Question

So I'm writing a little interpreter in C at the moment for a language I have created (which is pretty similar to Python). I have written the lexer and parser and currently my program outputs an AST, an now I am attempting to turn this AST into bytecode. Currently my algorithm traverses the AST (depth-first) and can generate bytecode for simple arithmetic, and now I am trying to implement if statements.

I cannot copy all my code in here because it's a pretty large amount of code, but currently the program takes an AST which might look something like

ADD
|-- 1
|-- MUL
    |-- 2
    |-- 3

and turns this into

LOAD 1 //the real code doesn't put the value here, but a number representing the position of this value in an array
LOAD 2
LOAD 3
MUL
ADD

This is easy for simple expressions but I really don't know how to generate the bytecode for an if statement. I know that I will have to jump to the else clause if the comparison is false, and also jump from the end of every if/else if block, but how do I deal with this if the jump is more than 256 bytes of bytecode?

Answer 1

您应该阅读SICP，Dragon Book，然后Lisp In Small Pieces。

我希望你能重新设计你的字节码。然后你可以有一些FARJUMP字节码，然后是四个字节a，b，c，d（考虑作为uint8_t无符号整数，每个8位），你将跳转到(a<<24) + (b<<16) + (c<<8) +d偏移量。

你可能希望能够前后跳跃。要么BACKFARJUMP要向后跳，要么使用一些签名的偏移...

使用这样的操作码，您将能够跳转到超过40亿个字节码（2 ³²）。这可能更容易。

如果40亿字节码偏移量不够，可以概括。

不要忘记您的计算机不可能有超过1TB的RAM（并且这样的计算机比汽车的成本更高）。

Converting an AST to bytecode

1 个答案: