The APL Orchard

12:00 AM

also, am I losing out much by using RIDE on OS X vs the windows software?

same interpreter i know but there seem to be some differences in the IDE

OK, weird. Winkey wasn't working, then all of a sudden it is. Now... is there a way to change from Winkey to ... oh, AltGr would be good...?

Winkey-L puts up QUAD and also immediately locks the computer :)

Adám

12:17 AM

For basic usage, you're not missing much. The Windows IDE does have some nice integrations missing in RIDE, e.g.:
∘ Workspace Explorer is much more powerful, allowing editing and moving items around
∘ There is a versatile search functionality
∘ Menu items to save/load session
∘ Menu items to save/load/clear log
∘ Much more customisable (undock or move language bar and other elements, font, )
∘ Copy and paste data with structure (e.g. spreadsheet ←→ matrix)
∘ GUI for many additional settings
On the other hand, RIDE has some modern features, e.g.:

@JeffZeitlin Well known clash with Linux in VM under Windows. I highly recommend using AltGr (on Windows too), especially for English layouts. I simply use my AltGr layout on Windows and it somehow propagates into my Ubuntu VM. I think that forum thread addresses switching to AltGr directly in Linux, but if it doesn't, let me know and I'll find out from someone.

Jeff Zeitlin

Ah! Got it! It took some groping around since I don't seem to be using exactly any of the interfaces discussed in that thread, but I tried what seemed to be the equivalent of one of the suggestions/instructions, and it worked.

Adám

Nice. Maybe you could document how to do it under Mint in a reply on that thread?

Jeff Zeitlin

On Windows, I use the default of Ctrl for the APL characters, with the input method set to the Dyalog US keyboard layout.

As far as documenting it, I couldn't tell you what interface I'm using - it's a windows-y interface, but it's not KDE, and it's not ttbomk Gnome.

Adám

And it doesn't bother you that you can't use Ctrl+S to Save, Ctrl+F to Find, Ctrl+A to select All, etc. etc.?

Jeff Zeitlin

Nope. It actually never occurred to me that I might want those. But then, remember that I'm the guy that uses the del editor for tradfns. :)

Adám

12:31 AM

Oh, right ⍨

Jeff Zeitlin

While I'm thinking of it... To the best of your knowledge, are there any browsers that inherently block Javascript from writing data to the computer? I'm thinking of the possibility of hacking on that 5110 emulator so that you can save (and load) the disk images [so that you can write your own programs in BASIC or APL and save them and keep them; right now, the disk images are only in the browser's memory].

Adám

Certainly not any mainstream browser. These days, you only have to worry about Chromium, Firefox, and Safari.

Jeff Zeitlin

Well... And EdgeHTML, for a while longer. Chromium Edge hasn't been fully deployed yet.

But EdgeHTML doesn't lock down, either.

Evan

4:01 AM

Hmm... is there a reason why when I call ]LINK.create # /my/dir, I wind up with Linked: # → ...... the arrow going one way?

Adám

5:47 AM

@Evan Yes, we don't have file system watching other than on Windows due to it using the .NET Framework. Basic support for .NET Core is coming in the next version.

RGS

6:52 PM

There's something that is bothering me. I have a regular vector (in my particular case, ⎕D)

and I want to box every scalar in the vector.

I was able to come up with a couple of solutions, like

(⍳∘≢⊆⊢) ⎕D

((1/⍨≢)⊂⊢) ⎕D

I came up with ((1/⍨≢)⊂⊢) ⎕D first and I thought it was too long, so I thought a bit more and came up with (⍳∘≢⊆⊢) ⎕D

But this still looks really long...

(on a completely unrelated note, sometimes my Dyalog interpreter tooltips stop showing; i.e. I hover a glyph and it doesn't show the little help section I usually get. Does anyone know why this happens/how to prevent it/how to fix it after it happened?)

Also, my WS is full? how is that even possible :P

Nevermind, I just realised my lack of parentheses meant I was trying to create a matrix of rank 10 or something like that.

Adám

7:24 PM

@RGS You can't enclose simple scalars, but the methods you've come up with anyway simply transform each scalar to a vector. You can do that with ,¨ too.

RGS

@Adám Alright, this is, in fact, much shorter and cleaner! Thanks

I mean the ,¨

Nathan

8:05 PM

@ngn @Adám do you know if there is a way to convert NGN Apl results to javascript data and back?

ngn

@Nathan you mean like dyalog's ⎕json?

there isn't but it shouldn't be too hard to write in apl

Adám

8:20 PM

@ngn how are apl arrays represented internally?

ngn

@Adám generally as js objects with a .s property for the shape and .a for the content, but there are gotchas - scalars are sometimes represented as ordinary numbers and those should be treated equivalently to {s:[],a:[thescalar]}

Nathan

What I mean is I don't understand how to use ngn apl results with existing javascript libraries without the means of collecting the result as javascript data.

I could write a wrapper for unpacking the data, but I was just curious if that already existed

ngn

@Nathan you could use those .s and .a properties of the result, if you really want, but i wouldn't recommend using ngn/apl for anything other than learning from my mistakes

Nathan

What mistakes are you referring to? I just think it would be useful if the returned object had a .toJS() method which would output the results in a useful format. That'd allow for some nice interop, and I could use ngn as part of a typical JS app

so apl('2 2⍴⍳4').toJS() would result in [[1,2],[3,4]] or something to that effect.

I'm just trying to see what ngn can do. I thought you could do things like var f = apl('{⍵+⍺}'); f(3,4) or something to that effect? I also thought that definitions in one call to apl('...') would be visible in the next call to apl('...'). Is that not the case?

ngn

@Nathan how would you distinguish between 2 2⍴⍳4 and (1 2)(3 4)?

@Nathan it's not as advanced as you think

Nathan

8:34 PM

For exporting to JS, I wouldn't. There's no such difference in javascript arrays.

[[1,2],[3,4]] is a 2x2 matrix so far as mainstream language notation is concerned

I could↓ the results, and get the .s from each of the resulting arrays. That seems simple enough

ngn

what if apl('2 2⍴⍳4') returns { isA: 1, a: Float64Array [ 0, 1, 2, 3 ], s: [ 2, 2 ] }?

Nathan

then the hypothetical .toJS function would return [[0,1],[2,3]]

Adám

And 0 2⍴0 would give []?

Nathan

I imagine it would result in [[]]

Adám

@Nathan That looks a lot more like 1 0⍴0

What would 'a'1 give?

Nathan

8:39 PM

Yes, I see that. You're right, it would return []

'a'1 I expect would give ['a',1]

Adám

Just like (,'a')1?

Nathan

[['a'],1]? Or not

Adám

What would 'abc' give?

Nathan

'abc'

Adám

'a' 'b' 'c'?

Nathan

8:41 PM

grr. Ok, then ⊂'abc' would give 'abc', but 'abc' would give 'abc'.split('')

Adám

@Nathan What would ⊂1 2 3 give?

Nathan

[1,2,3]

Adám

And 'a' 'b' 'c' 1 2 3?

Nathan

['a','b','c',1,2,3]

Adám

How about ⊂'a' 'b' 'c' 1 2 3 ?

Nathan

8:42 PM

That's the same thing

Adám

Uh, nope. 'a' 'b' 'c' 1 2 3 ≢ ⊂'a' 'b' 'c' 1 2 3

Nathan

I suppose it would mean [['a','b',..., 3]]

Adám

What would ,⊂'a' 'b' 'c' 1 2 3 give?

Nathan

[['a','b',..., 3]]?

Adám

So ⊂'a' 'b' 'c' 1 2 3 and ,⊂'a' 'b' 'c' 1 2 3 would give the same?

Nathan

8:45 PM

I would imagine

Adám

Would you also say that ,⊂1 2 should give the same as ⊂1 2?

Nathan

[[1,2]]

Adám

Fair, then ,⊂'abc' and ⊂'abc' should be the same too, I imagine?

Nathan

if a←1 2 ≡ [1,2], then ⊂a would mean [[1,2]].

I see

Adám

Good for you. Seeing is believing.

Nathan

8:48 PM

Why not just handle the type separately? When returning any sort of 'string' it should be returned from JS as a string

Adám

Define "string" (on the APL side).

Nathan

so then a 2 2⍴⎕A ≡ ['ab','cd']

I wouldn't

I would describe it on the .toJS side

Adám

How about 2 2⍴0,⎕A?

Nathan

[[0,'A'],'BC']

Adám

Seems a bit inconsistent, no?

Nathan

8:49 PM

That's annoying. Stop being socrates

:P

Adám

Stop proposing things you haven't thought through.

Jeff Zeitlin

@Adám - Wouldn't the canonical response to that be along the lines of "There's no such thing, just vectors where all the members happen to be characters"?

Nathan

It isn't a matter of thinking it through or not, it's a matter of I would like to use APL for javascript interoperatbility, and I can't, so I was asking if it was possible. And if it isn't, it would be nice if it were

Rumination doesn't need to be rigorously proven

Adám

@JeffZeitlin Sure, but that doesn't help us in converting to JSON.

@Nathan It'd be nice to have a way to call something that'd work like JSONserver, but I wouldn't venture into auto-mangling arrays. You'd call result=APL("{⌽⍵}",myJsArray) or result=APL(jsArray1,"{⍺⌽⍵}",jsArray2).

Even better if we could also pass JSON objects.

@ngn How hard do you think it'd be to implement ^ based on ngn/apl?

E.g. sometimes I want to .map() over two JS lists, so I'd want to be able to do APL(list1,"{code here}¨",list2)

ngn

@Nathan @Adám such interop would have been a nice thing to have, but i'm no longer interested in developing ngn/apl. i put all my efforts in k now.

Adám

8:57 PM

@ngn And K'd be much better for this, it having a much closer array model.

ngn

@Adám there are issues there too

char atom vs char list for example

Adám

Sure, like ("a",1)

ngn

.. vs symbols

Nathan

I seem to recall a K demo using canvas in the browser. Does anyone know where I can find that offhand?

Adám

@Nathan johnearnest.github.io/ok/ike/ike.html

ngn

9:02 PM

@Nathan also k7 and ktye/w

@Nathan comparison of k impls

dzaima

@Adám converting to APL would be much more problematic, whereas i can see an APL to JS being mostly usable even with 'a' 'b' 'c' returning a string

Adám

@dzaima tbf, ⎕JSON does an OK job.

dzaima

a .toJS in JS could even do fancy stuff like add array methods (map, filter, etc) to outputted strings, hiding the difference while keeping stringiness

Adám

11:43 PM

Feedback welcome.

dzaima

@Adám r←(2↑a),b - i (in my personal opinion) would leave that as-is (maybe sprinkle a couple spaces if it's not too unimportant), but would be okay with r ← b,⍨ 2↑a - spaces help my brain parse a lot. nested parenthesis, though, warrant a lot more to remove

Adám

@dzaima I find it interesting that you read b,⍨ as a unit. I read ,⍨ as "appended to", so I'd space it r ← b ,⍨ 2↑a

Bubbler

@Adám I personally tend to use parens over ⍨s because it is the natural direction for most of the primitives (one exception being ÷).

dzaima

@Adám it becomes more useful in succession, and especially in trains - (¯1↓ 1, A× ⊢) (maybe even things like (1↓ ∘, 10× ⊢) would make sense to do?)

Adám

@dzaima Yeah, I do that in my explanations of code golf.

Bubbler

11:58 PM

And I think you omitted one acceptable (and actually recommended) use case of parens: (a b c)←array

Adám

@Bubbler Oh, see that's the type of feedback I wanted.

Transcript for