From 4c467b118d12ca6fabd018e4c0295096fa4d399b Mon Sep 17 00:00:00 2001
From: gingerBill <bill+github@gingerbill.org>
Date: Wed, 10 Aug 2016 20:05:45 +0100
Subject: [PATCH] copy(...)

---
 examples/basic.odin        |  36 ++-
 examples/main.ll           | 619 +++++++++++++++++++++++--------------
 examples/main.odin         |  46 +--
 src/checker/checker.cpp    |   8 +
 src/checker/expr.cpp       |  12 +
 src/codegen/codegen.cpp    |  11 +-
 src/codegen/print_llvm.cpp |  57 +++-
 src/codegen/ssa.cpp        | 244 +++++++++++----
 src/exact_value.cpp        |   7 +-
 src/gb/gb.h                |   7 +
 10 files changed, 710 insertions(+), 337 deletions(-)

diff --git a/examples/basic.odin b/examples/basic.odin
index df1ef29d1..c088b85f9 100644
--- a/examples/basic.odin
+++ b/examples/basic.odin
@@ -14,17 +14,18 @@ string_byte_reverse :: proc(s: string) {
 	}
 }
 
-encode_rune :: proc(buf : []u8, r : rune) -> int {
+encode_rune :: proc(r : rune) -> ([4]u8, int) {
+	buf : [4]u8;
 	i := cast(u32)r;
 	mask : u8 : 0x3f;
 	if i <= 1<<7-1 {
 		buf[0] = cast(u8)r;
-		return 1;
+		return buf, 1;
 	}
 	if i <= 1<<11-1 {
 		buf[0] = 0xc0 | cast(u8)(r>>6);
 		buf[1] = 0x80 | cast(u8)(r)&mask;
-		return 2;
+		return buf, 2;
 	}
 
 	// Invalid or Surrogate range
@@ -37,26 +38,28 @@ encode_rune :: proc(buf : []u8, r : rune) -> int {
 		buf[0] = 0xe0 | cast(u8)(r>>12);
 		buf[1] = 0x80 | cast(u8)(r>>6)&mask;
 		buf[2] = 0x80 | cast(u8)(r)&mask;
-		return 3;
+		return buf, 3;
 	}
 
 	buf[0] = 0xf0 | cast(u8)(r>>18);
 	buf[1] = 0x80 | cast(u8)(r>>12)&mask;
 	buf[2] = 0x80 | cast(u8)(r>>6)&mask;
 	buf[3] = 0x80 | cast(u8)(r)&mask;
-	return 4;
+	return buf, 4;
 }
 
 print_rune :: proc(r : rune) {
-	buf : [4]u8;
-	n := encode_rune(buf[:], r);
+	buf, n := encode_rune(r);
 	str := cast(string)buf[:n];
-
 	print_string(str);
 }
 
-print_int :: proc(i, base : int) {
+print_int :: proc(i : int) {
+	print_int_base(i, 10);
+}
+print_int_base :: proc(i, base : int) {
 	NUM_TO_CHAR_TABLE :: "0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz@$";
+	NEG :: "-";
 
 	buf: [21]u8;
 	len := 0;
@@ -65,17 +68,16 @@ print_int :: proc(i, base : int) {
 		negative = true;
 		i = -i;
 	}
-	if i > 0 {
-		for i > 0 {
-			c : u8 = NUM_TO_CHAR_TABLE[i % base];
-			buf[len] = c;
-			len++;
-			i /= base;
-		}
-	} else {
+	if i == 0 {
 		buf[len] = '0';
 		len++;
 	}
+	for i > 0 {
+		c : u8 = NUM_TO_CHAR_TABLE[i % base];
+		buf[len] = c;
+		len++;
+		i /= base;
+	}
 
 	if negative {
 		buf[len] = '-';
diff --git a/examples/main.ll b/examples/main.ll
index 9d5e274e9..412f9c49b 100644
--- a/examples/main.ll
+++ b/examples/main.ll
@@ -2,56 +2,197 @@
 
 %.rawptr = type i8* ; Basic_rawptr
 
-define {i64, i64} @tuple() {
-"entry - 0":
-	%0 = alloca {i64, i64}, align 8 
-	store {i64, i64} zeroinitializer, {i64, i64}* %0
-	%1 = getelementptr inbounds {i64, i64}, {i64, i64}* %0, i64 0, i32 0
-	store i64 1, i64* %1
-	%2 = getelementptr inbounds {i64, i64}, {i64, i64}* %0, i64 0, i32 1
-	store i64 2, i64* %2
-	%3 = load {i64, i64}, {i64, i64}* %0
-	ret {i64, i64} %3
-}
+declare void @llvm.memmove.p0i8.p0i8.i64(i8*, i8*, i64, i32, i1)
 
 define void @main() {
 "entry - 0":
-	%0 = alloca i64, align 8 ; a
-	store i64 zeroinitializer, i64* %0
-	%1 = alloca i64, align 8 ; b
-	store i64 zeroinitializer, i64* %1
-	%2 = call {i64, i64} @tuple()
-	%3 = alloca {i64, i64}, align 8 
-	store {i64, i64} zeroinitializer, {i64, i64}* %3
-	store {i64, i64} %2, {i64, i64}* %3
-	%4 = getelementptr inbounds {i64, i64}, {i64, i64}* %3, i64 0, i32 0
-	%5 = load i64, i64* %4
-	%6 = getelementptr inbounds {i64, i64}, {i64, i64}* %3, i64 0, i32 1
-	%7 = load i64, i64* %6
-	store i64 %5, i64* %0
-	store i64 %7, i64* %1
-	%8 = load i64, i64* %0
-	call void @print_int(i64 %8, i64 10)
-	%9 = getelementptr inbounds [1 x i8], [1 x i8]* @.str0, i64 0, i64 0
-	%10 = alloca %.string, align 8 
-	store %.string zeroinitializer, %.string* %10
-	%11 = getelementptr inbounds %.string, %.string* %10, i64 0, i32 0
-	%12 = getelementptr inbounds %.string, %.string* %10, i64 0, i32 1
-	store i8* %9, i8** %11
-	store i64 1, i64* %12
-	%13 = load %.string, %.string* %10
-	call void @print_string(%.string %13)
-	%14 = load i64, i64* %1
-	call void @print_int(i64 %14, i64 10)
-	%15 = getelementptr inbounds [1 x i8], [1 x i8]* @.str1, i64 0, i64 0
-	%16 = alloca %.string, align 8 
-	store %.string zeroinitializer, %.string* %16
-	%17 = getelementptr inbounds %.string, %.string* %16, i64 0, i32 0
-	%18 = getelementptr inbounds %.string, %.string* %16, i64 0, i32 1
-	store i8* %15, i8** %17
-	store i64 1, i64* %18
-	%19 = load %.string, %.string* %16
-	call void @print_string(%.string %19)
+	%0 = alloca [4 x i64], align 8 ; dst
+	store [4 x i64] zeroinitializer, [4 x i64]* %0
+	%1 = alloca [2 x i64], align 8 ; src
+	store [2 x i64] zeroinitializer, [2 x i64]* %1
+	%2 = getelementptr inbounds [2 x i64], [2 x i64]* %1, i64 0, i64 0
+	%3 = getelementptr i64, i64* %2, i64 0
+	store i64 7, i64* %3
+	%4 = getelementptr inbounds [2 x i64], [2 x i64]* %1, i64 0, i64 0
+	%5 = getelementptr i64, i64* %4, i64 1
+	store i64 5, i64* %5
+	%6 = sub i64 4, 1
+	%7 = sub i64 4, 1
+	%8 = getelementptr inbounds [4 x i64], [4 x i64]* %0, i64 0, i64 0
+	%9 = getelementptr i64, i64* %8, i64 1
+	%10 = alloca {i64*, i64, i64}, align 8 
+	store {i64*, i64, i64} zeroinitializer, {i64*, i64, i64}* %10
+	%11 = getelementptr inbounds {i64*, i64, i64}, {i64*, i64, i64}* %10, i64 0, i32 0
+	store i64* %9, i64** %11
+	%12 = getelementptr inbounds {i64*, i64, i64}, {i64*, i64, i64}* %10, i64 0, i32 1
+	store i64 %6, i64* %12
+	%13 = getelementptr inbounds {i64*, i64, i64}, {i64*, i64, i64}* %10, i64 0, i32 2
+	store i64 %7, i64* %13
+	%14 = sub i64 2, 0
+	%15 = sub i64 2, 0
+	%16 = getelementptr inbounds [2 x i64], [2 x i64]* %1, i64 0, i64 0
+	%17 = getelementptr i64, i64* %16, i64 0
+	%18 = alloca {i64*, i64, i64}, align 8 
+	store {i64*, i64, i64} zeroinitializer, {i64*, i64, i64}* %18
+	%19 = getelementptr inbounds {i64*, i64, i64}, {i64*, i64, i64}* %18, i64 0, i32 0
+	store i64* %17, i64** %19
+	%20 = getelementptr inbounds {i64*, i64, i64}, {i64*, i64, i64}* %18, i64 0, i32 1
+	store i64 %14, i64* %20
+	%21 = getelementptr inbounds {i64*, i64, i64}, {i64*, i64, i64}* %18, i64 0, i32 2
+	store i64 %15, i64* %21
+	%22 = getelementptr inbounds {i64*, i64, i64}, {i64*, i64, i64}* %10, i64 0, i32 0
+	%23 = load i64*, i64** %22
+	%24 = bitcast i64* %23 to %.rawptr
+	%25 = getelementptr inbounds {i64*, i64, i64}, {i64*, i64, i64}* %18, i64 0, i32 0
+	%26 = load i64*, i64** %25
+	%27 = bitcast i64* %26 to %.rawptr
+	%28 = getelementptr inbounds {i64*, i64, i64}, {i64*, i64, i64}* %10, i64 0, i32 1
+	%29 = load i64, i64* %28
+	%30 = getelementptr inbounds {i64*, i64, i64}, {i64*, i64, i64}* %18, i64 0, i32 1
+	%31 = load i64, i64* %30
+	%32 = icmp slt i64 %29, %31
+	%33 = select i1 %32, i64 %29, i64 %31
+	%34 = mul i64 %33, 8
+	call void @llvm.memmove.p0i8.p0i8.i64(i8* %24, i8* %27, i64 %34, i32 8, i1 false)
+	%35 = alloca i64, align 8 ; i
+	store i64 zeroinitializer, i64* %35
+	store i64 0, i64* %35
+	br label %"for.loop - 2"
+
+"for.body - 1":
+	%36 = getelementptr inbounds [4 x i64], [4 x i64]* %0, i64 0, i64 0
+	%37 = load i64, i64* %35
+	%38 = getelementptr i64, i64* %36, i64 %37
+	%39 = load i64, i64* %38
+	call void @print_int(i64 %39)
+	br label %"for.post - 3"
+
+"for.loop - 2":
+	%40 = load i64, i64* %35
+	%41 = icmp slt i64 %40, 4
+	br i1 %41, label %"for.body - 1", label %"for.done - 4"
+
+"for.post - 3":
+	%42 = load i64, i64* %35
+	%43 = add i64 %42, 1
+	store i64 %43, i64* %35
+	br label %"for.loop - 2"
+
+"for.done - 4":
+	br i1 false, label %"if.then - 5", label %"if.done - 6"
+
+"if.then - 5":
+	%44 = getelementptr inbounds [26 x i8], [26 x i8]* @.str0, i64 0, i64 0
+	%45 = alloca %.string, align 8 
+	store %.string zeroinitializer, %.string* %45
+	%46 = getelementptr inbounds %.string, %.string* %45, i64 0, i32 0
+	%47 = getelementptr inbounds %.string, %.string* %45, i64 0, i32 1
+	store i8* %44, i8** %46
+	store i64 26, i64* %47
+	%48 = load %.string, %.string* %45
+	call void @print_string(%.string %48)
+	%49 = getelementptr inbounds [26 x i8], [26 x i8]* @.str1, i64 0, i64 0
+	%50 = alloca %.string, align 8 
+	store %.string zeroinitializer, %.string* %50
+	%51 = getelementptr inbounds %.string, %.string* %50, i64 0, i32 0
+	%52 = getelementptr inbounds %.string, %.string* %50, i64 0, i32 1
+	store i8* %49, i8** %51
+	store i64 26, i64* %52
+	%53 = load %.string, %.string* %50
+	call void @print_string(%.string %53)
+	%54 = getelementptr inbounds [25 x i8], [25 x i8]* @.str2, i64 0, i64 0
+	%55 = alloca %.string, align 8 
+	store %.string zeroinitializer, %.string* %55
+	%56 = getelementptr inbounds %.string, %.string* %55, i64 0, i32 0
+	%57 = getelementptr inbounds %.string, %.string* %55, i64 0, i32 1
+	store i8* %54, i8** %56
+	store i64 25, i64* %57
+	%58 = load %.string, %.string* %55
+	call void @print_string(%.string %58)
+	%59 = getelementptr inbounds [27 x i8], [27 x i8]* @.str3, i64 0, i64 0
+	%60 = alloca %.string, align 8 
+	store %.string zeroinitializer, %.string* %60
+	%61 = getelementptr inbounds %.string, %.string* %60, i64 0, i32 0
+	%62 = getelementptr inbounds %.string, %.string* %60, i64 0, i32 1
+	store i8* %59, i8** %61
+	store i64 27, i64* %62
+	%63 = load %.string, %.string* %60
+	call void @print_string(%.string %63)
+	%64 = getelementptr inbounds [24 x i8], [24 x i8]* @.str4, i64 0, i64 0
+	%65 = alloca %.string, align 8 
+	store %.string zeroinitializer, %.string* %65
+	%66 = getelementptr inbounds %.string, %.string* %65, i64 0, i32 0
+	%67 = getelementptr inbounds %.string, %.string* %65, i64 0, i32 1
+	store i8* %64, i8** %66
+	store i64 24, i64* %67
+	%68 = load %.string, %.string* %65
+	call void @print_string(%.string %68)
+	%69 = getelementptr inbounds [42 x i8], [42 x i8]* @.str5, i64 0, i64 0
+	%70 = alloca %.string, align 8 
+	store %.string zeroinitializer, %.string* %70
+	%71 = getelementptr inbounds %.string, %.string* %70, i64 0, i32 0
+	%72 = getelementptr inbounds %.string, %.string* %70, i64 0, i32 1
+	store i8* %69, i8** %71
+	store i64 42, i64* %72
+	%73 = load %.string, %.string* %70
+	call void @print_string(%.string %73)
+	%74 = getelementptr inbounds [24 x i8], [24 x i8]* @.str6, i64 0, i64 0
+	%75 = alloca %.string, align 8 
+	store %.string zeroinitializer, %.string* %75
+	%76 = getelementptr inbounds %.string, %.string* %75, i64 0, i32 0
+	%77 = getelementptr inbounds %.string, %.string* %75, i64 0, i32 1
+	store i8* %74, i8** %76
+	store i64 24, i64* %77
+	%78 = load %.string, %.string* %75
+	call void @print_string(%.string %78)
+	%79 = getelementptr inbounds [35 x i8], [35 x i8]* @.str7, i64 0, i64 0
+	%80 = alloca %.string, align 8 
+	store %.string zeroinitializer, %.string* %80
+	%81 = getelementptr inbounds %.string, %.string* %80, i64 0, i32 0
+	%82 = getelementptr inbounds %.string, %.string* %80, i64 0, i32 1
+	store i8* %79, i8** %81
+	store i64 35, i64* %82
+	%83 = load %.string, %.string* %80
+	call void @print_string(%.string %83)
+	%84 = getelementptr inbounds [33 x i8], [33 x i8]* @.str8, i64 0, i64 0
+	%85 = alloca %.string, align 8 
+	store %.string zeroinitializer, %.string* %85
+	%86 = getelementptr inbounds %.string, %.string* %85, i64 0, i32 0
+	%87 = getelementptr inbounds %.string, %.string* %85, i64 0, i32 1
+	store i8* %84, i8** %86
+	store i64 33, i64* %87
+	%88 = load %.string, %.string* %85
+	call void @print_string(%.string %88)
+	%89 = getelementptr inbounds [24 x i8], [24 x i8]* @.str9, i64 0, i64 0
+	%90 = alloca %.string, align 8 
+	store %.string zeroinitializer, %.string* %90
+	%91 = getelementptr inbounds %.string, %.string* %90, i64 0, i32 0
+	%92 = getelementptr inbounds %.string, %.string* %90, i64 0, i32 1
+	store i8* %89, i8** %91
+	store i64 24, i64* %92
+	%93 = load %.string, %.string* %90
+	call void @print_string(%.string %93)
+	%94 = getelementptr inbounds [45 x i8], [45 x i8]* @.stra, i64 0, i64 0
+	%95 = alloca %.string, align 8 
+	store %.string zeroinitializer, %.string* %95
+	%96 = getelementptr inbounds %.string, %.string* %95, i64 0, i32 0
+	%97 = getelementptr inbounds %.string, %.string* %95, i64 0, i32 1
+	store i8* %94, i8** %96
+	store i64 45, i64* %97
+	%98 = load %.string, %.string* %95
+	call void @print_string(%.string %98)
+	%99 = getelementptr inbounds [24 x i8], [24 x i8]* @.strb, i64 0, i64 0
+	%100 = alloca %.string, align 8 
+	store %.string zeroinitializer, %.string* %100
+	%101 = getelementptr inbounds %.string, %.string* %100, i64 0, i32 0
+	%102 = getelementptr inbounds %.string, %.string* %100, i64 0, i32 1
+	store i8* %99, i8** %101
+	store i64 24, i64* %102
+	%103 = load %.string, %.string* %100
+	call void @print_string(%.string %103)
+	br label %"if.done - 6"
+
+"if.done - 6":
 	ret void
 }
 
@@ -159,143 +300,164 @@ define void @string_byte_reverse(%.string %s) {
 	ret void
 }
 
-define i64 @encode_rune({i8*, i64, i64} %buf, i32 %r) {
+define {[4 x i8], i64} @encode_rune(i32 %r) {
 "entry - 0":
-	%0 = alloca {i8*, i64, i64}, align 8 ; buf
-	store {i8*, i64, i64} zeroinitializer, {i8*, i64, i64}* %0
-	store {i8*, i64, i64} %buf, {i8*, i64, i64}* %0
-	%1 = alloca i32, align 4 ; r
-	store i32 zeroinitializer, i32* %1
-	store i32 %r, i32* %1
+	%0 = alloca i32, align 4 ; r
+	store i32 zeroinitializer, i32* %0
+	store i32 %r, i32* %0
+	%1 = alloca [4 x i8], align 1 ; buf
+	store [4 x i8] zeroinitializer, [4 x i8]* %1
 	%2 = alloca i32, align 4 ; i
 	store i32 zeroinitializer, i32* %2
-	%3 = load i32, i32* %1
+	%3 = load i32, i32* %0
 	store i32 %3, i32* %2
 	%4 = load i32, i32* %2
 	%5 = icmp ule i32 %4, 127
 	br i1 %5, label %"if.then - 1", label %"if.done - 2"
 
 "if.then - 1":
-	%6 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %0, i64 0, i32 0
-	%7 = load i8*, i8** %6
-	%8 = getelementptr i8, i8* %7, i64 0
-	%9 = load i32, i32* %1
-	%10 = trunc i32 %9 to i8
-	store i8 %10, i8* %8
-	ret i64 1
+	%6 = getelementptr inbounds [4 x i8], [4 x i8]* %1, i64 0, i64 0
+	%7 = getelementptr i8, i8* %6, i64 0
+	%8 = load i32, i32* %0
+	%9 = trunc i32 %8 to i8
+	store i8 %9, i8* %7
+	%10 = alloca {[4 x i8], i64}, align 8 
+	store {[4 x i8], i64} zeroinitializer, {[4 x i8], i64}* %10
+	%11 = load [4 x i8], [4 x i8]* %1
+	%12 = getelementptr inbounds {[4 x i8], i64}, {[4 x i8], i64}* %10, i64 0, i32 0
+	store [4 x i8] %11, [4 x i8]* %12
+	%13 = getelementptr inbounds {[4 x i8], i64}, {[4 x i8], i64}* %10, i64 0, i32 1
+	store i64 1, i64* %13
+	%14 = load {[4 x i8], i64}, {[4 x i8], i64}* %10
+	ret {[4 x i8], i64} %14
 
 "if.done - 2":
-	%11 = load i32, i32* %2
-	%12 = icmp ule i32 %11, 2047
-	br i1 %12, label %"if.then - 3", label %"if.done - 4"
+	%15 = load i32, i32* %2
+	%16 = icmp ule i32 %15, 2047
+	br i1 %16, label %"if.then - 3", label %"if.done - 4"
 
 "if.then - 3":
-	%13 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %0, i64 0, i32 0
-	%14 = load i8*, i8** %13
-	%15 = getelementptr i8, i8* %14, i64 0
-	%16 = load i32, i32* %1
-	%17 = lshr i32 %16, 6
-	%18 = trunc i32 %17 to i8
-	%19 = or i8 192, %18
-	store i8 %19, i8* %15
-	%20 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %0, i64 0, i32 0
-	%21 = load i8*, i8** %20
-	%22 = getelementptr i8, i8* %21, i64 1
-	%23 = load i32, i32* %1
-	%24 = trunc i32 %23 to i8
-	%25 = and i8 %24, 63
-	%26 = or i8 128, %25
-	store i8 %26, i8* %22
-	ret i64 2
+	%17 = getelementptr inbounds [4 x i8], [4 x i8]* %1, i64 0, i64 0
+	%18 = getelementptr i8, i8* %17, i64 0
+	%19 = load i32, i32* %0
+	%20 = lshr i32 %19, 6
+	%21 = trunc i32 %20 to i8
+	%22 = or i8 192, %21
+	store i8 %22, i8* %18
+	%23 = getelementptr inbounds [4 x i8], [4 x i8]* %1, i64 0, i64 0
+	%24 = getelementptr i8, i8* %23, i64 1
+	%25 = load i32, i32* %0
+	%26 = trunc i32 %25 to i8
+	%27 = and i8 %26, 63
+	%28 = or i8 128, %27
+	store i8 %28, i8* %24
+	%29 = alloca {[4 x i8], i64}, align 8 
+	store {[4 x i8], i64} zeroinitializer, {[4 x i8], i64}* %29
+	%30 = load [4 x i8], [4 x i8]* %1
+	%31 = getelementptr inbounds {[4 x i8], i64}, {[4 x i8], i64}* %29, i64 0, i32 0
+	store [4 x i8] %30, [4 x i8]* %31
+	%32 = getelementptr inbounds {[4 x i8], i64}, {[4 x i8], i64}* %29, i64 0, i32 1
+	store i64 2, i64* %32
+	%33 = load {[4 x i8], i64}, {[4 x i8], i64}* %29
+	ret {[4 x i8], i64} %33
 
 "if.done - 4":
-	%27 = load i32, i32* %2
-	%28 = icmp ugt i32 %27, 1114111
-	br i1 %28, label %"if.then - 5", label %"cmp-or - 6"
+	%34 = load i32, i32* %2
+	%35 = icmp ugt i32 %34, 1114111
+	br i1 %35, label %"if.then - 5", label %"cmp-or - 6"
 
 "if.then - 5":
-	store i32 65533, i32* %1
+	store i32 65533, i32* %0
 	br label %"if.done - 8"
 
 "cmp-or - 6":
-	%29 = load i32, i32* %2
-	%30 = icmp uge i32 %29, 55296
-	br i1 %30, label %"cmp-and - 7", label %"if.done - 8"
+	%36 = load i32, i32* %2
+	%37 = icmp uge i32 %36, 55296
+	br i1 %37, label %"cmp-and - 7", label %"if.done - 8"
 
 "cmp-and - 7":
-	%31 = load i32, i32* %2
-	%32 = icmp ule i32 %31, 57343
-	br i1 %32, label %"if.then - 5", label %"if.done - 8"
+	%38 = load i32, i32* %2
+	%39 = icmp ule i32 %38, 57343
+	br i1 %39, label %"if.then - 5", label %"if.done - 8"
 
 "if.done - 8":
-	%33 = load i32, i32* %2
-	%34 = icmp ule i32 %33, 65535
-	br i1 %34, label %"if.then - 9", label %"if.done - 10"
+	%40 = load i32, i32* %2
+	%41 = icmp ule i32 %40, 65535
+	br i1 %41, label %"if.then - 9", label %"if.done - 10"
 
 "if.then - 9":
-	%35 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %0, i64 0, i32 0
-	%36 = load i8*, i8** %35
-	%37 = getelementptr i8, i8* %36, i64 0
-	%38 = load i32, i32* %1
-	%39 = lshr i32 %38, 12
-	%40 = trunc i32 %39 to i8
-	%41 = or i8 224, %40
-	store i8 %41, i8* %37
-	%42 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %0, i64 0, i32 0
-	%43 = load i8*, i8** %42
-	%44 = getelementptr i8, i8* %43, i64 1
-	%45 = load i32, i32* %1
-	%46 = lshr i32 %45, 6
-	%47 = trunc i32 %46 to i8
-	%48 = and i8 %47, 63
-	%49 = or i8 128, %48
-	store i8 %49, i8* %44
-	%50 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %0, i64 0, i32 0
-	%51 = load i8*, i8** %50
-	%52 = getelementptr i8, i8* %51, i64 2
-	%53 = load i32, i32* %1
-	%54 = trunc i32 %53 to i8
-	%55 = and i8 %54, 63
-	%56 = or i8 128, %55
-	store i8 %56, i8* %52
-	ret i64 3
+	%42 = getelementptr inbounds [4 x i8], [4 x i8]* %1, i64 0, i64 0
+	%43 = getelementptr i8, i8* %42, i64 0
+	%44 = load i32, i32* %0
+	%45 = lshr i32 %44, 12
+	%46 = trunc i32 %45 to i8
+	%47 = or i8 224, %46
+	store i8 %47, i8* %43
+	%48 = getelementptr inbounds [4 x i8], [4 x i8]* %1, i64 0, i64 0
+	%49 = getelementptr i8, i8* %48, i64 1
+	%50 = load i32, i32* %0
+	%51 = lshr i32 %50, 6
+	%52 = trunc i32 %51 to i8
+	%53 = and i8 %52, 63
+	%54 = or i8 128, %53
+	store i8 %54, i8* %49
+	%55 = getelementptr inbounds [4 x i8], [4 x i8]* %1, i64 0, i64 0
+	%56 = getelementptr i8, i8* %55, i64 2
+	%57 = load i32, i32* %0
+	%58 = trunc i32 %57 to i8
+	%59 = and i8 %58, 63
+	%60 = or i8 128, %59
+	store i8 %60, i8* %56
+	%61 = alloca {[4 x i8], i64}, align 8 
+	store {[4 x i8], i64} zeroinitializer, {[4 x i8], i64}* %61
+	%62 = load [4 x i8], [4 x i8]* %1
+	%63 = getelementptr inbounds {[4 x i8], i64}, {[4 x i8], i64}* %61, i64 0, i32 0
+	store [4 x i8] %62, [4 x i8]* %63
+	%64 = getelementptr inbounds {[4 x i8], i64}, {[4 x i8], i64}* %61, i64 0, i32 1
+	store i64 3, i64* %64
+	%65 = load {[4 x i8], i64}, {[4 x i8], i64}* %61
+	ret {[4 x i8], i64} %65
 
 "if.done - 10":
-	%57 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %0, i64 0, i32 0
-	%58 = load i8*, i8** %57
-	%59 = getelementptr i8, i8* %58, i64 0
-	%60 = load i32, i32* %1
-	%61 = lshr i32 %60, 18
-	%62 = trunc i32 %61 to i8
-	%63 = or i8 240, %62
-	store i8 %63, i8* %59
-	%64 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %0, i64 0, i32 0
-	%65 = load i8*, i8** %64
-	%66 = getelementptr i8, i8* %65, i64 1
-	%67 = load i32, i32* %1
-	%68 = lshr i32 %67, 12
-	%69 = trunc i32 %68 to i8
-	%70 = and i8 %69, 63
-	%71 = or i8 128, %70
-	store i8 %71, i8* %66
-	%72 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %0, i64 0, i32 0
-	%73 = load i8*, i8** %72
-	%74 = getelementptr i8, i8* %73, i64 2
-	%75 = load i32, i32* %1
-	%76 = lshr i32 %75, 6
-	%77 = trunc i32 %76 to i8
-	%78 = and i8 %77, 63
-	%79 = or i8 128, %78
-	store i8 %79, i8* %74
-	%80 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %0, i64 0, i32 0
-	%81 = load i8*, i8** %80
-	%82 = getelementptr i8, i8* %81, i64 3
-	%83 = load i32, i32* %1
-	%84 = trunc i32 %83 to i8
-	%85 = and i8 %84, 63
-	%86 = or i8 128, %85
-	store i8 %86, i8* %82
-	ret i64 4
+	%66 = getelementptr inbounds [4 x i8], [4 x i8]* %1, i64 0, i64 0
+	%67 = getelementptr i8, i8* %66, i64 0
+	%68 = load i32, i32* %0
+	%69 = lshr i32 %68, 18
+	%70 = trunc i32 %69 to i8
+	%71 = or i8 240, %70
+	store i8 %71, i8* %67
+	%72 = getelementptr inbounds [4 x i8], [4 x i8]* %1, i64 0, i64 0
+	%73 = getelementptr i8, i8* %72, i64 1
+	%74 = load i32, i32* %0
+	%75 = lshr i32 %74, 12
+	%76 = trunc i32 %75 to i8
+	%77 = and i8 %76, 63
+	%78 = or i8 128, %77
+	store i8 %78, i8* %73
+	%79 = getelementptr inbounds [4 x i8], [4 x i8]* %1, i64 0, i64 0
+	%80 = getelementptr i8, i8* %79, i64 2
+	%81 = load i32, i32* %0
+	%82 = lshr i32 %81, 6
+	%83 = trunc i32 %82 to i8
+	%84 = and i8 %83, 63
+	%85 = or i8 128, %84
+	store i8 %85, i8* %80
+	%86 = getelementptr inbounds [4 x i8], [4 x i8]* %1, i64 0, i64 0
+	%87 = getelementptr i8, i8* %86, i64 3
+	%88 = load i32, i32* %0
+	%89 = trunc i32 %88 to i8
+	%90 = and i8 %89, 63
+	%91 = or i8 128, %90
+	store i8 %91, i8* %87
+	%92 = alloca {[4 x i8], i64}, align 8 
+	store {[4 x i8], i64} zeroinitializer, {[4 x i8], i64}* %92
+	%93 = load [4 x i8], [4 x i8]* %1
+	%94 = getelementptr inbounds {[4 x i8], i64}, {[4 x i8], i64}* %92, i64 0, i32 0
+	store [4 x i8] %93, [4 x i8]* %94
+	%95 = getelementptr inbounds {[4 x i8], i64}, {[4 x i8], i64}* %92, i64 0, i32 1
+	store i64 4, i64* %95
+	%96 = load {[4 x i8], i64}, {[4 x i8], i64}* %92
+	ret {[4 x i8], i64} %96
 }
 
 define void @print_rune(i32 %r) {
@@ -307,59 +469,59 @@ define void @print_rune(i32 %r) {
 	store [4 x i8] zeroinitializer, [4 x i8]* %1
 	%2 = alloca i64, align 8 ; n
 	store i64 zeroinitializer, i64* %2
-	%3 = sub i64 4, 0
-	%4 = sub i64 4, 0
-	%5 = getelementptr inbounds [4 x i8], [4 x i8]* %1, i64 0, i64 0
-	%6 = getelementptr i8, i8* %5, i64 0
-	%7 = alloca {i8*, i64, i64}, align 8 
-	store {i8*, i64, i64} zeroinitializer, {i8*, i64, i64}* %7
-	%8 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %7, i64 0, i32 0
-	store i8* %6, i8** %8
-	%9 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %7, i64 0, i32 1
-	store i64 %3, i64* %9
-	%10 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %7, i64 0, i32 2
-	store i64 %4, i64* %10
-	%11 = load {i8*, i64, i64}, {i8*, i64, i64}* %7
-	%12 = load i32, i32* %0
-	%13 = call i64 @encode_rune({i8*, i64, i64} %11, i32 %12)
-	store i64 %13, i64* %2
-	%14 = alloca %.string, align 8 ; str
-	store %.string zeroinitializer, %.string* %14
-	%15 = load i64, i64* %2
-	%16 = sub i64 %15, 0
-	%17 = sub i64 4, 0
-	%18 = getelementptr inbounds [4 x i8], [4 x i8]* %1, i64 0, i64 0
-	%19 = getelementptr i8, i8* %18, i64 0
-	%20 = alloca {i8*, i64, i64}, align 8 
-	store {i8*, i64, i64} zeroinitializer, {i8*, i64, i64}* %20
-	%21 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %20, i64 0, i32 0
-	store i8* %19, i8** %21
-	%22 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %20, i64 0, i32 1
-	store i64 %16, i64* %22
-	%23 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %20, i64 0, i32 2
-	store i64 %17, i64* %23
-	%24 = load {i8*, i64, i64}, {i8*, i64, i64}* %20
-	%25 = alloca {i8*, i64, i64}, align 8 
-	store {i8*, i64, i64} zeroinitializer, {i8*, i64, i64}* %25
-	store {i8*, i64, i64} %24, {i8*, i64, i64}* %25
-	%26 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %25, i64 0, i32 0
-	%27 = load i8*, i8** %26
-	%28 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %25, i64 0, i32 1
-	%29 = load i64, i64* %28
-	%30 = alloca %.string, align 8 
-	store %.string zeroinitializer, %.string* %30
-	%31 = getelementptr inbounds %.string, %.string* %30, i64 0, i32 0
-	%32 = getelementptr inbounds %.string, %.string* %30, i64 0, i32 1
-	store i8* %27, i8** %31
-	store i64 %29, i64* %32
-	%33 = load %.string, %.string* %30
-	store %.string %33, %.string* %14
-	%34 = load %.string, %.string* %14
-	call void @print_string(%.string %34)
+	%3 = load i32, i32* %0
+	%4 = call {[4 x i8], i64} @encode_rune(i32 %3)
+	%5 = extractvalue {[4 x i8], i64} %4, 0
+	%6 = extractvalue {[4 x i8], i64} %4, 1
+	store [4 x i8] %5, [4 x i8]* %1
+	store i64 %6, i64* %2
+	%7 = alloca %.string, align 8 ; str
+	store %.string zeroinitializer, %.string* %7
+	%8 = load i64, i64* %2
+	%9 = sub i64 %8, 0
+	%10 = sub i64 4, 0
+	%11 = getelementptr inbounds [4 x i8], [4 x i8]* %1, i64 0, i64 0
+	%12 = getelementptr i8, i8* %11, i64 0
+	%13 = alloca {i8*, i64, i64}, align 8 
+	store {i8*, i64, i64} zeroinitializer, {i8*, i64, i64}* %13
+	%14 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %13, i64 0, i32 0
+	store i8* %12, i8** %14
+	%15 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %13, i64 0, i32 1
+	store i64 %9, i64* %15
+	%16 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %13, i64 0, i32 2
+	store i64 %10, i64* %16
+	%17 = load {i8*, i64, i64}, {i8*, i64, i64}* %13
+	%18 = alloca {i8*, i64, i64}, align 8 
+	store {i8*, i64, i64} zeroinitializer, {i8*, i64, i64}* %18
+	store {i8*, i64, i64} %17, {i8*, i64, i64}* %18
+	%19 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %18, i64 0, i32 0
+	%20 = load i8*, i8** %19
+	%21 = getelementptr inbounds {i8*, i64, i64}, {i8*, i64, i64}* %18, i64 0, i32 1
+	%22 = load i64, i64* %21
+	%23 = alloca %.string, align 8 
+	store %.string zeroinitializer, %.string* %23
+	%24 = getelementptr inbounds %.string, %.string* %23, i64 0, i32 0
+	%25 = getelementptr inbounds %.string, %.string* %23, i64 0, i32 1
+	store i8* %20, i8** %24
+	store i64 %22, i64* %25
+	%26 = load %.string, %.string* %23
+	store %.string %26, %.string* %7
+	%27 = load %.string, %.string* %7
+	call void @print_string(%.string %27)
 	ret void
 }
 
-define void @print_int(i64 %i, i64 %base) {
+define void @print_int(i64 %i) {
+"entry - 0":
+	%0 = alloca i64, align 8 ; i
+	store i64 zeroinitializer, i64* %0
+	store i64 %i, i64* %0
+	%1 = load i64, i64* %0
+	call void @print_int_base(i64 %1, i64 10)
+	ret void
+}
+
+define void @print_int_base(i64 %i, i64 %base) {
 "entry - 0":
 	%0 = alloca i64, align 8 ; i
 	store i64 zeroinitializer, i64* %0
@@ -388,26 +550,26 @@ define void @print_int(i64 %i, i64 %base) {
 
 "if.done - 2":
 	%9 = load i64, i64* %0
-	%10 = icmp sgt i64 %9, 0
-	br i1 %10, label %"if.then - 3", label %"if.else - 4"
+	%10 = icmp eq i64 %9, 0
+	br i1 %10, label %"if.then - 3", label %"if.done - 4"
 
 "if.then - 3":
-	br label %"for.loop - 6"
-
-"if.else - 4":
 	%11 = getelementptr inbounds [21 x i8], [21 x i8]* %2, i64 0, i64 0
 	%12 = load i64, i64* %3
 	%13 = getelementptr i8, i8* %11, i64 %12
-	store i8 0, i8* %13
+	store i8 48, i8* %13
 	%14 = load i64, i64* %3
 	%15 = add i64 %14, 1
 	store i64 %15, i64* %3
-	br label %"if.done - 8"
+	br label %"if.done - 4"
+
+"if.done - 4":
+	br label %"for.loop - 6"
 
 "for.body - 5":
 	%16 = alloca i8, align 1 ; c
 	store i8 zeroinitializer, i8* %16
-	%17 = getelementptr inbounds [64 x i8], [64 x i8]* @.str2, i64 0, i64 0
+	%17 = getelementptr inbounds [64 x i8], [64 x i8]* @.strc, i64 0, i64 0
 	%18 = load i64, i64* %1
 	%19 = load i64, i64* %0
 	%20 = srem i64 %19, %18
@@ -434,23 +596,20 @@ define void @print_int(i64 %i, i64 %base) {
 	br i1 %33, label %"for.body - 5", label %"for.done - 7"
 
 "for.done - 7":
-	br label %"if.done - 8"
-
-"if.done - 8":
 	%34 = load i1, i1* %4
-	br i1 %34, label %"if.then - 9", label %"if.done - 10"
+	br i1 %34, label %"if.then - 8", label %"if.done - 9"
 
-"if.then - 9":
+"if.then - 8":
 	%35 = getelementptr inbounds [21 x i8], [21 x i8]* %2, i64 0, i64 0
 	%36 = load i64, i64* %3
 	%37 = getelementptr i8, i8* %35, i64 %36
-	store i8 0, i8* %37
+	store i8 45, i8* %37
 	%38 = load i64, i64* %3
 	%39 = add i64 %38, 1
 	store i64 %39, i64* %3
-	br label %"if.done - 10"
+	br label %"if.done - 9"
 
-"if.done - 10":
+"if.done - 9":
 	%40 = alloca %.string, align 8 ; str
 	store %.string zeroinitializer, %.string* %40
 	%41 = load i64, i64* %3
@@ -489,6 +648,16 @@ define void @print_int(i64 %i, i64 %base) {
 	ret void
 }
 
-@.str0 = global [1 x i8] c"\0A"
-@.str1 = global [1 x i8] c"\0A"
-@.str2 = global [64 x i8] c"0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz\40$"
+@.str0 = global [26 x i8] c"Chinese\20\20\20\20-\20\E4\BD\A0\E5\A5\BD\E4\B8\96\E7\95\8C\0A"
+@.str1 = global [26 x i8] c"Dutch\20\20\20\20\20\20-\20Hello\20wereld\0A"
+@.str2 = global [25 x i8] c"English\20\20\20\20-\20Hello\20world\0A"
+@.str3 = global [27 x i8] c"French\20\20\20\20\20-\20Bonjour\20monde\0A"
+@.str4 = global [24 x i8] c"German\20\20\20\20\20-\20Hallo\20Welt\0A"
+@.str5 = global [42 x i8] c"Greek\20\20\20\20\20\20-\20\CE\B3\CE\B5\CE\B9\CE\AC\20\CF\83\CE\BF\CF\85\20\CE\BA\CF\8C\CF\83\CE\BC\CE\BF\CF\82\0A"
+@.str6 = global [24 x i8] c"Italian\20\20\20\20-\20Ciao\20mondo\0A"
+@.str7 = global [35 x i8] c"Japanese\20\20\20-\20\E3\81\93\E3\82\93\E3\81\AB\E3\81\A1\E3\81\AF\E4\B8\96\E7\95\8C\0A"
+@.str8 = global [33 x i8] c"Korean\20\20\20\20\20-\20\EC\97\AC\EB\B3\B4\EC\84\B8\EC\9A\94\20\EC\84\B8\EA\B3\84\0A"
+@.str9 = global [24 x i8] c"Portuguese\20-\20Ol\C3\A1\20mundo\0A"
+@.stra = global [45 x i8] c"Russian\20\20\20\20-\20\D0\97\D0\B4\D1\80\D0\B0\D0\B2\D1\81\D1\82\D0\B2\D1\83\D0\BB\D1\82\D0\B5\20\D0\BC\D0\B8\D1\80\0A"
+@.strb = global [24 x i8] c"Spanish\20\20\20\20-\20Hola\20mundo\0A"
+@.strc = global [64 x i8] c"0123456789ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz\40$"
diff --git a/examples/main.odin b/examples/main.odin
index bc354a33b..1a6d9b443 100644
--- a/examples/main.odin
+++ b/examples/main.odin
@@ -2,31 +2,31 @@ import "basic"
 
 TWO_HEARTS :: '💕';
 
-tuple :: proc() -> (int, int) {
-	return 1, 2;
-}
-
 main :: proc() {
-	a, b : int = tuple();
+	dst : [4]int;
+	src : [2]int;
+	src[0] = 7;
+	src[1] = 5;
 
-	print_int(a, 10);
-	print_string("\n");
-	print_int(b, 10);
-	print_string("\n");
+	_ = copy(dst[1:], src[:]);
 
-/*
-	print_string("Chinese    - 你好世界\n");
-	print_string("Dutch      - Hello wereld\n");
-	print_string("English    - Hello world\n");
-	print_string("French     - Bonjour monde\n");
-	print_string("German     - Hallo Welt\n");
-	print_string("Greek      - γειά σου κόσμος\n");
-	print_string("Italian    - Ciao mondo\n");
-	print_string("Japanese   - こんにちは世界\n");
-	print_string("Korean     - 여보세요 세계\n");
-	print_string("Portuguese - Olá mundo\n");
-	print_string("Russian    - Здравствулте мир\n");
-	print_string("Spanish    - Hola mundo\n");
-*/
+	for i := 0; i < len(dst); i++ {
+		print_int(dst[i]);
+	}
+
+	if false {
+		print_string("Chinese    - 你好世界\n");
+		print_string("Dutch      - Hello wereld\n");
+		print_string("English    - Hello world\n");
+		print_string("French     - Bonjour monde\n");
+		print_string("German     - Hallo Welt\n");
+		print_string("Greek      - γειά σου κόσμος\n");
+		print_string("Italian    - Ciao mondo\n");
+		print_string("Japanese   - こんにちは世界\n");
+		print_string("Korean     - 여보세요 세계\n");
+		print_string("Portuguese - Olá mundo\n");
+		print_string("Russian    - Здравствулте мир\n");
+		print_string("Spanish    - Hola mundo\n");
+	}
 }
 
diff --git a/src/checker/checker.cpp b/src/checker/checker.cpp
index 5ae2ec417..6a4d4266e 100644
--- a/src/checker/checker.cpp
+++ b/src/checker/checker.cpp
@@ -522,6 +522,14 @@ void pop_procedure(Checker *c) {
 	gb_array_pop(c->proc_stack);
 }
 
+Type *const curr_procedure(Checker *c) {
+	isize count = gb_array_count(c->proc_stack);
+	if (count > 0) {
+		return c->proc_stack[count-1];
+	}
+	return NULL;
+}
+
 void add_curr_ast_file(Checker *c, AstFile *file) {
 	TokenPos zero_pos = {};
 	c->error_collector.prev = zero_pos;
diff --git a/src/checker/expr.cpp b/src/checker/expr.cpp
index d39c86ee0..2fc55a294 100644
--- a/src/checker/expr.cpp
+++ b/src/checker/expr.cpp
@@ -1553,6 +1553,7 @@ ExpressionKind check_call_expr(Checker *c, Operand *operand, AstNode *call) {
 		return Expression_Statement;
 	}
 
+
 	if (operand->mode == Addressing_Builtin) {
 		i32 id = operand->builtin_id;
 		if (!check_builtin_procedure(c, operand, call, id))
@@ -1574,6 +1575,17 @@ ExpressionKind check_call_expr(Checker *c, Operand *operand, AstNode *call) {
 		return Expression_Statement;
 	}
 
+	if (curr_procedure(c) == NULL) {
+		AstNode *e = operand->expr;
+		gbString str = expr_to_string(e);
+		defer (gb_string_free(str));
+		error(&c->error_collector, ast_node_token(e), "Can ony call procedure within a procedure: `%s`", str);
+
+		operand->mode = Addressing_Invalid;
+		operand->expr = call;
+
+		return Expression_Statement;
+	}
 
 	check_call_arguments(c, operand, proc_type, call);
 
diff --git a/src/codegen/codegen.cpp b/src/codegen/codegen.cpp
index 7af2c01cd..cde7951df 100644
--- a/src/codegen/codegen.cpp
+++ b/src/codegen/codegen.cpp
@@ -46,6 +46,8 @@ void ssa_gen_code(ssaGen *s) {
 	ssaModule *m = &s->module;
 	CheckerInfo *info = m->info;
 	gbAllocator a = m->allocator;
+	ssaProcedure dummy_proc = {};
+	dummy_proc.module = m;
 
 	gb_for_array(i, info->entities.entries) {
 		auto *entry = &info->entities.entries[i];
@@ -61,7 +63,14 @@ void ssa_gen_code(ssaGen *s) {
 		} break;
 
 		case Entity_Variable: {
-			ssaValue *g = ssa_make_value_global(a, e, NULL);
+			ssaValue *value = ssa_build_expr(&dummy_proc, decl->init_expr);
+			if (value->kind == ssaValue_Instr) {
+				ssaInstr *i = &value->instr;
+				if (i->kind == ssaInstr_Load) {
+					value = i->load.address;
+				}
+			}
+			ssaValue *g = ssa_make_value_global(a, e, value);
 			map_set(&m->values, hash_pointer(e), g);
 			map_set(&m->members, hash_string(name), g);
 		} break;
diff --git a/src/codegen/print_llvm.cpp b/src/codegen/print_llvm.cpp
index d4765ceb7..ce84cb45a 100644
--- a/src/codegen/print_llvm.cpp
+++ b/src/codegen/print_llvm.cpp
@@ -312,8 +312,9 @@ void ssa_print_instr(gbFile *f, ssaModule *m, ssaValue *value) {
 	case ssaInstr_GetElementPtr: {
 		Type *et = instr->get_element_ptr.elem_type;
 		ssa_fprintf(f, "%%%d = getelementptr ", value->id);
-		if (instr->get_element_ptr.inbounds)
-		ssa_fprintf(f, "inbounds ");
+		if (instr->get_element_ptr.inbounds) {
+			ssa_fprintf(f, "inbounds ");
+		}
 
 		ssa_print_type(f, m->sizes, et);
 		ssa_fprintf(f, ", ");
@@ -331,6 +332,16 @@ void ssa_print_instr(gbFile *f, ssaModule *m, ssaValue *value) {
 		ssa_fprintf(f, "\n");
 	} break;
 
+	case ssaInstr_ExtractValue: {
+		Type *et = instr->extract_value.elem_type;
+		ssa_fprintf(f, "%%%d = extractvalue ", value->id);
+
+		ssa_print_type(f, m->sizes, et);
+		ssa_fprintf(f, " ");
+		ssa_print_value(f, m, instr->extract_value.address, et);
+		ssa_fprintf(f, ", %d\n", instr->extract_value.index);
+	} break;
+
 	case ssaInstr_Br: {
 		ssa_fprintf(f, "br ");
 		if (instr->br.cond != NULL) {
@@ -494,7 +505,41 @@ void ssa_print_instr(gbFile *f, ssaModule *m, ssaValue *value) {
 
 	} break;
 
+	case ssaInstr_Select: {
+		ssa_fprintf(f, "%%%d = select i1 ", value->id);
+		ssa_print_value(f, m, instr->select.cond, t_bool);
+		ssa_fprintf(f, ", ");
+		ssa_print_type(f, m->sizes, ssa_value_type(instr->select.true_value));
+		ssa_fprintf(f, " ");
+		ssa_print_value(f, m, instr->select.true_value, ssa_value_type(instr->select.true_value));
+		ssa_fprintf(f, ", ");
+		ssa_print_type(f, m->sizes, ssa_value_type(instr->select.false_value));
+		ssa_fprintf(f, " ");
+		ssa_print_value(f, m, instr->select.false_value, ssa_value_type(instr->select.false_value));
+		ssa_fprintf(f, "\n");
+	} break;
+
+	case ssaInstr_CopyMemory: {
+		ssa_fprintf(f, "call void @llvm.memmove.p0i8.p0i8.");
+		ssa_print_type(f, m->sizes, t_int);
+		ssa_fprintf(f, "(i8* ");
+		ssa_print_value(f, m, instr->copy_memory.dst, t_rawptr);
+		ssa_fprintf(f, ", i8* ");
+		ssa_print_value(f, m, instr->copy_memory.src, t_rawptr);
+		ssa_fprintf(f, ", ");
+		ssa_print_type(f, m->sizes, t_int);
+		ssa_fprintf(f, " ");
+		ssa_print_value(f, m, instr->copy_memory.len, t_int);
+		char *vol_str = "false";
+		if (instr->copy_memory.is_volatile) {
+			vol_str = "true";
+		}
+		ssa_fprintf(f, ", i32 %d, i1 %s)\n", instr->copy_memory.align, vol_str);
+
+	} break;
+
 	default:
+		GB_PANIC("<unknown instr> %d\n", instr->kind);
 		ssa_fprintf(f, "; <unknown instr> %d\n", instr->kind);
 		break;
 	}
@@ -511,8 +556,12 @@ void ssa_print_llvm_ir(gbFile *f, ssaModule *m) {
 	ssa_fprintf(f, "} ; Basic_string\n\n");
 
 	ssa_print_encoded_local(f, make_string(".rawptr"));
-	ssa_fprintf(f, " = type i8*");
-	ssa_fprintf(f, " ; Basic_rawptr\n\n");
+	ssa_fprintf(f, " = type i8* ; Basic_rawptr\n\n");
+	ssa_fprintf(f, "declare void @llvm.memmove.p0i8.p0i8.");
+	ssa_print_type(f, m->sizes, t_int);
+	ssa_fprintf(f, "(i8*, i8*, ");
+	ssa_print_type(f, m->sizes, t_int);
+	ssa_fprintf(f, ", i32, i1)\n\n");
 
 	gb_for_array(member_index, m->members.entries) {
 		auto *entry = &m->members.entries[member_index];
diff --git a/src/codegen/ssa.cpp b/src/codegen/ssa.cpp
index 972f8dc72..90adf2c7d 100644
--- a/src/codegen/ssa.cpp
+++ b/src/codegen/ssa.cpp
@@ -60,12 +60,15 @@ struct ssaProcedure {
 	SSA_INSTR_KIND(Store), \
 	SSA_INSTR_KIND(Load), \
 	SSA_INSTR_KIND(GetElementPtr), \
+	SSA_INSTR_KIND(ExtractValue), \
 	SSA_INSTR_KIND(Conv), \
 	SSA_INSTR_KIND(Br), \
 	SSA_INSTR_KIND(Ret), \
+	SSA_INSTR_KIND(Select), \
 	SSA_INSTR_KIND(Unreachable), \
 	SSA_INSTR_KIND(BinaryOp), \
 	SSA_INSTR_KIND(Call), \
+	SSA_INSTR_KIND(CopyMemory), \
 	SSA_INSTR_KIND(Count),
 
 enum ssaInstrKind {
@@ -135,6 +138,12 @@ struct ssaInstr {
 			isize     index_count;
 			b32       inbounds;
 		} get_element_ptr;
+		struct {
+			ssaValue *address;
+			Type *    result_type;
+			Type *    elem_type;
+			i32       index;
+		} extract_value;
 		struct {
 			ssaConvKind kind;
 			ssaValue *value;
@@ -147,6 +156,11 @@ struct ssaInstr {
 		} br;
 		struct { ssaValue *value; } ret;
 		struct {} unreachable;
+		struct {
+			ssaValue *cond;
+			ssaValue *true_value;
+			ssaValue *false_value;
+		} select;
 		struct {
 			Type *type;
 			Token op;
@@ -158,6 +172,12 @@ struct ssaInstr {
 			ssaValue **args;
 			isize arg_count;
 		} call;
+		struct {
+			ssaValue *dst, *src;
+			ssaValue *len;
+			i32 align;
+			b32 is_volatile;
+		} copy_memory;
 	};
 };
 
@@ -280,10 +300,14 @@ Type *ssa_instr_type(ssaInstr *instr) {
 		return instr->load.type;
 	case ssaInstr_GetElementPtr:
 		return instr->get_element_ptr.result_type;
+	case ssaInstr_ExtractValue:
+		return instr->extract_value.result_type;
 	case ssaInstr_BinaryOp:
 		return instr->binary_op.type;
 	case ssaInstr_Conv:
 		return instr->conv.to;
+	case ssaInstr_Select:
+		return ssa_value_type(instr->select.true_value);
 	case ssaInstr_Call: {
 		Type *pt = instr->call.type;
 		GB_ASSERT(pt->kind == Type_Proc);
@@ -293,6 +317,8 @@ Type *ssa_instr_type(ssaInstr *instr) {
 		else
 			return tuple->variables[0]->type;
 	}
+	case ssaInstr_CopyMemory:
+		return t_int;
 	}
 	return NULL;
 }
@@ -311,6 +337,9 @@ void ssa_instr_set_type(ssaInstr *instr, Type *type) {
 	case ssaInstr_GetElementPtr:
 		instr->get_element_ptr.result_type = type;
 		break;
+	case ssaInstr_ExtractValue:
+		instr->extract_value.result_type = type;
+		break;
 	case ssaInstr_BinaryOp:
 		instr->binary_op.type = type;
 		break;
@@ -459,6 +488,22 @@ ssaValue *ssa_make_instr_get_element_ptr(ssaProcedure *p, ssaValue *address,
 	return v;
 }
 
+ssaValue *ssa_make_instr_extract_value(ssaProcedure *p, ssaValue *address, i32 index, Type *result_type) {
+	ssaValue *v = ssa_alloc_instr(p->module->allocator, ssaInstr_ExtractValue);
+	ssaInstr *i = &v->instr;
+	i->extract_value.address = address;
+	i->extract_value.index = index;
+	i->extract_value.result_type = result_type;
+	Type *et = ssa_value_type(address);
+	i->extract_value.elem_type = et;
+	GB_ASSERT(et->kind == Type_Structure || et->kind == Type_Array || et->kind == Type_Tuple);
+	if (p->curr_block) {
+		gb_array_append(p->curr_block->values, v);
+	}
+	return v;
+}
+
+
 ssaValue *ssa_make_instr_binary_op(ssaProcedure *p, Token op, ssaValue *left, ssaValue *right) {
 	ssaValue *v = ssa_alloc_instr(p->module->allocator, ssaInstr_BinaryOp);
 	ssaInstr *i = &v->instr;
@@ -500,6 +545,17 @@ ssaValue *ssa_make_instr_ret(ssaProcedure *p, ssaValue *value) {
 	return v;
 }
 
+ssaValue *ssa_make_instr_select(ssaProcedure *p, ssaValue *cond, ssaValue *t, ssaValue *f) {
+	ssaValue *v = ssa_alloc_instr(p->module->allocator, ssaInstr_Select);
+	v->instr.select.cond = cond;
+	v->instr.select.true_value = t;
+	v->instr.select.false_value = f;
+	if (p->curr_block) {
+		gb_array_append(p->curr_block->values, v);
+	}
+	return v;
+}
+
 ssaValue *ssa_make_instr_call(ssaProcedure *p, ssaValue *value, ssaValue **args, isize arg_count, Type *result_type) {
 	ssaValue *v = ssa_alloc_instr(p->module->allocator, ssaInstr_Call);
 	v->instr.call.value = value;
@@ -512,6 +568,19 @@ ssaValue *ssa_make_instr_call(ssaProcedure *p, ssaValue *value, ssaValue **args,
 	return v;
 }
 
+ssaValue *ssa_make_instr_copy_memory(ssaProcedure *p, ssaValue *dst, ssaValue *src, ssaValue *len, i32 align, b32 is_volatile) {
+	ssaValue *v = ssa_alloc_instr(p->module->allocator, ssaInstr_CopyMemory);
+	v->instr.copy_memory.dst = dst;
+	v->instr.copy_memory.src = src;
+	v->instr.copy_memory.len = len;
+	v->instr.copy_memory.align = align;
+	v->instr.copy_memory.is_volatile = is_volatile;
+	if (p->curr_block) {
+		gb_array_append(p->curr_block->values, v);
+	}
+	return v;
+}
+
 ssaValue *ssa_make_instr_conv(ssaProcedure *p, ssaConvKind kind, ssaValue *value, Type *from, Type *to) {
 	ssaValue *v = ssa_alloc_instr(p->module->allocator, ssaInstr_Conv);
 	v->instr.conv.kind = kind;
@@ -557,7 +626,6 @@ ssaValue *ssa_make_value_block(ssaProcedure *proc, AstNode *node, Scope *scope,
 	return v;
 }
 
-
 b32 ssa_is_blank_ident(AstNode *node) {
 	if (node->kind == AstNode_Ident) {
 		ast_node(i, Ident, node);
@@ -578,15 +646,24 @@ ssaInstr *ssa_get_last_instr(ssaBlock *block) {
 
 }
 
+b32 ssa_is_instr_terminating(ssaInstr *i) {
+	if (i != NULL) {
+		switch (i->kind) {
+		case ssaInstr_Ret:
+		case ssaInstr_Unreachable:
+			return true;
+		}
+	}
+
+	return false;
+}
+
 ssaValue *ssa_emit(ssaProcedure *proc, ssaValue *instr) {
 	ssaBlock *b = proc->curr_block;
 	instr->instr.parent = b;
 	if (b) {
 		ssaInstr *i = ssa_get_last_instr(b);
-		if (i && (i->kind == ssaInstr_Ret || i->kind == ssaInstr_Unreachable)) {
-			// NOTE(bill): any instruction in the current block after a `ret`
-			// or an `unreachable`, is never executed
-		} else {
+		if (!ssa_is_instr_terminating(i)) {
 			gb_array_append(b->instrs, instr);
 		}
 	}
@@ -598,7 +675,9 @@ ssaValue *ssa_emit_store(ssaProcedure *p, ssaValue *address, ssaValue *value) {
 ssaValue *ssa_emit_load(ssaProcedure *p, ssaValue *address) {
 	return ssa_emit(p, ssa_make_instr_load(p, address));
 }
-
+ssaValue *ssa_emit_select(ssaProcedure *p, ssaValue *cond, ssaValue *t, ssaValue *f) {
+	return ssa_emit(p, ssa_make_instr_select(p, cond, t, f));
+}
 
 
 ssaValue *ssa_add_local(ssaProcedure *proc, Entity *e) {
@@ -764,6 +843,7 @@ void ssa_end_procedure_body(ssaProcedure *proc) {
 			case ssaInstr_Br:
 			case ssaInstr_Ret:
 			case ssaInstr_Unreachable:
+			case ssaInstr_CopyMemory:
 				continue;
 			case ssaInstr_Call:
 				if (instr->call.type->proc.results == NULL) {
@@ -859,6 +939,18 @@ ssaValue *ssa_emit_struct_gep(ssaProcedure *proc, ssaValue *s, ssaValue *index,
 	return ssa_emit(proc, gep);
 }
 
+ssaValue *ssa_emit_struct_gep(ssaProcedure *proc, ssaValue *s, i32 index, Type *result_type) {
+	ssaValue *i = ssa_make_value_constant(proc->module->allocator, t_i32, make_exact_value_integer(index));
+	return ssa_emit_struct_gep(proc, s, i, result_type);
+}
+
+
+ssaValue *ssa_emit_struct_ev(ssaProcedure *proc, ssaValue *s, i32 index, Type *result_type) {
+	// NOTE(bill): For some weird legacy reason in LLVM, structure elements must be accessed as an i32
+	return ssa_emit(proc, ssa_make_instr_extract_value(proc, s, index, result_type));
+}
+
+
 
 ssaValue *ssa_array_elem(ssaProcedure *proc, ssaValue *array) {
 	Type *t = ssa_value_type(array);
@@ -967,7 +1059,7 @@ ssaValue *ssa_emit_slice(ssaProcedure *proc, Type *slice_type, ssaValue *base, s
 	gep = ssa_emit_struct_gep(proc, slice, v_two32, t_int);
 	ssa_emit_store(proc, gep, cap);
 
-	return ssa_emit_load(proc, slice);
+	return slice;
 }
 
 ssaValue *ssa_emit_substring(ssaProcedure *proc, ssaValue *base, ssaValue *low, ssaValue *high) {
@@ -981,20 +1073,21 @@ ssaValue *ssa_emit_substring(ssaProcedure *proc, ssaValue *base, ssaValue *low,
 	}
 
 	Token op_sub = {Token_Sub};
-	ssaValue *len = ssa_emit_arith(proc, op_sub, high, low, t_int);
+	ssaValue *elem, *len;
+	len = ssa_emit_arith(proc, op_sub, high, low, t_int);
 
-	ssaValue *elem = ssa_string_elem(proc, base);
+	elem = ssa_string_elem(proc, base);
 	elem = ssa_emit_ptr_offset(proc, elem, low);
 
-	ssaValue *str = ssa_add_local_generated(proc, t_string);
-	ssaValue *gep = NULL;
+	ssaValue *str, *gep;
+	str = ssa_add_local_generated(proc, t_string);
 	gep = ssa_emit_struct_gep(proc, str, v_zero32, ssa_value_type(elem));
 	ssa_emit_store(proc, gep, elem);
 
 	gep = ssa_emit_struct_gep(proc, str, v_one32, t_int);
 	ssa_emit_store(proc, gep, len);
 
-	return ssa_emit_load(proc, str);
+	return str;
 }
 
 
@@ -1012,9 +1105,7 @@ ssaValue *ssa_add_global_string_array(ssaProcedure *proc, ExactValue value) {
 	token.string = name;
 	Type *type = make_type_array(a, t_u8, value.value_string.len);
 	Entity *entity = make_entity_constant(a, NULL, token, type, value);
-	ssaValue *v = ssa_make_value_constant(a, type, value);
-
-	ssaValue *g = ssa_make_value_global(a, entity, v);
+	ssaValue *g = ssa_make_value_global(a, entity, ssa_make_value_constant(a, type, value));
 
 	map_set(&proc->module->values, hash_pointer(entity), g);
 	map_set(&proc->module->members, hash_string(name), g);
@@ -1033,7 +1124,7 @@ ssaValue *ssa_emit_string(ssaProcedure *proc, ssaValue *elem, ssaValue *len) {
 	ssaValue *str_len = ssa_emit_struct_gep(proc, str, v_one32, t_int);
 	ssa_emit_store(proc, str_elem, elem);
 	ssa_emit_store(proc, str_len, len);
-	return ssa_emit_load(proc, str);
+	return str;
 }
 
 
@@ -1120,15 +1211,14 @@ ssaValue *ssa_emit_conv(ssaProcedure *proc, ssaValue *value, Type *t) {
 		ssa_emit_store(proc, slice, value);
 		ssaValue *elem = ssa_slice_elem(proc, slice);
 		ssaValue *len  = ssa_slice_len(proc, slice);
-		return ssa_emit_string(proc, elem, len);
+		return ssa_emit_load(proc, ssa_emit_string(proc, elem, len));
 	}
 	if (is_type_string(src) && is_type_u8_slice(dst)) {
 		ssaValue *str = ssa_add_local_generated(proc, src);
 		ssa_emit_store(proc, str, value);
 		ssaValue *elem = ssa_string_elem(proc, str);
 		ssaValue *len  = ssa_string_len(proc, str);
-		ssaValue *v = ssa_emit_slice(proc, dst, elem, v_zero, len, len);
-		return v;
+		return ssa_emit_load(proc, ssa_emit_slice(proc, dst, elem, v_zero, len, len));
 	}
 
 
@@ -1261,6 +1351,7 @@ ssaValue *ssa_build_single_expr(ssaProcedure *proc, AstNode *expr, TypeAndValue
 				Entity *e = *found;
 				switch (e->builtin.id) {
 				case BuiltinProc_len: {
+					// len :: proc(Type) -> int
 					// NOTE(bill): len of an array is a constant expression
 					ssaValue *v = ssa_lvalue_address(ssa_build_addr(proc, ce->arg_list), proc);
 					Type *t = get_base_type(ssa_value_type(v));
@@ -1270,25 +1361,54 @@ ssaValue *ssa_build_single_expr(ssaProcedure *proc, AstNode *expr, TypeAndValue
 						return ssa_slice_len(proc, v);
 				} break;
 				case BuiltinProc_cap: {
+					// cap :: proc(Type) -> int
 					// NOTE(bill): cap of an array is a constant expression
 					ssaValue *v = ssa_lvalue_address(ssa_build_addr(proc, ce->arg_list), proc);
 					Type *t = get_base_type(ssa_value_type(v));
 					return ssa_slice_cap(proc, v);
 				} break;
 				case BuiltinProc_copy: {
-					GB_PANIC("TODO(bill): BuiltinProc_copy");
-					// TODO(bill): Should this be llvm.memmove internally?
-					// http://llvm.org/docs/LangRef.html#llvm-memmove-intrinsic
-					// declare void @llvm.memmove.p0i8.p0i8.i32(i8* <dest>, i8* <src>, i32 <len>, i32 <align>, i1 <isvolatile>)
-					// declare void @llvm.memmove.p0i8.p0i8.i64(i8* <dest>, i8* <src>, i64 <len>, i32 <align>, i1 <isvolatile>)
+					// copy :: proc(dst, src: []Type) -> int
+					AstNode *dst_node = ce->arg_list;
+					AstNode *src_node = ce->arg_list->next;
+					ssaValue *dst_slice = ssa_lvalue_address(ssa_build_addr(proc, dst_node), proc);
+					ssaValue *src_slice = ssa_lvalue_address(ssa_build_addr(proc, src_node), proc);
+					Type *slice_type = get_base_type(ssa_value_type(dst_slice));
+					GB_ASSERT(slice_type->kind == Type_Slice);
+					Type *elem_type = slice_type->slice.elem;
+					i64 size_of_elem = type_size_of(proc->module->sizes, proc->module->allocator, elem_type);
+
+					ssaValue *dst = ssa_emit_conv(proc, ssa_slice_elem(proc, dst_slice), t_rawptr);
+					ssaValue *src = ssa_emit_conv(proc, ssa_slice_elem(proc, src_slice), t_rawptr);
+
+					ssaValue *len_dst = ssa_slice_len(proc, dst_slice);
+					ssaValue *len_src = ssa_slice_len(proc, src_slice);
+
+					Token lt = {Token_Lt};
+					ssaValue *cond = ssa_emit_comp(proc, lt, len_dst, len_src);
+					ssaValue *len = ssa_emit_select(proc, cond, len_dst, len_src);
+					Token mul = {Token_Mul};
+					ssaValue *elem_size = ssa_make_value_constant(proc->module->allocator, t_int,
+					                                              make_exact_value_integer(size_of_elem));
+					ssaValue *byte_count = ssa_emit_arith(proc, mul, len, elem_size, t_int);
+
+
+					i32 align = cast(i32)type_align_of(proc->module->sizes, proc->module->allocator, elem_type);
+					b32 is_volatile = false;
+
+					ssa_emit(proc, ssa_make_instr_copy_memory(proc, dst, src, byte_count, align, is_volatile));
+					return len;
 				} break;
 				case BuiltinProc_append: {
+					// copy :: proc(s: ^[]Type, value: Type) -> bool
 					GB_PANIC("TODO(bill): BuiltinProc_append");
 				} break;
 				case BuiltinProc_print: {
+					// print :: proc(...)
 					GB_PANIC("TODO(bill): BuiltinProc_print");
 				} break;
 				case BuiltinProc_println: {
+					// println :: proc(...)
 					GB_PANIC("TODO(bill): BuiltinProc_println");
 				} break;
 				}
@@ -1310,13 +1430,9 @@ ssaValue *ssa_build_single_expr(ssaProcedure *proc, AstNode *expr, TypeAndValue
 			ssaValue *a = ssa_build_expr(proc, arg);
 			Type *at = ssa_value_type(a);
 			if (at->kind == Type_Tuple) {
-				ssaValue *tuple = ssa_add_local_generated(proc, at);
-				ssa_emit_store(proc, tuple, a);
 				for (isize i = 0; i < at->tuple.variable_count; i++) {
 					Entity *e = at->tuple.variables[i];
-					ssaValue *index = ssa_make_value_constant(proc->module->allocator, t_i32, make_exact_value_integer(i));
-					ssaValue *v = ssa_emit_struct_gep(proc, tuple, index, e->type);
-					v = ssa_emit_load(proc, v);
+					ssaValue *v = ssa_emit_struct_ev(proc, a, i, e->type);
 					args[arg_index++] = v;
 				}
 			} else {
@@ -1330,28 +1446,7 @@ ssaValue *ssa_build_single_expr(ssaProcedure *proc, AstNode *expr, TypeAndValue
 	case_end;
 
 	case_ast_node(se, SliceExpr, expr);
-		ssaValue *low  = NULL;
-		ssaValue *high = NULL;
-		ssaValue *max  = NULL;
-
-		if (se->low  != NULL)    low  = ssa_build_expr(proc, se->low);
-		if (se->high != NULL)    high = ssa_build_expr(proc, se->high);
-		if (se->triple_indexed)  max  = ssa_build_expr(proc, se->max);
-
-		switch (tv->type->kind) {
-		case Type_Slice:
-		case Type_Array: {
-			ssaValue *base = ssa_lvalue_address(ssa_build_addr(proc, se->expr), proc);
-			return ssa_emit_slice(proc, tv->type, base, low, high, max);
-		} break;
-		case Type_Basic: {
-			// NOTE(bill): max is not needed
-			ssaValue *base = ssa_lvalue_address(ssa_build_addr(proc, se->expr), proc);
-			return ssa_emit_substring(proc, base, low, high);
-		} break;
-		}
-
-		GB_PANIC("Unknown slicable type");
+		return ssa_emit_load(proc, ssa_lvalue_address(ssa_build_addr(proc, expr), proc));
 	case_end;
 
 	case_ast_node(ie, IndexExpr, expr);
@@ -1372,9 +1467,10 @@ ssaValue *ssa_build_expr(ssaProcedure *proc, AstNode *expr) {
 
 	if (tv->value.kind != ExactValue_Invalid) {
 		if (tv->value.kind == ExactValue_String) {
+			// TODO(bill): Optimize by not allocating everytime
 			ssaValue *array = ssa_add_global_string_array(proc, tv->value);
 			ssaValue *elem = ssa_array_elem(proc, array);
-			return ssa_emit_string(proc, elem, ssa_array_len(proc, array));
+			return ssa_emit_load(proc, ssa_emit_string(proc, elem, ssa_array_len(proc, array)));
 		}
 		return ssa_make_value_constant(proc->module->allocator, tv->type, tv->value);
 	}
@@ -1424,8 +1520,7 @@ ssaLvalue ssa_build_addr(ssaProcedure *proc, AstNode *expr) {
 			e = ssa_emit_load(proc, e);
 		}
 
-		ssaValue *index = ssa_make_value_constant(proc->module->allocator, t_i32, make_exact_value_integer(field_index));
-		ssaValue *v = ssa_emit_struct_gep(proc, e, index, entity->type);
+		ssaValue *v = ssa_emit_struct_gep(proc, e, field_index, entity->type);
 		return ssa_make_lvalue_address(v, expr);
 	case_end;
 
@@ -1465,6 +1560,32 @@ ssaLvalue ssa_build_addr(ssaProcedure *proc, AstNode *expr) {
 		return ssa_make_lvalue_address(v, expr);
 	case_end;
 
+	case_ast_node(se, SliceExpr, expr);
+		ssaValue *low  = NULL;
+		ssaValue *high = NULL;
+		ssaValue *max  = NULL;
+
+		if (se->low  != NULL)    low  = ssa_build_expr(proc, se->low);
+		if (se->high != NULL)    high = ssa_build_expr(proc, se->high);
+		if (se->triple_indexed)  max  = ssa_build_expr(proc, se->max);
+		Type *type = type_of_expr(proc->module->info, expr);
+
+		switch (type->kind) {
+		case Type_Slice:
+		case Type_Array: {
+			ssaValue *base = ssa_lvalue_address(ssa_build_addr(proc, se->expr), proc);
+			return ssa_make_lvalue_address(ssa_emit_slice(proc, type, base, low, high, max), expr);
+		} break;
+		case Type_Basic: {
+			// NOTE(bill): max is not needed
+			ssaValue *base = ssa_lvalue_address(ssa_build_addr(proc, se->expr), proc);
+			return ssa_make_lvalue_address(ssa_emit_substring(proc, base, low, high), expr);
+		} break;
+		}
+
+		GB_PANIC("Unknown slicable type");
+	case_end;
+
 	case_ast_node(de, DerefExpr, expr);
 		ssaValue *e = ssa_emit_load(proc, ssa_lvalue_address(ssa_build_addr(proc, de->expr), proc));
 		ssaValue *gep = ssa_make_instr_get_element_ptr(proc, e, NULL, NULL, 0, false);
@@ -1596,13 +1717,9 @@ void ssa_build_stmt(ssaProcedure *proc, AstNode *node) {
 					ssaValue *init = ssa_build_expr(proc, value);
 					Type *t = ssa_value_type(init);
 					if (t->kind == Type_Tuple) {
-						ssaValue *tuple = ssa_add_local_generated(proc, t);
-						ssa_emit_store(proc, tuple, init);
 						for (isize i = 0; i < t->tuple.variable_count; i++) {
 							Entity *e = t->tuple.variables[i];
-							ssaValue *index = ssa_make_value_constant(proc->module->allocator, t_i32, make_exact_value_integer(i));
-							ssaValue *v = ssa_emit_struct_gep(proc, tuple, index, e->type);
-							v = ssa_emit_load(proc, v);
+							ssaValue *v = ssa_emit_struct_ev(proc, init, i, e->type);
 							gb_array_append(inits, v);
 						}
 					} else {
@@ -1678,13 +1795,9 @@ void ssa_build_stmt(ssaProcedure *proc, AstNode *node) {
 					Type *t = ssa_value_type(init);
 					// TODO(bill): refactor for code reuse as this is repeated a bit
 					if (t->kind == Type_Tuple) {
-						ssaValue *tuple = ssa_add_local_generated(proc, t);
-						ssa_emit_store(proc, tuple, init);
 						for (isize i = 0; i < t->tuple.variable_count; i++) {
 							Entity *e = t->tuple.variables[i];
-							ssaValue *index = ssa_make_value_constant(proc->module->allocator, t_i32, make_exact_value_integer(i));
-							ssaValue *v = ssa_emit_struct_gep(proc, tuple, index, e->type);
-							v = ssa_emit_load(proc, v);
+							ssaValue *v = ssa_emit_struct_ev(proc, init, i, e->type);
 							gb_array_append(inits, v);
 						}
 					} else {
@@ -1750,8 +1863,7 @@ void ssa_build_stmt(ssaProcedure *proc, AstNode *node) {
 				Entity *e = return_type_tuple->variables[i];
 				ssaValue *res = ssa_build_expr(proc, r);
 				ssa_value_set_type(res, e->type);
-				ssaValue *index = ssa_make_value_constant(proc->module->allocator, t_int, make_exact_value_integer(i));
-				ssaValue *field = ssa_emit_struct_gep(proc, v, index, e->type);
+				ssaValue *field = ssa_emit_struct_gep(proc, v, i, e->type);
 				ssa_emit_store(proc, field, res);
 			}
 			v = ssa_emit_load(proc, v);
diff --git a/src/exact_value.cpp b/src/exact_value.cpp
index d62779462..b774d8656 100644
--- a/src/exact_value.cpp
+++ b/src/exact_value.cpp
@@ -87,7 +87,12 @@ ExactValue make_exact_value_from_basic_literal(Token token) {
 	case Token_String:  return make_exact_value_string(token.string);
 	case Token_Integer: return make_exact_value_integer(token.string);
 	case Token_Float:   return make_exact_value_float(token.string);
-	case Token_Rune:    return make_exact_value_integer(token.string);
+	case Token_Rune: {
+		Rune r = GB_RUNE_INVALID;
+		gb_utf8_decode(token.string.text, token.string.len, &r);
+		// gb_printf("%.*s rune: %d\n", LIT(token.string), r);
+		return make_exact_value_integer(r);
+	}
 	default:
 		GB_PANIC("Invalid token for basic literal");
 		break;
diff --git a/src/gb/gb.h b/src/gb/gb.h
index 2afc3b131..b7c5a9b9c 100644
--- a/src/gb/gb.h
+++ b/src/gb/gb.h
@@ -6625,6 +6625,8 @@ gb_global gbUtf8AcceptRange const gb__utf8_accept_ranges[] = {
 
 
 isize gb_utf8_decode(u8 const *str, isize str_len, Rune *codepoint_out) {
+
+
 	isize width = 0;
 	Rune codepoint = GB_RUNE_INVALID;
 
@@ -6639,6 +6641,11 @@ isize gb_utf8_decode(u8 const *str, isize str_len, Rune *codepoint_out) {
 			width = 1;
 			goto end;
 		}
+		if (s0 < 0x80) {
+			codepoint = s0;
+			width = 1;
+			goto end;
+		}
 
 		sz = x&7;
 		accept = gb__utf8_accept_ranges[x>>4];