progress on resolving deps

code/base/memory.h

@@ -0,0 +1,38 @@
+#if MD_INTELLISENE_DIRECTIVES
+#pragma once
+#include "base_types.h"
+#endif
+
+#define MD_KILOBYTES( x ) (             ( x ) * ( S64 )( 1024 ) )
+#define MD_MEGABYTES( x ) ( MD_KILOBYTES( x ) * ( S64 )( 1024 ) )
+#define MD_GIGABYTES( x ) ( MD_MEGABYTES( x ) * ( S64 )( 1024 ) )
+#define MD_TERABYTES( x ) ( MD_GIGABYTES( x ) * ( S64 )( 1024 ) )
+
+#define MD__ONES          ( scast( GEN_NS usize, - 1) / MD_U8_MAX )
+#define MD__HIGHS         ( MD__ONES * ( MD_U8_MAX / 2 + 1 ) )
+#define MD__HAS_ZERO( x ) ( ( ( x ) - MD__ONES ) & ~( x ) & MD__HIGHS )
+
+#define swap(a, b)     \
+do                     \
+{                      \
+	typeof(a) tmp = a; \
+	a = b;             \
+	b = tmp;           \
+}                      \
+while (0)
+
+#if MD_COMPILER_MSVC || (MD_COMPILER_CLANG && MD_OS_WINDOWS)
+# pragma section(".rdata$", read)
+# define read_only __declspec(allocate(".rdata$"))
+#elif (MD_COMPILER_CLANG && MD_OS_LINUX)
+# define read_only __attribute__((section(".rodata")))
+#else
+// NOTE(rjf): I don't know of a useful way to do this in GCC land.
+// __attribute__((section(".rodata"))) looked promising, but it introduces a
+// strange warning about malformed section attributes, and it doesn't look
+// like writing to that section reliably produces access violations, strangely
+// enough. (It does on Clang)
+# define read_only
+#endif
+
+