Rework struct & class syntax

[ctxcode]

New syntax for classes, structs & traits.

class A {
    // properties
    my_pub_prop: i32 = 5;
    - my_private_prop: i32;  // Private
    ~ my_readonly_prop: i32; // Readonly

    // Trait
    trait MyTrait;

    // Functions
    fn get_prop() {
        return this.my_pub_prop;
    }
    static fn get_prop() String {
        return "";
    }
}

trait MyTrait {
    new_prop: u32;

    fn another_func() {
    }
}

extend A {
    // functions & traits
}

- for private and + for readonly. extend to add more functions to a class.

[sparky4pro]

Thank you for including us in the discussion.

Question: the first property in the struct has neither - nor +, so what is it?
read and write?

IMHO + for readonly seems weird as functionality of this property is subtracted not added...
On the flip side - for private seems adequate.

The best approach would be to answer the question:
Is Ki a safety oriented language?
If the answer is yes then a normal struct field/property should be private.
This would result in the following:

struct A {
    // properties
    + my_pub_prop: i32 = 5; // Public readable and writable
    my_private_prop: i32;  // Private
    - my_readonly_prop: i32; // Readonly
}

How do you want to differentiate private and public methods?

static for a keyword is fine, but the question arises:
Why mix things up like this:
private, public, readonly with '1 special char' but use a 'keyword' for static?
It doesn't seem consistent.
IMHO finding a 'proper char' that resembles static would be a better choice (like $ if not already assigned for something else).
Or increase the number of keywords and use proper keywords for everything...

[ctxcode]

sorry i've been changing my example alot.

so by default it is public. i changed + to ~ although, i dont know how easy it is to type on mac keyboards.

rust is safe and it doesnt have private properties.

- would also work on functions

i use signs private & readonly so the parser doesnt need to be confused about it being a property name or access type. The compiler could probably figure it out using private / readonly... It's just an idea and it requires less typing.

[sparky4pro]

Rust is safe and it doesn't have private properties.

Shame on them :D

This is a design decision, like if a variable is always immutable or not...
So, you can decide to do it differently than Crab/Rust.
:P

[ctxcode]

true, that's why i have private/readonly, because i think they are useful.

i think private by default will cause more problems. e.g. a package maintainer that didnt make something public for no good reason. It's easier to think about what needs to be private than think about what needs to be public (in my opinion)

[ctxcode]

also, something not being public is a bigger issue than something not being private.

[ctxcode]

I think i will settle down on this syntax:

class A {
    my_public_prop_1: i32;
    my_public_prop_2: i32 = 5;
    - my_private_prop: i32;  // Private
    ~ my_readonly_prop: i32; // Readonly

    trait MyTrait;

    fn my_public_func() {
        return this.my_pub_prop;
    }
    - fn my_private_func() {
        return this.my_private_prop;
    }
    static fn my_public_static_func() String {
        return "";
    }
    - static fn my_private_static_func() String {
        return "";
    }
}

trait MyTrait {
    // properties & functions
    new_prop: u32;
    - new_private_prop: u32;

    fn another_func() {
    }
}

extend A {
    // functions & traits
    // traits are not allowed to add more properties when using extend
}

[sparky4pro]

~ is better than + for readonly.
Suppose the readonly properties would be public?

[ctxcode]

yes, but you cant change value of the property. e.g. the port of a running http server or a length from a buffer.