Allow "@" character as a prefix to enable keywords to be used as identifiers.

[Hyrtwol]

Idea from C#
Allow "@" character as a prefix to enable keywords to be used as identifiers.
https://learn.microsoft.com/en-us/dotnet/csharp/language-reference/language-specification/lexical-structure#643-identifiers
e.g. keywords like

@packages
@context
@map

Useful for json serialization so you don't need to add attributes to map the fields.
C# example

    [Fact]
    public void KeywordsAsIdentifiers()
    {
        int @int = 666;
        Assert.Equal(666, @int);
        Assert.Equal("int", nameof(@int));
    }

not sure if using @ will clash with the Odin syntax

@(...)

if so another char could be chosen ofc

[JesseRMeyer]

int : int = 0 is already valid Odin syntax. So at least shadowing of type names is already a feature. I'd be shocked if shadowing of keywords more generally becomes supported.

[Hyrtwol]

Oh didn't know int : int = 0 was allowed but that looks like a tough debugging session if you're not careful ;)
Tbh I would also be shocked to see this in Odin but you never know, maybe down the road. Can't remember exactly when it was added to C# but pretty sure it was not in version 1.0 :p
Also I don't think you can do that trick with all keywords, this don't work

map : int = 0
context : int = 0

[perrylets]

Can't you use another character like _?

[flysand7]

_ is a black-hole variable, it consumes everything, so it is a special name which you can't declare too.

[perrylets]

I meant prefixing the identifier with _

[flysand7]

I'm pretty sure our JSON marshaller supports alt-names for fields and it can understand when a field in a struct differs from the encoded field. That said, serialization is the only place where this syntax is useful. If you were fine with @foo being special, maybe this isn't too bad:

JSON_Entity :: struct {
   name: string,
   value: f64,
   _context: string, `json:"context"`
}

To specify a different encoded field name. The benefit of this feature doesn't extend beyond the problem of serialization and can't be abused like the @kw thing.

[gingerBill]

I know this is an old post but there is zero need for this whatsoever. Why do you need to use a keyword as an identifier? Every place where I think people would think it is necessary is the following:

• foreign procedures/variables. @(link_name=<string>) exists so that the link name and the usage name can be separate
• Struct fields. The struct field name doesn't have to match some foreign code, and if you need it to be different for serialization/marshalling purposes, use struct field tags.
• You just want to use that work. Just use a different one. context is usually written with ctx, or even _context or context_.