How to implement the execution of a language that I am writing?
I am writing my own programming language using C# just for fun. It's called SPL. It has only 12 keywords/commands/instructions and nothing more. I know exactly what each command is going to do. This is the class diagram:
This is not finished of course. But the idea is, we have a runtime ISplRuntime. All the things that the commands do, is done on the runtime. For example, the OutputCommand outputs the stuff in ISplRuntime.Current (This is just an object) to ISplRuntime.Output (This is a TextWriter).
So far so good. But then I need to implement the IGotoCommand which is just like the goto keyword in C#. I am going to make three kinds of gotos. But I cannot figure out how to change the flow of the program.
Here's the ideas I tried:
Add a method called RunNextCommand() in ISplRuntime. And in the end of every command's Execute() method, call RunNextCommand. However, if the user writes a lot of SPL code, the stack trace will get larger and larger and result in a Stack Overflow.
Add a delegate to ISplRuntime. Each time a command has finished executing, Invoke() the delegate. The delegate will look for the next command to execute and call Execute() on that ICommand. But I am not sure if this will also result in a stack overflow. Will it result in a stack overflow?
I am still not very sure if this is the right way to do it. I just want an general answer about how to do this kind of thing as this is the first time for me to write a language. Also please tell me whether using delegates will cause a stack overflow.
Normally, you would just do a loop; within the loop, execute the command at the current line, then for all non-flow-control commands increment the current line. For goto, just set the current line to the parameter. There are no stack issues because the flow is not recursive - everything gets launched from the instruction loop (which would belong to the runtime, given your class diagram).