What's in an assembler?

How does a compiler work, anyway?

Lexer? Parser?

Difference between lexer and parser

Consider:

• fjshdfoilu gdiyuktf ggdbkifucqjuhzsiudmfppoe arzej
• now a sentence this like, but none sense makes

→ token ≅ word

Codegen

After parsing: semantics

Roughly, instructions for code generation

A (really tough) science and an art

Thankfully trivial for an assembler 🙃

Background

Born in 1997

Pile of 90's-style C code 🐛

Claim to fame: used for some commercial releases!

Very 90's design...

Design points

Macro assembler

Unlimited relocation complexity

Human-oriented

Tangent: two kinds of assembler

Sometimes at odds, but possible to marry both with enough effort

Let's witness some design lessons

...primarily negative.

PURGE

Metaprogramming facilities

Compile-time variables to store info

Uh-oh, no namespacing!

PURGE

Let's allow deleting symbols!

Metaprogramming

Metaprogramming

EQUS = compile-time eval

Imagine C macros, but there is no preprocessor

Fully text-based, bypassing tokens

→ Can't even lex a line until previous one is processed

Metaprogramming

Structs

RGBASM doesn't support structs

And now it does!

Brainfuck interpreter

Yes, really.

; hello world
	bf ++++++++[>++++[>++>+++>+++>+<<<<-]>+>+>->>+[<]<-]>>.>---.+++++++..+++.>>.<-.<.+++.------.--------.>>+.>++.
; echo
	bf \,+[-.\,+], HELLO WORLD!\n

(Image missing)

Language design

• Processing in passes = better
• Metaprogramming = good, actually

Design in general

• Simplicity is a lie
• Good to think about implications of a design
• Oh how far we've come