Neat: x⌹y ←→ (⌹(⍉y)+.×y)+.×(⍉y)+.×x. The hand side is just (YⷮY)⁻¹YⷮX, which is equal to Y⁻¹X if Y⁻¹ exists, i.e. Y is square and non-singular. However, the former works for non-square matrices. Nifty.

Still not sure how this relates to least squares exactly.

Hrm. And ⎕DR tells us the storage width of an element and how it's interpreted. The final digit seems to just be an odd number, except for pointers. What's the design decision on that? Is ⎕DR exposing some internal type info which just happens to use the least-significant bit as a rapid "this is a pointer" check or something?

1010⌶ is cool. Kind of like set -x/set -o xtrace. Very nice.

Hrm. It appears that dyalogscript treats Character Output (⍞←blah), depending on if it's executed at the top level or from within a function. At the top level, separate lines with ⍞ get newlines between the output.

For example, a line with ⍞←'Neo-Passéist' ⋄ ⍞←' drudgery', would print out "Neo-Passéist drudgery" on a single line, but if you replace the Diamond with a newline, then the string also gets split between two lines.

Is this expected/intended behaviour?

Is there any particular reason that the Dyalog CLI doesn't respond to SIGWINCH? It's mildly surprising that one has to send FF (Ctrl-L) just to get it to pick up width changes and the like.

Also, apologies for spamming the channel. Just methodically going through the docs and kind of braindumped here.

@B.Wilson Sounds like a bug. I've emailed John, who's implementing all this.

@B.Wilson The terminal version doesn't really see development these days (we're always surprised when people prefer it over RIDE), but I've emailed support@dyalog.com about it.

@B.Wilson No worries, but you're always welcome to email support@ too.

@B.Wilson 0 1⊤9 11○⊢

Hi everyone :) I'm still new to APL and I'm having a hard time trying to come up with a tacit version of this expression (+\+⌿2↑p),⍥↓+\+⌿2↓p Tried to use forks but I guess I still didn't really understand how to go about it, any suggestions? I would love to get rid of all this duplication

@ViniciusCouto 2(↑,⍥(+\+⌿)↓)p

Also, welcome. Happy to have you here!

Oops, no there's missing an inner ⍥↓

2(↑,⍥↓⍥(+\+⌿)↓)p

@Adám Beautiful, thanks a bunch! I'll play around with it until I figure out how it works haha and thank you for the welcome!!

Do you know about this? ⎕←2(↑,⍥↓⍥(+\+⌿)↓)⊢

@Adám
┌───────┼───────┐
2 ┌─────┼─────┐ ⊢
  ↑ ┌───┼───┐ ↓
  ┌─┼─┐ ⍥ ┌─┴─┐
  , ⍥ ↓   \   ⌿
        ┌─┘ ┌─┘
        +   +

Kinda, I'm still getting used to interpreting this, but it's a great idea!

My first attempt at making it simpler was trying to get rid of the duplicate mention of p with ((+\+⌿2↑⊢),⍥↓(+\+⌿2↓⊢))p but I kept running into LENGTH ERRORs and I couldn't understand what was going on from the diagram alone (and I still don't)

The problem here is that both +\ and +⌿ are monadic, while those trains get parsed as +\+⌿(2↓⊢) i.e. +\ computes the left argument to +⌿ which is all wrong.

Oh I see the difference now that I've compared the diagrams, but I didn't quite understand how the usage of two monadic functions would cause this

@TryAPL Also, how can I get diagrams to display like this? Mine looks different with ]box on and without it it is even harder to follow

It isn't their nature of being monadic that's the problem, indeed, APL will happily try to use any function monadically or dyadically.

@ViniciusCouto ]box -t=tree

Are you aware of the parity structure of trains?

Thanks! Unfortunately no, I've only recently started learning about them. I'm taking a learn-as-you-go approach but perhaps I should look a bit more into it

@ViniciusCouto Look at apl.wiki/Train and the external resources at the bottom.

@Adám I will! Thank you very much

i have my apl ahk script setup to act just like tryapl, with the exception of some communative stuff. are you still interested in having it for your wiki? @Adám

Absolutely.

how would you like me to get it to you?

Maybe you can put it on GitHub or similar?

sure, i'll go ahead and do that

And then feel free to add an entry on the wiki.

Should I just add a link to it in the ASCII symbol composition section of the Typing glyphs page?

sounds right

Okay, I added it

I'm trying to use gnuapl, but when I print a vector, it wraps it even though it's not at the edge of my terminal. Has anyone else encountered this and knows how to fix it ?

@sushi try querying and, if successful, setting ⎕PW

great, thank you

are there apl interpreters written in apl?

Dyalog APL and ngn/apl are partially written in APL.

I've made a JIT transpiler.

very cool

does anyone use apl as mathematical notation on paper? I've been wanting to experiment with using APL on paper, but I don't understand enough of it yet

@sushi Yes. I've seen people do that, and on occasion done so myself. This isn't exactly paper, but maybe can inspire you.