LLVM Native Codegen & Linking

Prady features a full native compiler backend utilizing LLVM (Low-Level Virtual Machine) infrastructure to generate standalone, highly optimized machine code.

Compilation Pipeline

The Prady compiler transforms source files through sequential lowering phases:

Source Code (.pr)
      │
      ▼
Lexer & AST Parser
      │
      ▼
Semantic Analysis & Type Checking
      │
      ▼
Architecture & Policy Validation
      │
      ▼
Prady Intermediate Representation (PradyIR)
      │
      ▼
LLVM IR Text Emission
      │
      ▼
LLVM Optimization Pipeline (-O1, -O2, -O3, -Oz)
      │
      ▼
System Linker (MSVC link.exe / LLD / Clang / GCC)
      │
      ▼
Native Executable (.exe / ELF / Mach-O)

Inspecting LLVM IR

You can view the exact LLVM IR emitted for any Prady source file using prady emit-llvm:

prady emit-llvm src/main.pr

Example output:

; Prady compiler - Module: main
target triple = "x86_64-pc-windows-msvc"

declare i32 @printf(i8* nocapture readonly, ...)
declare i8* @malloc(i64)
declare void @free(i8*)

define i64 @add(i64 %a, i64 %b) {
entry:
  %t0 = add i64 %a, %b
  ret i64 %t0
}

define void @main() {
entry:
  %t1 = call i64 @add(i64 10, i64 20)
  %t2 = call i64 @print(i64 %t1)
  ret void
}

Cross-Compilation & Target Triples

Prady supports targeting multiple host and embedded environments: