What are the existing solutions for creating object graphs in an interpreter?

Question

Please let me know if this isn't the right SE site (or otherwise) for this kind of question, it's the best match I could find.

I'm working on a humble interpreter written in C for a simple language I'm designing. I ran into a design problem in the interpreter which I'm sure people have already solved, and I would like to learn about the existing solutions to this.

Consider there's a builtin String type in the language. And in the interpreter each object has a attributes hash table associated with it. It maps a String* key to a Object* value.

Where do I have problems? Each String instance should have methods which operate on it: E.g. .length(), .at_index() etc.

The way I thought to implement this, is have the C function which is responsible to create String objects, also put the methods (Function objects) in the new object's attributes table, under the matching String keys.

This crashes the interpreter (the C program itself) with a Stack Overflow, because it means creating a new String has to create a new String, and so on...

I can probably work around this issue, maybe by specifically caching Strings (which is a good idea anyway), but I get the feeling that maybe this design is problematic in the first place.

Is it? How do existing language implementations, mainly simple ones, accomplish this?

Answer 1

Your interpreter sits outside the language. It doesn't have to abide by the rules of the language. If it cannot initialize a property of an object because that property requires something else that doesn't exist yet, just don't initialize it, and wait until whatever you need to initialize it exists.