The APL Orchard

nathan rogers

20:00

that's odd

ngn

"subject" :)

nathan rogers

oh wait you have the rotate ( for joining data

Marshall

And the name and value have to have the same role in assignment, so for most large expressions you can tell the final type immediately.

@nathanrogers And you get to not panic and look rightward every time you apply a function with rank 1.

dzaima

@ngn it's quite similar, but either a font or coloring fix that

and speaking of coloring, if one wanted to, syntax-highlighting for BQN could color parentheses by whether they group an array, function, 1-modifier or 2-modifier. (but i don't think that's particularly useful)

Marshall

@dzaima I think it's useful but it would be very hard to get any editor to do it.

ngn

20:04

@dzaima mumbles: why have the problem and fix the problem instead of not having the problem in the first place..

^this is a common theme here

dzaima

@ngn there aren't many other options - there aren't (m)any other usable paired unicode characters, and not having array notation is also stupid

nathan rogers

@dzaima I use lisp, and paredit makes telling which scope you're in really easy through colored parens

ngn

@dzaima is (a⋄b) meaningful in bqn?

Marshall

@ngn No, it's banned.

dzaima

@Marshall anything that allows coloring things by non-context-free rules should easily allow for it (though i don't know what, if any, of such cases would be; and i also might or might not at some point make my own editor too so ¯\_(ツ)_/¯ (though i'm at the "might or might not" stage of way too many things))

@ngn but (a) wouldn't be a single-item list then, and that's stupid

ngn

20:08

@Marshall @dzaima there you go. why not make that the array notation?

@dzaima (⋄a) or something.. there are ways

Marshall

@ngn Because being able to write one-element lists is very important.

dzaima

@ngn that's just ugly

ngn

no uglier than that thing that looks like the illegitimate child of < and (

dzaima

@ngn imo it is less ugly

nathan rogers

can someone help me find another way to determine if all substrings are part of a string?

('ac' 'cd' 'pq' 'xy')∨.∊⊂'accdpqxy'

dzaima

20:10

@nathanrogers wouldn't you want ⍷ instead of ∊ there?

nathan rogers

     ('ac' 'cd' 'pq' 'xy')∨.⍷⊂'accdpqxy'
┌───────────────┐
│1 0 1 0 1 0 1 0│
└───────────────┘

dzaima

@nathanrogers you still want the reduction - ∨.(∨/⍷)⊂

nathan rogers

oh, so I was right

thanks. I never can remember the difference between ∊ and ⍷

ngn

@Marshall you must have a primitive for making a 1-element list?

dzaima

∨/ a ∨/⍤⍷¨ ⊂b is an alternative

@ngn for some reason, Marshall decided to not. I still definitely want one

ngn

20:13

oh.. i was going to suggest (thatprimitive ..) as the obvious solution

dzaima

and (a⋄b⋄c) (a⋄b) ⍮a (or whatever) isn't a nice progression either

Marshall

@ngn I can't find a good glyph to use. But it is absolutely not okay that you would delete an item from a literal two-element list and not end up with a literal one-element list. So it's not really relevant.

≍○< is the standard function to make a list of one or two arguments.

ngn

@Marshall 0_0 that's a 2-element list, right? (imagine the _ is curved)

dzaima

@ngn the thing being that the creation and solving of a given thing you're whining about also solves something else (that you don't care about, but others do)

ngn

@dzaima right. i whine about many things. it's a matter of deciding which is more important. like in real life.

Marshall

20:17

@ngn Stranding does have that issue, but the programmer can think about whether they're likely to change the number of elements and decide whether to use it. There has to be some option that adapts gracefully to all cases though.

dzaima

@ngn I'm not exactly a fan of ‿-notation, but the alternatives aren't good (always using ⟨⟩-or-whatever-notation is a lot of overhead, and k-style array tokens and APL strands both make for weird syntactic things)

nathan rogers

trying to look at the tree for this guy

(3≤+/,∊'aeiou')

but I get DOMAIN ERROR

ngn

@nathanrogers "the tree"?

nathan rogers

um... I'm trying to evalute it

and get the representation

yes

dzaima

@ngn creating and solving a problem, "in real life", is net zero. If doing that also solves another problem, it's a net positive.

ngn

20:26

@nathanrogers , and ∊ are no-ops there. so, you're trying to sum a string.

nathan rogers

well

i'd like to understand why it won't even evaluate to show me the representation at least

dzaima

@nathanrogers representation of what?

nathan rogers

it doesn't look like completely invalid APL

the train

come on

dzaima

@nathanrogers there's no trains here

nathan rogers

      (+/÷≢)
┌─────┬─┬─┐
│┌─┬─┐│÷│≢│
││+│/││ │ │
│└─┴─┘│ │ │
└─────┴─┴─┘

yes the representation

dzaima

20:28

@nathanrogers right, that's a train. (3≤+/,∊'aeiou') isn't a train

nathan rogers

right

I understand that

but it isn't parseable?

dzaima

@nathanrogers it's "parseable", but entering a line in the Dyalog REPL executes it, not gives you a parse tree (which isn't even a thing in Dyalog)

Bubbler

domain error ≠ not parseable

nathan rogers

oh, I get it

nvm

ngn

@nathanrogers you seem very confused about trains. read carefully: a train is a sequence of terms that ends with a function. it must be isolated in some way, like being alone on the right-hand side of ← or being enclosed in ( )

3≤+/,∊'aeiou' is not a train because the last term 'aeiou' is not a function

nathan rogers

20:31

yeah I got that

ngn

the 3≤+/ part of it is not a train either, because it's not in isolation, and the sequence of terms continues

Marshall

BQN also has an extensive document on trains.

ngn

@nathanrogers so, did you mean something like (3≤+/),∊{'aeiou'} or (3≤+/),∊'aeiou'⍨ or (3≤+/)'aeiou'∊⍨, ?

nathan rogers

20:44

@ngn ('aeiou'(3≤⊣).(+⌿∊)⍨⊂) 'bunaaaac'

I don't understand how this works

and I wrote it

dzaima

@nathanrogers that happens with trains

ngn

@nathanrogers golfed: 3≤1⊥∊∘'aeiou'

nathan rogers

in particular, I'm confused why 3≤⊣ and not ⊢

dzaima

@nathanrogers it also doesn't work - try 'aeiou' as an input

ngn

@nathanrogers i don't think it's correct. what should it return for 'banuuuuc'?

nathan rogers

20:49

(3≤ 1⊥ ∊∘'aeiou')'bunaaac'
1

how?

it works for me

more than 3 or more vowels 1 less than 3 0

dzaima

@nathanrogers yours doesn't - ('aeiou'(3≤⊣).(+⌿∊)⍨⊂)'aeiou' gives 0

nathan rogers

uh

wtf

Adám

@nathanrogers Known bug, being fixed.

ngn

@nathanrogers ∊∘'aeiou' is a composition that means {⍵∊'aeiou'}. 1⊥.. is sum. 3≤.. i'm sure you understand that.

nathan rogers

@Adám what's a known bug?

dzaima

20:51

@nathanrogers click the reply indicator

nathan rogers

oh thanks

so... ∘ means ⍵ gets applied to whichever side is thefunction?

or you're saying that f∘A is just a curried function awaiting an argument

ngn

@nathanrogers a∘f ←→ {af⍵}; f∘a ←→ {⍵fa}; f∘g←→{f g ⍵} (or {⍺ f g ⍵} when applied dyadically)

dzaima

@nathanrogers yep, that's precisely what that is. and A∘f does the obvious thing of currying the left arg

nathan rogers

cool

didn't know that

I thought it was positional

i'm not sure how this works either

(∨/ 2=/ ⊢)

I would have assumed this would (∨/⍵) 2=/ {⍵}

oh

never mend

dzaima

@nathanrogers with ∨/→f and =/→g, does (f 2 g ⊢) become clearer?

nathan rogers

20:58

its ∨/ {2} =/ {⍵}

i'm getting there, its slow, but I wrote a few trains, ok?

ngn

22 hours ago, by ngn

if you have more than 3 functions in a row, the last 3 become a fork

nathan rogers

a fork being f g h i (f (g ⍵) h (i ⍵))

ngn

@nathanrogers f 2 g ⊢ ←→ f (2 g ⊢)

nathan rogers

yes I saw that already

4 mins ago, by nathan rogers

its ∨/ {2} =/ {⍵}

dzaima

@nathanrogers f g h i is an atop and a fork. f g h is a fork

nathan rogers

21:04

I thought g h i was the fork

with f (g h i) as the atop

dzaima

@nathanrogers right, g h i there is the fork. i was saying that in general

nathan rogers

oh

hah thanks

now this confuses me

nvm again

s05_1  ← 1⊥{                             ⍝ 5_1 How many total where:
    pair ← ∨/2=/⊢                        ⍝ ... Any neighbors match?
    vowl ← 3≤1⊥∊∘'aeiou'                 ⍝ ... 3 or more vowels?
    subs ← 'ac' 'cd' 'pq' 'xy'∨.(∨/⍷)⊂   ⍝ ... Substrings exist?
    (pair ∧ vowl ∧ subs) ⍵               ⍝ ... All true?
}¨

@ngn problems?

looks like errors

nevermind... again

^
line(36,0) : error AC0008: error (SYNTAX ERROR) executing line "}¨"

ok now I know this has to wrong

I've done this a lot

what's the problem here

is this a v18 thing?

dzaima

@nathanrogers dyalog may or may not Sometimes™ allow multiline dfns only in abc←{…} and nothing else

nathan rogers

I don't follow

dzaima

21:19

@nathanrogers multiline dfns are special, in that in some contexts you can only assign them and nothing else

nathan rogers

I am assigning

dzaima

@nathanrogers but you're also using it as the operand of ¨

nathan rogers

I removed the 1⊥ and I know for a fact I have code that does
f ←{
...
}¨

¨ is a function now?

rak1507

no, an operator

nathan rogers

whatever

I didn't know it was an operator in the first place

rak1507

21:21

what else would it be

nathan rogers

syntax?

dzaima

@nathanrogers it's a regular monadic operator like ⌸ or / or ⍨

nathan rogers

yeah well, I have other operators defined similarly

sorry, operator derived functions

s02_1 ← +/{{(⌊/⍵)+ 2× +/⍵} 2×/ ⍵,⊃⍵}¨

like that one

no errors

dzaima

@nathanrogers that dfn isn't multiline though

nathan rogers

i mean

dzaima

21:24

if you have a multiline dfn, you must assign it immediately and nothing else because Reasons™

actually, it seems you can use multiline dfns mid-expression inside dfns, but not inside :namespace. Dyalog yay

rak1507

who needs more than 1 line anyway /s

nathan rogers

I know I have done this

I thought for sure in APL2XL I had a dfn or 10 that was defined with ¨

but I can't find it!

I've truly stepped into another timeline

dzaima

@nathanrogers maybe inside another dfn, like i said

nathan rogers

no I mean, I know I"ve defined
f←{
...
}¨

}¨) can't even do this

@Adám is this intentional?

this has to be a v18 bug, I'm downloading 17.1 to check

dzaima

@nathanrogers as i said, maybe you wrote that inside another dfn. see this

nathan rogers

21:27

oh I see

bleh

yes everything besides the external mutation and error handling is done inside a dfn

so that's probably right

dzaima

it's absolutely stupid that this is the case, but what can ya do

rak1507

oh yeah fun fact it doesn't work with ]dinput either

nathan rogers

agitate

rak1507

it's like the parser sees the } and goes 'yep we're done here ok time to go home'

nathan rogers

s05_1  ← {{                             ⍝ 5_1 How many total where:
    pair ← ∨/2=/⊢                        ⍝ ... Any neighbors match?
    vowl ← 3≤1⊥∊∘'aeiou'                 ⍝ ... 3 or more vowels?
    subs ← 'ac' 'cd' 'pq' 'xy'∨.(∨/⍷)⊂   ⍝ ... Substrings exist?
    (pair ∧ vowl ∧ subs)⍵               ⍝ ... All true?
}¨⍵}

oh yeah, this is dandy

dzaima

21:29

@nathanrogers yay for workarounds

nathan rogers

dzaima

also you could just (pair ∧ vowl ∧ subs)¨⍵ instead

nathan rogers

true

ok here's something strange

     ('ac' 'cd' 'pq' 'xy'⍱.(∨/⍷)⊂)¨ 'acasdf' 'cdasdf' 'pqasdf' 'xyasdf'
0 1 0 0

why is cd ok, but the rest aren't?

ngn

@nathanrogers why ⍱ instead of ∨

rak1507

⍱/x is not the same as ~∨/x

Adám

21:43

@nathanrogers It is a known bug in namespaces, but it works in the new multi-line mode, and thus it works in !# scripts too, so… yay?

rak1507

what new multiline mode (]dinput?)

nathan rogers

!# scripts are real now?

@ngn because I want a 1 where not includes

Adám

@nathanrogers ⍤0 1 pairs up scalars from the left with vectors from the right. ⍤0⍤1 pairs up vectors, then inside those, it pairs up scalars.

rak1507

@nathanrogers see my message

Adám

@rak1507 That's obsolete now.

rak1507

21:44

what's replaced it?

Adám

@nathanrogers In 18.1, yes.

nathan rogers

@rak1507 its giving me the same results now

rak1507

⋄ ~('ac' 'cd' 'pq' 'xy'∨.(∨/⍷)⊂)¨ 'acasdf' 'cdasdf' 'pqasdf' 'xyasdf'

nathan rogers

@rak1507

      ('ac' 'cd' 'pq' 'xy'(~∨).(∨/⍷)⊂)¨'acasdf' 'cdasdf' 'pqasdf' 'xyasdf'
0 1 0 0

rak1507

yes

nathan rogers

21:45

cd is the only one returning problems

rak1507

(~'ac' 'cd' 'pq' 'xy'∨.(∨/⍷)⊂)¨'acasdf' 'cdasdf' 'pqasdf' 'xyasdf'

nathan rogers

the others are returning accurate results

Adám

@rak1507 Proper multi-line input mode:

rak1507

the others are getting the right thing for the wrong reason

@Adám oh cool not a hack

Adám

@rak1507 Not at all.

rak1507

21:45

great!

dzaima

@nathanrogers (~∨)/a is still not what you want

Adám

Plus, if you edit things that are indicated by that green [ then the whole block is run again.

nathan rogers

@Adám so !# scripts are in 18.1?

Adám

@nathanrogers Yes, yes, they are.

nathan rogers

:D

Bubbler

21:46

Not !#, it's #!

rak1507

@nathanrogers ⍱/∘.=⍨⍳4

Marshall

Added a large section on the history of APL control structures. Can anyone confirm that GNU APL doesn't have them?

Adám

@Bubbler Yes, of course :-)

@Marshall Confirmed:

      ∇ foo
      :if 1
      :endif
      ∇
Line 1
Offending token: (tag > TC_MAX_PERM) 0x43030019 :
Statement: `:  `'if  `VALUE3«1»
Fatal error in defined function. Please correct (or else delete with [∆1])
 the offending line indicated above.

ngn

this isn't an error message, this is an error novel :)

nathan rogers

@Adám stackoverflow.com/questions/37845074/…

comment on second answer says that the → is correct control structures in gnu apl

Adám

21:51

@nathanrogers Not sure what you mean. Can you link to the comment?

nathan rogers

idk how?

6

A: If-Else conditions with APL?

If your APL supports control structures then this should work: ∇ generateAll number :If number>100 ⎕←'Too high' :else ⎕←⍳ number :endif ∇ If it does NOT support control structures (like APL2) you will need to branch: ∇ generateAll number →(number>100)/error ⎕←⍳ number →0 error: ⎕←'T...

dzaima

The second example (without control structures) should be correct for GNU APL. — Lobachevsky Jun 22 '16 at 10:19

nathan rogers

comment by lobachevsky

Marshall

@Adám Thanks. I guess it's unlikely it would use a different syntax when : is so common.

Would be nice to know where that syntax comes from as well. Looks like APL*PLUS had them (at least :for) before Dyalog, based on this paper.

Adám

@nathanrogers Oh, that wasn't the second answer in what SE showed me :-)

@nathanrogers fyi, for next time: click the time next to the comment.

nathan rogers

21:59

oh ok

ty

Marshall

22:24

@Marshall Definitely APL*PLUS; I've added some Vector articles from when it introduced them as references.

From Adrian Smith: "This apart, I can see no problems with the basic idea, so the sooner everyone else copies it, the happier I will be."

nathan rogers

@Adám I'm getting all kinds of import errors

nonstop

s05_1_tests ← {,⊂{(⊂0⌷⍵),(+/1⌷⍵)}⍉↑⍵} s05_1a_tests

it imports this line just fine at ]link time

but when the file is modified, errors for days

Adám

@nathanrogers Wait, that looks awfully recursive.

rak1507

1a

nathan rogers

line(73,0) : error AC0008: error (DOMAIN ERROR) executing line "s05_1_tests ← {,⊂{(⊂0⌷⍵),(+/1⌷⍵)}⍉↑⍵}s05_1a_tests"
^
line(74,0) : error AC0008: error (VALUE ERROR) executing line "s05_1_tests ,←(i05 258)"
^
Complete: 2 errors.
line(73,0) : error AC0008: error (DOMAIN ERROR) executing line "s05_1_tests ← {,⊂{(⊂0⌷⍵),(+/1⌷⍵)}⍉↑⍵}s05_1a_tests"
^
line(74,0) : error AC0008: error (VALUE ERROR) executing line "s05_1_tests ,←(i05 258)"

Adám

@rak1507 Oh, missed that.

nathan rogers

22:28

it runs just fine from the session

Adám

@nathanrogers So you're running code straight in the namespace script?

nathan rogers

I'm assigning variables in the namespace

in text files

Adám

@user13730524 Hi alphatask. If you want to participate here, then email [email protected]

@nathanrogers Namespace scripts don't stop on error, but try to continue fixing the script. Having skipped an erroring line, you get this sequence of errors. Try to avoid running code at fix time.

nathan rogers

I'm not trying to run code, but I already have data defined and just want to reshape the values......

Adám

Don't do that at fix time.

nathan rogers

22:33

I want the _tests to be assigned some value, that isn't static

... its in a text file

when else can I do it

Adám

At run time.

nathan rogers

I have the project linked, and its in a text file

and I'm updating the text file...

Adám

You really shouldn't use Link to run code whenever a file is changed.

nathan rogers

I'm not

Adám

If your namespace script contains s05_1_tests ← {,⊂{(⊂0⌷⍵),(+/1⌷⍵)}⍉↑⍵} s05_1a_tests outside a function, then you are.

nathan rogers

22:35

a←('a' 1)('b' 2)
is just fine for a
but now c
c ← values in the first column paired with the sum of the 2nd column

I mean, I'm not, IT is

I'm just editing text

Adám

@nathanrogers Well, obviously it is the interpreter that executes APL, not you, but you are relying on it to do so.

nathan rogers

I'm relying on it

Adám

Namespace scripts are for declaring namespace members.

nathan rogers

to read text files

Adám

@nathanrogers And do what with them?

nathan rogers

22:37

whatever it needs to in order to properly read the text thats in the files

ok

lets reframe this

Adám

May I ask what extension your text files have?

nathan rogers

.apln

HOW do I define a bunch of variable names known to a namespace such that I can just assign them arbitrary expressions?

Adám

@nathanrogers That's going to give you trouble because Link will think that .apln files contain namespace declarations. How about using .txt for text?

nathan rogers

does link read .txt files for the purpose of importing source code to be executed in the session?

Adám

@nathanrogers I don't understand.

@nathanrogers No, not by default.

nathan rogers

22:40

right

THATS whiy

rak1507

you want to use APL like any other language where you can put things in files and run them? too bad :D

nathan rogers

what's not to understand. I have variables that depend on other data, and I need those variables to have those values

and those varaibles are scoped to a namespace

@Adám so HOW do I do that?

in such a way that it doesn't crash every time I make a text change

Adám

@rak1507 Yup, if that's what he wants, then tough luck. Wait for 18.1.

@nathanrogers You want to make changes to basic data, and have APL re-compute final data on the fly?

nathan rogers

i don't care when it does it that is of no importance to me, I don't care what is happening under the hood, I care that myNS.var has the value I expect

WITHOUT CRASHING

so HOW do I do that?

rak1507

you don't

nathan rogers

22:43

no

Adám

@nathanrogers Simple: Write the correct code!

nathan rogers

yes

so what is that code

obv

that is literally the question I"m asking

Adám

The one that will cause the effect you want.

Obv.

rak1507

@Adám have you considered helping

nathan rogers

no

Adám

22:44

@rak1507 Yes, I've considered it.

nathan rogers

because that code is crashing on import

Adám

Then it isn't the correct code. Obv.

dzaima

@nathanrogers what is your code

nathan rogers

:Namespace myNS
data←('a' 1) ('b' 2) ⍝ ... etc
var←{column 1 , sum col 2} ↑data
:EndNamespace

Adám

@nathanrogers So when you fix the namespace. the 3rd line is processing the data from the second line.

That's more than just declaration.

That's running code.

Don't do that at fix time.

nathan rogers

22:46

again

how do I write it

Adám

Namespaces are for declaring members.

nathan rogers

such that I'm not doing something that causes a crash

Adám

Not running code.

nathan rogers

I need var to have that data

Adám

May I venture to say that you need the name var to return that data?

nathan rogers

22:47

that's what I would expect from the above

because its valid APL

Adám

Try:

:Namespace myNS
data←('a' 1) ('b' 2) ⍝ ... etc
∇var←var
 var←{column 1 , sum col 2} ↑data
∇
:EndNamespace

@nathanrogers Clearly it isn't.

nathan rogers

clearly it is, but its a problem with loading the files as namespaces

dzaima

@Adám what i expected. That'll re-evaluate that on every instance of var which is usually stupid

ngn

@nathanrogers are you using linux?

nathan rogers

windows

Adám

22:48

@dzaima But that's fairly easy to fix.

dzaima

@nathanrogers you're loading code, not a namespace

Adám

@dzaima ^ 1000×this ^

ngn

@nathanrogers sadly, no #! there

nathan rogers

you keep telling me what I'm doing

I'm not doing it

I didn't design the thing that's doing the thing

rak1507

@Adám hacky solutions to fix hacky solutions to fix...

nathan rogers

22:49

IT is doing it

I'm not against a workaround

but this isn't it

Adám

@nathanrogers You asked for it with :Namespace

@nathanrogers Did you try my code?

nathan rogers

I mean, that's because I need to namespace the data

it isn't a matter of it working or not

can I define a DFN like so:

dzaima

@nathanrogers you intentionally wrote code that executes stuff. Namespace files are for declaring stuff, not executing it

Adám

@dzaima 1000×this ^

nathan rogers

load←{
myNS.var←{evaluation}↑data
}

Adám

22:50

@nathanrogers Yes, and that won't execute at fix time. So that's fine.

nathan rogers

ok

do I need ∘←

Adám

No. Never.

nathan rogers

fine

@dzaima obviously I wrote code that executes

that's what you do with programming

dzaima

@nathanrogers but that's not what a namespace file is for

rak1507

there shouldn't be a fundamental difference between a ← {⍵} and a ← somevalue

ngn

22:51

@dzaima why does it allow initializers then?

dzaima

you may not like that, I may not like that, but that's the way it is

nathan rogers

@Adám in that load dfn is ⎕THIS still known?

dzaima

@ngn no clue

Adám

@nathanrogers Yes, ⎕THIS is always known.

nathan rogers

let me rephrase

if load is defined inside myNS will ⎕THIS reference myNS

Adám

22:53

@nathanrogers Yes, what's the definition of ⎕THIS.

nathan rogers

in other languages, this changes context inside of bodies

so I was clarifying

Adám

@nathanrogers bodies?

nathan rogers

function body, class body, blocks of code

Adám

Ah, but ⎕THIS just refers to the current namespace, nothing else.

It can never refer to a function.

nathan rogers

that's good

ngn

22:55

@Adám it should've been called ⎕somethingelse

nathan rogers

I was planning to optionally load tests anyway, so this just means I don't have to do that later

Adám

@ngn Agreed.

nathan rogers

wait

Adám

Naming is apparently hard.

@nathanrogers We are waiting.

nathan rogers

so you're saying that its fine so long as I don't modify the expression?

i can link it, but don't change the values

Adám

22:56

No, I'm not saying that.

nathan rogers

well

then your axiom doesn't hold

Adám

What axiom?

nathan rogers

because elsewhere I'm assigning variables to the results of reading a file

the axiom that I shouldn't evaluate code inside namespace

Adám

@nathanrogers I have no clue what that sentence means.

@nathanrogers ~~What‽~~ Oh.

nathan rogers

i01 ← ⊃ 2015 #.day 1

this reads text from a file

ngn

22:57

@nathanrogers in soviet russia variables assign to values :)

dzaima

@Adám presumably :namespace \n abc←⎕nget'something' \n :endnamespace

Adám

@dzaima That looks like a truly terrible idea.

dzaima

@Adám but why not do it? it's initializing constant data, just not as constant as you may like it to be

Adám

@dzaima Code like that would fail to load if its companion file is missing.

nathan rogers

but the companion file is part of the project

Transcript for