The APL Orchard

4:01 AM

@Adám I put an issue in the dracula github repo to change errors to appear as Red (#FF5555), per the Dracula theme spec, if I interpreted it correctly :)

RGS

9:09 AM

@Adám interesting thing going on :)

I am pretty sure your dracula colour scheme messed a tad too much with my Windows registry

Adám

@RGS Oh no. What happened?

RGS

My computer thinks I have 18.1 installed :)

Adám

Ah, right.

That won't happen if we package it with the IDE.

RGS

And this is annoying because the jupyter kernel is trying to start and is looking for an 18.1 installation :P Do you know how I can fix this?

Adám

Just regedit and remove that branch of the registry.

RGS

9:12 AM

lol

That was easy 😅😂

Best way to start a week of work! Having to figure out why things that were working last week are no longer working :P

Adám

9:23 AM

Sorry about that. The repo with the theme is currently experimental. I should have foreseen and warned about this.

RGS

9:37 AM

I am not blaming you!

I was able to pinpoint the problem quickly enough so this wasn't frustrating.

RGS

1:23 PM

In RIDE there's no option to prevent the trace window to open automatically upon error, right?

(In the Windows IDE one can unselect the "Show trace stack on error" option but I can't find anything similar on RIDE.)

Adám

1:35 PM

@RGS Remember: RIDE is dumb. Configure the interpreter instead.

RGS

Interesting

Adám

@user11786133 Hi Edwin Schut. If you want to participate here, please email access@apl.chat

RGS

@Adám Erm the link says the default is 0, but that does not seem to be the case.

Or am I missing something?

Adám

I suspect the docs are wrong on that. Can you try setting it to 0?

RGS

I set trace_on_error to 0 on RIDE and the trace window stopped popping up automatically

Adám

1:43 PM

Do you know how to log an issue for that?

RGS

Yup

Ugh, the setting of the Windows IDE also influences RIDE

So I am not sure how to check the actual default value

Adám

1:58 PM

@RGS No it doesn't. RIDE is dumb. It has no idea about what's going on. The interpreter (which is always an IDE too) dictates its movements, including when to open and close windows. Really, RIDE isn't an IDE at all; it is simply a rendering library with a JSON protocol.

@RGS Simply erase the registry entry!

RGS

@Adám ok, then I am missing something.

If I go to Ooptions > configure > trace/edit on the Windows IDE and tick "show trace stack on error"

And then open RIDE and start a new connection without specifying the Trace_on_error variable, it will automatically open the trace upon hitting an error

If I close RIDE, untick the option on the Windows IDE, and then reopen RIDE and hit an error, the trace window will not open

Adám

@RGS The IDE doesn't write the values back right away. Try clicking OK and closing the IDE first.

RGS

Sure, I skipped a couple of steps in my explanation

But I just showed that the settings I define on the Windows IDE end up influencing RIDE...

I had said "Ugh, the setting of the Windows IDE also influences RIDE" and you replied "No it doesn't."

dzaima

@Adám (i don't think that message does/says literally anything except cause confusion)

Adám

@RGS I should have explained better then. Changing a setting in the IDE influences that instance of the IDE and any subsequent instances of the IDE. When you start APL from RIDE, it spins up a new IDE instance, with the currently active settings. If the IDE you used to set settings with hasn't written them to the registry yet, then RIDE will spin up an IDE with the previous settings.

dzaima

2:04 PM

I believe Adám was just pointing out a random technicality about RIDE, not strictly related to your observation that, as can be seen, the Windows IDE apparently influences RIDE

RGS

Ah ok, I understand what you meant now

Adám

@user11786133 You should have access now.

dzaima

(also imo this global settings stuff Dyalog does is awful)

user11786133

@ad

@Adám thanks!

Adám

@dzaima Yeah. Agreed. At least I succeeded in getting rid of that for the new #! scripting mode.

@user11786133 Welcome. With what can we be of service?

RGS

2:10 PM

@Adám turns out the default value is 0

user11786133

I have some problems with ¨ in combination with ⍨. example: w ← 1 2 3
⍝ how to ⊂ (partitioned enclosed) with, for example, three masks 0 0 1 1 0 0 0 1 1

three mask or split in 0 0 1, 1 0 0, 0 1 1

Adám

Ah, I understand. First, the solution:

      (0 0 1)(1 0 0)(0 1 1)⊂¨⊂1 2 3
┌───┬───────┬─────┐
│┌─┐│┌─────┐│┌─┬─┐│
││3│││1 2 3│││2│3││
│└─┘│└─────┘│└─┴─┘│
└───┴───────┴─────┘

Now, do you want an explanation too?

rak1507

@Adám well not really... you can still use ⎕IO right? or is it locked at 1?

user11786133

i tried: w⊂¨⍨0 0 1,1 0 0, 0 1 1
LENGTH ERROR

that was wrong :)

Adám

@rak1507 "global settings stuff" is fixed in the sense that the environment won't affect your code. We can't really change the core language for this.

rak1507

2:13 PM

Ah, ok, got it

Adám

@user11786133 Ah, but you should note that , simply concatenates the three vectors into a single flat vector.

RGS

@user11786133 try running just 0 0 1,1 0 0, 0 1 1 in the session

user11786133

explanation for the ⊂¨⊂ would be fine.

RGS

@user11786133 It's just someone extending their arms to the left :D

⊃¨⊃ is the same person but extending the arms to the right

Probably looking for a hug :P

Adám

It will work if you use ,⍥⊆ instead of , but that's pretty specific to this case.

user11786133

2:16 PM

ok, i tried: w⊂¨⍨(0 0 1)(1 0 0)(0 1 1)
┌───┬───────┬─────┐
│┌─┐│┌─────┐│┌─┬─┐│
││1│││2 2 2│││3│3││
│└─┘│└─────┘│└─┴─┘│
└───┴───────┴─────┘
0 0 1⊂w
┌─┐
│3│
└─┘

Adám

@user11786133 ¨ pairs up elements of arguments in much the same way as + does. E,g, 1 2 3+10 20 30 gives 11 22 33 and 1 2 3+10 gives 11 12 13.

user11786133

but only 0 0 1 gives a different outcome with the first result of each

Adám

However, for the pairing up to work, as you want, you need a vector on one side, i.e. (0 0 1)(1 0 0)(0 1 1) and a scalar on the other, i.e. ⊂w

(⊂w)⊂¨⍨(0 0 1)(1 0 0)(0 1 1) will work.

If you do not enclose w then APL pairs up w's 1 with the 0 0 1 and w's 2 with the 1 0 0 etc.

user11786133

Thanks! that's also the extra ⊂ in your solution.

Adám

Exactly. In this case, you could also write ⍥⊆ as in w⊂¨⍥⊆⍨(0 0 1)(1 0 0)(0 1 1).

Which means you can get rid of all parentheses with w⊂¨⍥⊆⍨0 0 1,⍥⊆1 0 0,⍥⊆0 1 1 if you really want to. But do mind that this isn't generally applicable.

user11786133

2:23 PM

thanks, i didn't used ⍥ yet.

Adám

It is quite a wonderful operator.

Btw, if you intend to frequent this room, I highly recommend having a look here for some tips and tricks.

user11786133

Thanks for the link. Why a combination of ⍨ and ⍥ in de last solution?

Adám

⊂ takes the data on the right, and the partitioning info on the left. ⍨ swaps arguments, so e.g. 2-⍨10 is 8.

Edwin Schut

thanks

Adám

Any time (almost).

Edwin Schut

2:38 PM

why is the result different when given only 1 element: (0 0 1)(1 0 0)(0 1 1)⊂¨⊂1 2 3
┌───┬───────┬─────┐
│┌─┐│┌─────┐│┌─┬─┐│
││3│││1 2 3│││2│3││
│└─┘│└─────┘│└─┴─┘│
└───┴───────┴─────┘
(0 1 1)⊂¨⊂1 2 3
┌┬───────┬───────┐
││┌─┬─┬─┐│┌─┬─┬─┐│
│││1│2│3│││1│2│3││
││└─┴─┴─┘│└─┴─┴─┘│
└┴───────┴───────┘

RGS

It's related to why you do ⊂1 2 3

You are doing ⊂1 2 3 to tell ⊂¨ that the thing on the right should be seen as a single element.

0 1 1 is being seen as a vector, but you also want it to be seen as a single element, so you need (⊂0 1 1)⊂¨⊂1 2 3

Ofc at this point you should just do 0 1 1⊂1 2 3.

Does that make sense?

Adám

It may be confusing that (0 1 1) is NOT the same as ⊂0 1 1 but rather IS the same as 0 1 1

0 1 1⊂¨⍥⊆⊂1 2 3 would work though, with the final result being enclosed. It'd be the same as ⊂0 1 1⊂1 2 3

Edwin Schut

i find it confusion that the function behave different for a single (0 0 1) or multiple inputs (0 0 1) (1 0 0) (0 1 1).

Adám

It isn't so much the function that behaves differently. Rather you need to be aware that 0 1 1 is a flat vector while (0 0 1) (1 0 0) (0 1 1) is a vector of vectors.

Edwin Schut

i would expect in this example that (0 1 1)⊂¨⊂1 2 3 gives the third part of (0 0 1)(1 0 0)(0 1 1)⊂¨⊂1 2 3

Adám

2:45 PM

The "single" equivalent of (0 0 1) (1 0 0) (0 1 1) is ⊂0 1 1 or possibly ,⊂0 0 1

@EdwinSchut But notice that you have an "each" (¨) so you are asking to pair up each element from the left with each element from the right.

Edwin Schut

a, and with a single, the function act 1 level deeper?

Adám

Almost. The function is consistent. The data has one less enclosing level.

We call it the depth.

Edwin Schut

And with ( )( )( ) there is a extra depth so "each" act on depth ( ). And with ( ) "each" act on the elements in ( ).

Adám

Right, except that the parentheses themselves don't add depth. It is the juxtaposition of vectors that adds the depth.

If yo named each component, a←0 0 1 ⋄ b←1 0 0 ⋄ c←0 1 1 then a b c would give you the same as (0 0 1) (1 0 0) (0 1 1)

Edwin Schut

ok , thanks. And the ⊂ for 1 2 3 is to make both arguments the same depth ( for the second ⊂) ?

Adám

2:55 PM

@EdwinSchut Which exact expression are you talking about now?

Edwin Schut

sorry, this one: (0 0 1)(1 0 0)(0 1 1)⊂¨⊂1 2 3

Adám

Ah yes, exactly so.

The result would be the same as (0 0 1)(1 0 0)(0 1 1)⊂¨(1 2 3)(1 2 3)(1 2 3)

Edwin Schut

Exactly, that make sense.

Adám

Great!

Edwin Schut

3:19 PM

A second question, where i like to use "each", i think :)
with w <- 1 2 3 4 you can +/ w and -/w.
what's the form to apply +/ and -/ to w with each? So how to apply a list of functions to w?

Adám

What result are you expecting?

+/w is 10 and -/w is ¯2, btw.

RGS

@Adám Edwin wants to "map over a list of functions"

Adám

Which you can't easily do in APL, but I find it rather rare to actually be needed. There's usually a neat alternative. In this case, maybe

      (+/,-/)w
10 ¯2

Edwin Schut

it's more a general example. It's indeed my way to ask how to map over a list of functions. In this case is the result a list with two results?

that's easy.

Adám

In general, you can write (f,g,h) but note that the results are concatenated, not juxtaposed.

I.e. you may need (f,⍥⊆g,⍥⊆h) for simple cases, or a custom solution for more complicated cases.

This latter formula works for getting the two scans:

      (+\,⍥⊆-\)w
┌────────┬─────────┐
│1 3 6 10│1 ¯1 2 ¯2│
└────────┴─────────┘

Edwin Schut

3:26 PM

Thanks, i believe there are neat alternatives; My frame of mind needs some adjustments .. :)

Adám

It takes time. Feel free to ask all the questions.

CMQ: Other than the core interpreter, which parts of the Dyalog system would you like to be opened up to user contributions? E.g. documentation, dfns.dws, colour schemes…

RGS

All of the above ↑ :P

rak1507

as much as possible

Adám

@rak1507 That's not an actionable item. This is your opportunity to actually get your voice heard, so specific things, please!

phantomics

Is it possible to use ∇ to self-reference inside a defined operator? The below code isn't working the way I want:

rlop←{⍵<2000 : ⍵,⍺⍺ ∇ 3 ⍺⍺ 2×2+⊃⍵ ⋄ ⍵} ⋄ × rlop 5

It gives 5 1 1 1 1, so the first ⍺⍺ is being applied monadically as ×, rather than being the left operand to operator ∇

Adám

3:42 PM

@phantomics This?

dzaima

∇∇ is self as modifier. ∇ is the derived function

rak1507

@Adám {(is_possible¨⍵) / ⍵}all_items

phantomics

Ok, that makes sense

Adám

@rak1507 If you don't want to help me gather specific actionable items to be forwarded to management, then that's of course fine.

rak1507

Documentation, dfns.dws, colour schemes, packages, package manager (if that ever exists), configs (stop the keyboarding issues), etc

phantomics

3:45 PM

In terms of actionable items, making the TryAPL software open and modular could be neat, giving people the ability to create APL interpreter-driven web apps

Adám

@rak1507 Thanks. The first three are already on my list, obviously. The next two are already in the open (though WIP). Can you explain what you mean by accepting user contributions to "configs"?

@phantomics Wouldn't it be enough to publish the API? (Something I think @RikedyP was planning to do anyway.) TryAPL Mini already uses it.

rak1507

@Adám if possible, control over keyboarding issues mostly

Adám

@rak1507 Again, I'm not sure what you mean. We are removing the SuperKey hijacking in 18.1 and I think we're also making it easier to choose your keyboarding. I'm not sure what type of community contributions you are envisioning for that area.

rak1507

I know people have contributed scripts and stuff to fix issues in the past. If they were able to contribute that directly to dyalog so you didn't have to get it from aplwiki or whatever, that would be good. (that's sort of what I mean, sorry for awkward wording)

If stuff is being fixed/changed in 18.1 it's probably irrelevant anyway

Adám

@phantomics Btw, you can't really use the TryAPL system to drive much of a web app, as there's no server state. You'd have to send the entire state back and forth between frontend and backend all the time. (This is indeed what TryAPL and TryAPL Mini do.)

phantomics

3:51 PM

@Adám Perhaps an editable config file that lets you specify how you want APL keybindings to work? An input mode plugin for common Linux input mode managers? That would allow for some customizability

@Adám I see, perhaps you could have an app that calls to the interpreter independently of the user's APL interaction, using it as a database

Adám

@phantomics We already have human editable config files (JSON5) as of 18.1 :-) and while a unified input control system would be nice, it isn't something we currently sit on, so we can't "just" open it up to the community.

@phantomics Well, as a computation engine, yes. Not as a database, since it doesn't store any data. Unless you mean that people could run the whole stack with a custom APL in the backend. But at that point, they are way better off using Jarvis directly. TryAPL in fact is a small (<300 LoC) APL service served by Jarvis together with some static HTML/CSS/JS.

Adám

8:51 PM

CMP: What do you all think of this adaption of New Moon?

RGS

9:08 PM

I think the label named EndFor hurts my soul.

rak1507

I agree with RGS

the theme looks nice though!

RGS

Other than that I like the Dracula one better, for some reason I don't like these colours as much.

It might be the fact that my screen currently has f.lux turned on and so everything is yellow-ish.

rak1507

I think I prefer this one to be honest, this one is a bit more muted which I prefer

RGS

Yeah but personal preferences are just like that, everyone has their own :)

rak1507

yeah, the more options the better!

RGS

9:11 PM

Btw I'm really sad no one thought my earlier remark about ⊂¨⊂ and ⊃¨⊃ being people reaching for a hug was hilarious

rak1507

@RGS I did

RGS

I should know better than that, but I laughed way too much at my own joke :P

dzaima

also functions and operators are still the same color, with which i still disagree

rak1507

good point, agreed ^

RGS

@dzaima Can you do something about it in the Windows IDE?

Adám

9:15 PM

No.

dzaima

@Adám oh god

Adám

I can ask John if we could start differentiating, but I doubt it.

dzaima

poor windows IDE users

Adám

Well, they all have the option to use RIDE, of course.

rak1507

will the windows ide ever be 'phased out'?

Adám

9:18 PM

I think that is a very long term option if RIDE significantly surpasses it.

rak1507

Seems like it would be easier from an implementation point of view to just focus on RIDE, especially considering how many features it has

RGS

@rak1507 Whut? RIDE has more features than the Windows IDe?

Have I been living under a rock..?

rak1507

not sure about more but it definitely has some features that the windows one doesn't

Adám

The development teams are entirely separate, but hey, RIDE is FOSS, so feel free to contribute!

rak1507

like tab autocomplete and backtick typing

Adám

9:21 PM

Uh, Windows IDE has both of those.

RGS

↑ that

rak1507

?! how

RGS

(But backtick requires the IME, right?)

OPtions → configure → Autocomplete

Adám

@RGS (yes)

rak1507

I mean tab completion for primitives

Adám

9:23 PM

Oh, you mean ``name

rak1507

yeah

Adám

Right. I guess the more experienced users don't really miss that.

@A-LevelStudent Hi and welcome. Interested in APL?

rak1507

yep, I wouldn't use backticks either but I know less experienced people think that those features are pretty useful

Adám

One thing the IDE doesn't have, but RIDE has, is the ability to bind most any keystroke to a "macro" (sequence of actions).

rak1507

oh, that's quite cool

I bet roger hui has a macro for ⎕IO ← 0 ⍝ io delenda est

Adám

9:26 PM

It also allows binding multiple keystrokes to a single action. IDE doesn't let.

It also allows evaluating the selection or name under the cursor. (But that can be added manually to the IDE.)

Oh, and RIDE can show your keystrokes for demo purposes. That's pretty cool and would be even cooler if it recognised AltGr :-(

rak1507

can someone go here 4502-aquamarine-horse-s66yby97.ws-eu03.gitpod.io and tell me what they see? I'm trying bubbler's gitpod thing and wondering if it can be used with two people at once

(feel free to type stuff, if you can)

RGS

@rak1507 to me it says the port is closed

Adám

I can type, but Enter seems to do nothing, and CSS is half-way messed up.

rak1507

alright, interesting, thanks both

RGS

Now it shows the Dyalog RIDE screen

Like the splashscreen

rak1507

9:35 PM

it looks like you can 'access' it from a second location, but it only works on one

there goes my idea of live apl without screensharing, oh well

dzaima

previous conversation

Adám

9:50 PM

@RGS I think the Portuguese APL Meetup isn't a "User Group". I see a user group as an established organisation that hosts multiple activities. A regularly recurring online meetup is an event or "activity", even if the attendees form a group. Further, an online meetup isn't tied to a particular geographical location, and even more so if the language of the event is spoken in multiple far-apart places.

RGS

@Adám Thanks for bringing it up.

Had forgotten about it by now. I actually meant to ask what I should do about that. Today I noticed I had placed it in a weird place, and ended up concluding I don't know where to place it.

Maybe it doesn't belong there at all.

Ah wait, you moved it to "activities"!

Adám

Yeah.

RGS

I thought I had done it by mistake, you fixed it and then I undid your changes xD

That's unfortunate on my end, sorry. Today at around 13 I opened APL wiki and saw there had been no changes to the pages I was following.

Adám

No worries. I'll roll back your latest change.

RGS

And later today I opened the wiki again and noticed that and it didn't occur to me someone might've changed it meanwhile :P

ty

Adám

9:55 PM

Btw, isn't the event's title "Weekly APL Meetup"?

Or "APL Meetup"?

@RGS Your Pt title says "Meetup de APL" and the En title has "APL Meetup", but your poster says "Meetup Semanal de APL" ⍨

RGS

The poster was meant to summarise some info and I included the frequency in big letters

I guess the title is just APL Meetup

But prepending "Portuguese" seems sensible

Adám

That shouldn't be in the title, if that isn't actually the name. I'll fix it.

RGS

Alright, then maybe I "should change the name of the event".

Adám

No, that's fine. Funny enough, the German APL group has a name in English.

And so does the Finnish one.

RGS

Well

"APL Meetup" is awfully generic

Adám

10:03 PM

You booked the name first!

Nitpick: Isn't the objective to teach Portuguese-speakers APL, rather than to promote such teaching?

RGS

The text says "promote learning", not "promote teaching"..?

Is that not something you would say in English?

Adám

IANAENS.

RGS

In that the sessions aren't strictly for me (or someone else, for that matter) to teach, but for me to create an environment where people want to learn

rak1507

@Adám what does that stand for?

RGS

@Adám Dunno what that means.

Adám

10:07 PM

Ah. OK, I've rephrased it. What do you think?

I Am Not An English Native Speaker.

RGS

@Adám I like it.

Adám

I guess it should have been IANANES, but then again, I'm not…

RGS

So now the community template should also be updated, no?

I can do it, just don't want to write over whatever you might be writing rn.

Adám

@RGS Done.

I wonder if there similarly would be a market for Nic and/or Ven teaching APL in French…

rak1507

if they do they should definitely call it l'appel de l'APL

(I don't speak french, that probably makes no sense)

Adám

10:13 PM

:-D I got it, and my High School French isn't strong.

RGS

@rak1507 is "appel" your attempt at saying apple in french?

If so, it's pomme :)

rak1507

@RGS no, l'appel means 'the call' in french

RGS

AH :laugh:

"appel" looks like a person is saying apple with a French accent

rak1507

lol

Adám

So if I create Dracula and/or New Moon themes for RIDE, should I differentiate between functions, hybrids+monadic operators, and dyadic operators?

dzaima

10:17 PM

i mean i'll of course say yes

Adám

@dzaima What colours should I give them then?

I guess I can then also colour ⋄ like a keyword rather than a primitive.

I kind-of feel bad giving any of them the same colour as numeric constants or strings, or control structures, or global variables, or errors.

dzaima

@Adám it's kind of hard to guess without being able to play around y'know

what's the current list of color←→thing mappings?

Adám

normal: white
global: orange
error: red
control: pink
number: purple
string: yellow
idiom: green
function: cyan

I could make idioms underlined instead of green, and that frees up green for hybrids and monadic operators. Dyadic operators can be yellow like strings, as they are hardly confusables.

Hm, but 'x'@ is unfortunate.

dzaima

global orange for dyadic operators might work

Adám

You don't think it'd detract from the warning that the orange is supposed to provide?

rak1507

10:29 PM

nahh it'd probably be fine

Adám

Dracula is really missing a blue.

dzaima

orange to me doesn't associate with "warning" at all

@Adám probably because blue is just an awful color to work with

Adám

Safety orange

Safety orange (also known as blaze orange, vivid orange, OSHA orange, hunter orange, or Caltrans orange) is a hue. Safety orange is used to set objects apart from their surroundings, particularly in complementary contrast to the azure color of the sky (azure is the complementary color of orange, and therefore there is a very strong contrast between the two colors). The color is commonly used for hunting during the rifle season, and is also used for quail hunting. Most places, such as construction sites, use safety orange to help ensure the safety of others. ANSI standard Z535.1–1998 states how...

dzaima

@Adám that's a much more stronger orange, and without context, even that orange doesn't associate with "warning"

rak1507

shades of orange: Hunyadi yellow
how is that a shade of orange?!

dzaima

10:33 PM

(paste uses orange for monadic operators)

(it's also kind of hard for me to give suggestions as obviously i prefer my theme, where green is definitely functions; also idioms are gray bold underlined)

rak1507

idioms aren't underlined in paste right?

Adám