! Thanks, never occurred to me there were extensions to mess around with sites' CSS!

@AviF.S. pretty much everything i use is in dark theme

What modifications did you add? I haven't noticed any differences yet...

@RGS a∘← will, if a is not found, create it in the global scope (global as in now "a" is a variable readable from the REPL regardless of how many nested levels of dfns you were in when calling a∘←1)

@AviF.S. well, for one, it should be dark