The Nineteenth Byte

Seggan

00:07

firing up the vm for the first time now

Ginger

did it crash

I bet it did

~~smh yall should've used 0xD00D1E as the magic number~~

Seggan

@Ginger does a compilation error count

Ginger

yes

Seggan

it only did so bc i renamed a function and forgot to update a usage

Ginger

lol

Seggan

00:09

i am currently stepping through it

Ginger

imagine having a working debugger, couldn't be me

Seggan

yes, the first run is a detailed debug, make sure Everything Runs Right™

Ginger

*screams in VSCode*

Seggan

vscode has a debugger too

no?

Bbrk24

I also use VS Code but it's worse on Windows because gradle is incompatible with MinGW bash so I have to use PowerShell

@Seggan It has a debugger but you need an extension to actually use it, it doesn't work by default

Seggan

00:10

erm i forgot to add the LOAD opcode

Ginger

^^, and for some reason the Kotlin plugin's debugger doesn't work right even though I compiled it from source

rest of the plugin works fine tho

Seggan

and apparently i malformed my program?

yes i did, its having a constant label of PRINT instead of STRING lol

Bbrk24

@Seggan point of clarification: Does the unpacker support unpacking things like 00 (fixint 0) and cc 00 (uint8 0) as Int, or do I have to encode it as d2 00 00 00 00 (int32 0)?

Seggan

it supports everything msgpack does

so, yes

ok, running it for the second time, wish me luck

Bbrk24

I wasn't sure if it would require you to unpack the first two as Byte instead of Int

Seggan

00:17

no it wouldnt

as you can see in the hexdump above

Bbrk24

Got it

What is msgpack's official C/C++ library doing? There are three main branches with unrelated histories: master contains the license and a README directing you to the other two branches, c_master contains the pure C implementation, and cpp_master contains the C++ implementation

Ginger

msgpack's a clever idea but the execution is too iffy for my tastes

Seggan

mhm, time for a third attempt. the last one failed because the base enums werent being classloaded in time, so they werent being registered

so im just registering them manually before execution

Bbrk24

class load order problems? oh no...

Seggan

nah, it was me forgetting that enums dont exist until you reference them :P

Bbrk24

00:23

they what

Seggan

classloading thingies

@Bbrk24 as in, they arent loaded or anything till you reference them

ok, 4th run

i accidentally made LOADC take 0 args

i can confirm programs are being loaded properly

but do they run?

Bubbler

> LOADC: loads random garbage

Bbrk24

Right now I'm honestly considering which would be harder: installing cmake or reimplementing msgpack from scratch

Seggan

@Bubbler yep

Bbrk24

I don't need the whole of msgpack -- the array, bin, and float types are completely unused, for example, and I only need an encoder, not a decoder

Seggan

00:28

@Bubbler idk why they call a garbage dump a "constant pool"

YES CGR HELLO WORLD WORKS

Bbrk24

YO

Seggan

0000: de ad c0 de 03 a5 50 52 49 4e 54 a5 4c 4f 41 44   ......PRINT.LOAD
0010: 43 a6 53 54 52 49 4e 47 01 02 ad 48 65 6c 6c 6f   C.STRING...Hello
0020: 2c 20 57 6f 72 6c 64 21 02 01 00 00               , World!....

slightly modified from the previous version

ya think i should make an asm for this for easy debugging?

@Bbrk24 for small changes when reviewing u can make a commit request

Bbrk24

Oh I forgot about that

Seggan

easier to do than me physically making it

Wheat Wizard

01:54

Ok, I was kind of thinking the whole flag thing was a little bit silly. But I actually just looked at a few of the Vyxal answers for the first time and it immediately drew up a knee jerk reaction. I conceptually understood what was going on, but I did not realize how just like infuriating it is. I don't think it's changed my opinion on anything, I still stick with the philosophical angle that I had. But it is definitely a different perspective.

I guess the big thing is that it feels like Vyxal is trying to cheat, and it feels like noone really minds. That feeling of unfairness, is something I didn't really expect.

Everything here, is obviously my personal feeling (not even really an opinion), not taking a stance on this "as a moderator".

ATaco

02:06

I don't think it's too bad that Vyxal does it, under the philosophy of each language is competing within its class, rather than against every other language.

Also, I didn't think the flags changed that much? mostly just IO stuff

Bbrk24

~~Except for when the accepted answer is "Vyxal s, 0 bytes"~~

ATaco

Eh, what does a 0 byte answer gain?

Unrelated String

yeah some of the flags are basically pre/appended commands lol

ATaco

Don't forget, Stop upvoting Trivial answers.

Wheat Wizard

02:21

@ATaco I guess the infuriating part is that it feels like that's not really the philosophy people are operating under.

ATaco

Fair, What can you do?

Restricting flags makes some languages practically unusable.

Wheat Wizard

Yeah, exactly. All I can do is downvote, and that is all I do.

ATaco

Demanding C users include their -l... would be waay too punishing.

Correct answer. Vote with your votes.

Bbrk24

@ATaco What? I can literally clang main.c and get an executable with no flags

even if I #include <stdio.h> and whatever

ATaco

Depends the required libs I suppose.

Bbrk24

02:24

I've never had to mess with include paths

Wheat Wizard

A better example is Klein which won't run unless you pass it one of 12 3 character flags.

ATaco

I've had to with GCC and windows.h, and also openGL in the past

Bbrk24

Okay that's fair

ATaco

The more I use C the more I enjoy the language.

Wheat Wizard

C is enjoyable to use. But I also hate it.

Seggan

02:56

@WheatWizard yes

Bbrk24

I have had a Thought about cgr: when/how are we going to flesh out the bytecode? What operations will we include?

New Posts

0

Q: Great Leap to the Right

Given a string s consisting of characters a-z lowercase, generate 5 arrays a1, a2, a3, a4, and a5 of size n (length of s) where the i-th element of each array represents the index of the next occurrence of the i-th character in s, with the following conditions: If no such occurrence exists, set ...

code-golf

lyxal

Would it be possible for cgr to provide a library/utility for generating lexers and parsers?

Bbrk24

03:17

(I have the parsing step down for Trilangle, the hard part is going to be converting bytecode formats)

lyxal

wouldn't it just be a matter of writing what you currently have, but just as bytecode instructions?

transpiling is usually easier than writing an interpreter

Bbrk24

Well, Trilangle bytecode has instructions inline, like assembly. CGR bytecode has all the constants clumped at the beginning. So, some instructions may need to be more than one, and I have to somehow account for that in my jump instructions

Perhaps not difficult, I've just never had to do this kind of thing before

lyxal

do you mean global state may be changed between function calls?

Bbrk24

Function calls? Oh no, Trilangle isn't nearly that high-level

lyxal

well what is it about the constants at the beginning that is troublesome?

Bbrk24

03:26

Maybe it won't be that difficult, it's just a problem I haven't had to think about before

The real problem is that some instructions will become more than one and others won't, and the fact that I have to account for that in my jump offsets

lyxal

will CGR have labels?

Bbrk24

Seems pretty unlikely given the bytecode format. @Seggan Are there any plans for that?

lyxal

@Bbrk24 if python can manage gotos in bytecode, so can cgr :p

Bbrk24

Yeah, but the current implementation makes it seem like gotos will be more like those of machine code than those of C

lyxal

well that seems okay. Just group things that are more than one command into some sort of block, have a label to the block as a whole and jump to that if needed.

Bbrk24

03:34

I might also just pad it with NOPs so that all instructions are the same number of CGR bytecode ops

Given the tendency for disassembled CGR bytecode to contain an excess of NOPs (mirrors, skips, and some branches are represented as NOP) that may not generate the most efficient code but it is an option

lyxal

Have a group command that allows you to either execute all or none of the commands in a group

And you can't jump to inside the group, only above or below

Bbrk24

The other option is that I store the CGR bytecode in a temporary buffer and do some work on that buffer to resolve gotos

(that is either going to be straightforward or impossible. I have no idea which.)

lyxal

label(characterN): group { doX, doY, doZ....} ... jump(characterN)

Bbrk24

Oh no
I hadn't even considered what the threading operator would look like in CGR bytecode

github.com/bbrk24/Trilangle#threading look at this entire mess

lyxal

Is it internally implemented as multiple threads?

Bbrk24

03:43

No. I forgot to document it there but it guarantees that ticks are synchronized between threads, so the interpreter actually just keeps a vector<thread> and calls .tick() on each one in a for loop

What I mean is that I have to reconcile that with other languages' implementations of concurrent processing somehow

If you want to look at Trilangle's implementation, github.com/bbrk24/Trilangle/blob/master/interpreter.cpp#L19 . Could probably be improved/optimized a fair bit though

case status::waiting should probably be moved into its own subroutine, now that I look at it

lyxal

Sounds like a discrete event simulation could be useful

Bbrk24

04:55

I just published a feature to my online interpreter that lets the program be stored in the URL. If I go back and edit all my Trilangle answers to use this, I'd flood the front page, so I don't think that's a good idea. Should I do it in batches? What's the best way to approach this?

lyxal

How many answers will be impacted?

Bbrk24

17. I've already updated one of them, though

lyxal

05:14

I would recommend a batch size of 7

Bbrk24

And how long between batches?

lyxal

05:31

Probably just until someone else makes a few edits or answers a few questions

Bbrk24

Fair enough

lyxal

There should probably be a meta proposal, because I don't think anyone really knows what to do in these situations

Not a proposal, a discussion

New Posts

0

Q: How to Handle Mass Edit Events

Every now and again, links to online interpreters will change, answer headers will become outdated and other circumstances will occur that impact the contents of a non-trivial amount of posts. There's nothing wrong with this - things change and are modified over time. Given that a whole bunch of ...

discussion editing

lyxal

Something like that

Because this also applies to things like mass question retagging or mass grammar changes or other things like that

mousetail

09:44

Random dumb idea for my practlang, what if in addition to ++ and -- you could do ** to multiply by 2 in place, // to divide by 2, and %% to mod 2

Adám

Makes sense.

lyxal

@mousetail go further and make it so that all operators can do this

Adám

What other operators are there?

mousetail

I guess bitwise ones but && and || already have another meaning

Adám

oof

mousetail

09:49

Suppose they didn't though, what should the default right value be?

Adám

Bi-glyphs should be limited to the minimum

mousetail

Also << and >> I guess, could be <<<< and >>>>

lyxal

@mousetail depends on the operator.

mousetail

For &&?

Adám

@mousetail Yes, and in general, such implicit arguments are just more to remember.

lyxal

09:50

@mousetail Well what does & do

mousetail

bitwise and

Adám

Just support var+=1 etc. for all operators and get rid of the strange 1-arg syntax

mousetail

That would be the sane option though

Adám

> practlang

Practical languages should be sane, imho.

mousetail

I'll probably not implement these things but theoretically how would they be used

lyxal

09:51

@mousetail something that makes a number like 16 or 8 bits at most

@Adám my suggestion to make it for everything was mostly in jest anyway

mousetail

I might make a fork of my practlang with lots of dumb useless stuff

esolang variation

Adám

APL doesn't have the var++ and ++var but all operators, even user-defined ones, can be used in the var f= arg syntax.

lyxal

More languages should support that

It's really cool

Vyxal 3 has arg f#>var for it

mousetail

Actually practically I might implement ???? that, for a type that implements default, assigns the default to a nullish value

Adám

Slightly hard to read, imo.

Should use ⁇ instead of ??

lyxal

09:57

@mousetail could I have an example? I'm not quite getting this

Adám

If myvar is a declared but unassigned boolean, myvar???? is equivalent to myvar=false

lyxal

Oh that's cool

Why not just do it at variable creation though?

Adám

If myvar is num, then it becomes 0, if string, it becomes ""

@lyxal Maybe you're getting it from some external source.

mousetail

Something I recently did in PHP, $arr??=[] except then you could write it shorter since arrays have a default value

Adám

But I don't see how this can work if everything isn't strongly typed.

lyxal

10:01

But the default is already defined for the type

mousetail

@Adám It is strongly typed

lyxal

So not assigning a value should set the default at creation

mousetail

This would redeclare a Option<T> as a T

Adám

@mousetail Can you import JSON?

mousetail

There will probably be a standard library module for JSON yes

Adám

10:03

But how do you infer types?

lyxal

@mousetail then use the default when getting the value

No need for an operator

Adám

E.g. what is the type of myvar = {foo:"bar"}.baz ?

mousetail

A enum of all all possible JSON types, like rust

Though normally you'd load JSON into a predefined structure

Adám

But what does myvar???? do?

mousetail

myvar in your example is not a Option<> so you couldn't use it

Adám

10:05

Will you have both var++ and ++var?

lyxal

@mousetail So then how about myvar = Some({pretend this is json I'm on mobile}.baz)

mousetail

@lyxal If the JSON enum defines a default value, that. So probably Some(JsonValue::null)

lyxal

fair enough

Adám

a????????b should be valid syntax, right?

mousetail

@Adám Yes

So you could do this:

if (value ????) {
    log_warning("value was null, corrected");
}

lyxal

10:09

@Adám what would the b do?

mousetail

@Adám It would fail the type checker since a???? is not nullable anymore, and ?? requires a nullable type

Adám

Oh.

lyxal

@mousetail but what if it was a double nested option?

mousetail

Hmmm true

lyxal

Not that you'd probably see something like that in a sane production system

But still

mousetail

10:11

Imagine a???????????? when unwrapping a tripple nested Option

lyxal

It would greatly reflect the questions the programmer has as to why there's a triple nested option

Adám

Add (?n) syntactic sugar so (?6) means ??????

mousetail

That would further encourage terrible practices, I love it

lyxal

@Adám Go all the way and make that work for every operator and syntax

Adám

Yesss!!! I mean Ye(s3)(!3)

mousetail

10:14

Then you could just remove normal addition, just do a(++b)

lyxal

(n(o)100)100) - the screams of programmers seeing this syntax

3

Totally worth it btw

Sandbox Posts

11:59

0

A: Sandbox for Proposed Challenges

Isolate the binary signals There are n binary signals, with intervals of 1 through n. For example, for n=3 the lists could be [1], [0, 1], [1, 0, 1]. Each list represents a repeating pattern. So [0,1] is really [0, 1, 0, 1, 0, 1, ...] The signals all start at the same time, but they are sent over...

lyxal

12:12

LDQ: What could sorting a number do that isn't sorting the digits?

Ginger

@lyxal sort by number of holes? :p

topology go brrrr

lyxal

somehow, I don't think that'll make it past PR review :p

Ginger

I'd approve it

lyxal

wait it's not even my PR

that's right

Ginger

time to go cry about rabbit

lyxal

12:19

I'm asking to make a suggestion

Ginger

specifically time to cry about Kotlin not letting me put whatever I want in annotation parameters

*angry static typing noises*

guess my built-ins only get to return primitive types :|

really loving the fact that I now have over 100 errors in my code

mousetail

@lyxal Very out there suggestion but it could return the next smallest power of 10, IDK why I associate that with sorting

Ginger

well now I've got a bonus problem

currently the classes representing Rabbit objects have type as an instance property, as opposed to a static one

lyxal

@Ginger bonuses in code golf questions are usually discouraged.

@mousetail I like that

Ginger

I'm inclined to say this is inevitable because technically classes like FunctionObject are meta-classes

so their types can't be fixed

but this means I can't figure out the Rabbit return-type from the Kotlin function signature

this is a symptom of ANOTHER issue, which is that I can't write Rabbit structs from Kotlin code; only functions

I'm getting kinda tired of this :|

really what I need to do is make a distinction between classes-with-known-type (like primitives) and meta-classes-with-no-known-type (like functions)

because I know the type of the primitives! it doesn't change!

well actually reviewing my code I actually can make Rabbit objects from Kotlin... just not structs

I can make things that behave as instances of structs but not the structs themselves

this is a problem because it creates a divide between "struct" and "object"

I guess I get to rejigger some things

really I should be able to construct a RabbitObject from a ResolvedType

so, here's what I'll do:

ResolvedType will get a new fields property, which will contain a Map from Identifiers to ResolvedTypes

upon constructing a RabbitObject from a ResolvedType, it'll automatically fill out its Context with RabbitObjects created from the type's fields

cool and good

although actually I'm not sure that fields should be a member of ResolvedType

maybe just ResolvedStructType, because structs are what have fields

lyxal

12:33

Just gonna interject with something I just experienced: gotta love how flask hosts on port 5000 while the http-server node package hosts on port 8080

Ginger

new problem: how do I fill out fields during resolution?

...or initialization?

I suppose that's the whole point of the resolution system

hmm, yeah

mousetail

@lyxal This is actually super convenient, it means you can run both at the same time

Ginger

during resolution the fields property will be filled out

lyxal

@mousetail it means I have to fight the URL autocomplete

and besides, I can't

mousetail

That's a pretty minor issue overall

Also it's nice that this prevents caching and service worker sharing between the two

lyxal

12:36

in my case, there'll never be flask and node running at the same time

mousetail

You can explicitly set the port

Ginger

really my first big red flag should be the fact that I have a file named PrimitiveObjects.kt

when they should be primitive types

Seggan

@Bbrk24 any that we need -- common golfing operations

@Bbrk24 i was thinking on it yesterday. i think yes, we should have at least numerical labels

Bbrk24

Numerical labels works perfectly for me.

Ginger

oooooooooh kotlin reflection lets me just create types out of thin air

kinda

okay, how should I turn types into objects? a builder function?

(@Seggan you may want to read through most of the above)

Bbrk24

12:46

idk about the JVM, but the Objective-C runtime let you register new classes, or even new methods on existing classes, at runtime

Ginger

._.

yes I think builder function

Bbrk24

@lyxal It turns out that stackexchange only lets you edit 10 posts in a day

Ginger

at this point I will probably have to rewrite a large majority of Rabbit to work with the massive numbers of breaking changes I've added

:|

lyxal

@Bbrk24 huh

interesting

don't think I knew that one

Bbrk24

So now I have to wait until "tomorrow" (i.e. this evening) to finish the second batch

New Posts

13:04

0

Q: Shortening this Code to process nested tuple even further without the use of max()

I have the code: marks = tuple(tuple((input("Enter subject: "), input("Enter marks: "))) for _ in range(int(input("Enter number of tuples: ")))) print(tuple((s[:3].upper(), max(int(m) for t, m in marks if t == s)) for s in set(t for t, m in marks))) #INPUT (("Physics",131),("Chemistry",30),("Phy...

code-golf code-challenge array python

Seggan

13:57

@Ginger i feel like you should rewrite rabbit

@Ginger thats a symptom of bad code

reflection should be used sparingly

Ginger

calm down :b

@Seggan I'm not actually using reflection for that, I was just observing it was possible

@Seggan I am rewriting it

Transcript for