2.3 KiB
clox
Description
My notes for Crafting Interpreters
Table of Contents
Chunks of Bytecode
Why Bytecode
Goes over the choice of bytecode vs machine code / assembly.
Compiling to native instruction set the chip supports is what the fastest languages do.
However, this is extremely low-level and time consuming. Portability is just exponential complexity at that level.
We're exchanging performance for portability by writing an emulator in C.
That emulator is a VM or virtual machine to run that bytecode one instruction at a time.
Dynamic Arrays
Goes over his interface and implementation of dynamic arrays.
The struct contains a pointer to items of "type". Ex. uint8_t *data, void *ptr, etc.
It also contains the current length and capacity of the array.
The implementation contains a reallocate, initType, freeType, and writeType function.
It exposes three macros GROW_CAP(cap), GROW_ARRAY(type,ptr,old*type,new*type), and FREE_ARRAY.
GROW_CAP is how you handle growing your capacity.
FREE_ARRAY and GROW_ARRAY utilize our reallocate implementation.
Chunks
Now we can build dynamic arrays or "Chunks" of instructions or operations. These will be uint8_t *ptr. Which is the instruction number or Enum.
This gives us 255 available operations or instructions we can have.
Once we have instructions, we also want to be able to have constant values. This is another dynamic array inside chunk.
These type "values" are really the C primitive double type. 64 bits.
Instead of writing to it like earlier make a function to add new values and return the index so we can track it.
For line debugging we add an array of integers to act as a LUT with index of the operation. Each operation maps to a line.
To disassemble, we just need to write a disassemble chunk function. Instead of iterating or incrementing the offset, disassembleInstruction returns the next offset.
The offset of the first chunk is it's length. From there disassembleInstruction performs a switch statement on the operation or instruction to return the proper offset.
disassembleChunk prints out the "name" which is going to be used for the FILE, the chunk maps to a line, operation, and possible value.