Vyxal

lyxal

04:54

1

Nekomata + -e, 3 bytes Ɗ¦ů Attempt This Online! Ɗ¦ů Ɗ Decimal digits ¦ Divided by, and check if the result are integers ů Check if all elements are unique

This has been making me unhappy

3 bytes and we're stuck with 7

And tbh I've been thinking about whether there'd be some benefit to incorporating some form of non determinism

But I've come up with a better idea

Uno flip

Or specifically, a flip modifier that makes elements return a null or fail type for certain inputs

So like you might write flatten flip: int-div all-unique?

And the flip modifier would make div by 0 and non-int results a fail type

My current working name for the fail type is the nuh-uh type

Like that kinda nuh uh

It's a bad name I know but it's good for now

If a flip modifier was given a function or modifier, flip mode would be applied to everything in the function

So I suppose it'd be a context thing too

And there could also be a flag to have everything flipped by default

The name flip comes from how uno flip has double sided cards

And you can switch which side gets played on with a flip card

Hence why it's called the flip modifier

lyxal

08:20

I think a nuh-uh type would also be beneficial for knowing when a thing has failed or whether it's just returned an empty value

Or even to have a simple way to know that something has errored without actually having errors

The_AH

A null type?

Nuh-uh seems like an interesting name for a null type.

lyxal

It's not null, as such

It's a "hey this operation failed" type

Nuh-uh isn't the final name btw

It's a stand in name to avoid bikeshedding over what it should be called

The_AH

An error type?

Interesting I would suggest the name of "error" or "fail".

lyxal

That's the point of the temporary name

To avoid getting caught up in what it should be called :p

The_AH

Oh.

But I think "error" or "fail" make more sense.

lyxal

08:29

That's for later

If even ever

I don't know whether it's worth adding yet

The whole point of mentioning it was to discuss the concept and the proposed implementation

Because it'd require a bit of a code refactoring

The_AH

Okay.

lyxal

Which automatically means I'd need user's input

The_AH

So we could just do nothing.

And make the programmer check themselves.

So "is this a nuh-uh type?"

lyxal

There currently is no such type. And the idea is more "is it a golfy thing"

The_AH

Oh right.

But we can just make checking it easier.

Maybe one builtin just for that.

lyxal

08:32

I'm saying should the type even be added at all in the first place

Forget checking for it

The_AH

Yes.

It's a requirement.

lyxal

This specific proposal isn't something I've thought about in depth you know :p

I've yet to play test it outside of a single use case

The_AH

IT'S A REQUIREMENT.

A REQUIREMENT.

lyxal

It's not

The_AH

Well yeah.

lyxal

08:33

I haven't decided that it needs to be a thing

The_AH

Error handling could happen other ways.

lyxal

Meaning it's not a requirement

@The_AH this is code golf. Traditional error handling isn't a thing

The_AH

@lyxal Traditional, not all error handling.

lyxal

This idea is only because non-deterministic programs seem like they might be shorter sometimes

Error handling is not a priority.

This would be a neat side feature

The_AH

What would happen in a zero by division error then?

How would the programmer respond?

If you were building the smallest calculator possible for example.

lyxal

08:35

@The_AH it returns 0

No error thrown

The_AH

So that means you already have error handling.

Okay.

So errors are just 0 values then.

lyxal

No

Division by 0 is special cased

Because it's useful for golf

The_AH

What does that mean?

Special cased?

lyxal

Theres an explicit check for 0

The_AH

What special case?

What happens then?

lyxal

08:37

There's an if statement that says if rhs == 0: return 0

The_AH

OH.

lyxal

Operators can optionally implement checks for edge cases like that and return a default value

But error handling isn't really a thing outside of places where an error case can be turned into useful behaviour

The_AH

Okay that seems like an interesting way of error handling.

lyxal

The programmer has no control over it. If there's ever a runtime error, it's either strange data, or a bug on our end

The_AH

That seems like a horrible way of error handling.

lyxal

08:40

It would be if the main concern wasn't code golf

But this is code golf, so it's not a problem

The_AH

Yeah ik.

lyxal

Anyway , I need to flesh the idea out a bit more

The_AH

Yeah.

Vyxal Bot

12:36

Lyxal pushed a commit to Vyxal/main in Vyxal/Vyxal: Update pyproject.toml

@Ginger An error occured while processing event a68297c0-70d7-11ee-8440-19cdb2351af9!

lyxal

Ruh roh

that's like, one of the features that shouldn't be broken

Ginger

12:49

@lyxal LLVM has this! it's called "poison values" and it does effectively what you're suggesting

thanks chat

lyxal

Anyhow

Version 2.21.13

2

Given that the bot errored on it

Ginger

13:19

@lyxal shit, that was a release event?

lyxal

It was

Ginger

>:|

Transcript for