reed-util

Vector/matrix math and other basic utilities.

Instructions:

• #include <util.h> for everything
  • Or, #include <util-basics.h> for just the stuff in the "Basics" section below
• Link with util.lib, or just add util.vcxproj to your solution and reference it

Basics:

• dim(), dim_field()—size of static arrays
• sizeof_field()—size of a struct member
• cassert()—compile-time assert (like static_assert but no separate error message)
• uint, byte, sized integer types i8, u8 through i64, u64
• Float constants: pi, epsilon = 1e-6f, infinity, NaN
• Generic swap, min, max, clamp, saturate, lerp, square
• Float utilities: isnear (comparisons with epsilon), isfinite, exp2f, log2f
• Integer utilities: round, modPositive, various power-of-2-related utilities
• offset()—do byte arithmetic on a pointer of any type
• exit("error message", ...)—printf a message to stderr, and exit with a nonzero code
• LoadFile()—just slurp an entire file into memory
• tokenize()—in-place destructive tokenization, à la strtok but stateless
• makeLowercase()—lowercase a string in-place
• TextParsingHelper—help parse a text file into lines and tokens, and issue warnings for syntax errors
• SerializeHelper, DeserializeHelper—help read/write common types from a byte stream
• Logging—to a file and/or a callback
• Asserts, checks, errors, and warnings—with log messages, callbacks, and debugbreaks

Math, based on design principles from On Vector Math Libraries:

• Vectors and matrices in any number of dimensions
• Construction of common transformations
• Functionality for working with affine transformations stored as homogeneous matrices
• Boxes in any number of dimensions
• Quaternions
• Color space conversions
• SIMD math using AOSOA (not well tested)
• Half-float from OpenEXR

Random number generation, from Quick and Easy GPU Random Numbers in D3D11 (although these are on the CPU):

• Xorshift PRNG (simple, fast 32-bit PRNG)
• Wang hash (simple, fast 32-bit hash for seeding Xorshift, etc.)

Todo list (in no particular order):

• Error context stack
• Allow choice of whether to break per individual error/warning/assert
• Range structs (one-dimensional boxes)?
• Ray/line/plane structs
• Primitive intersection tests (e.g. box vs ray, box vs frustum, etc.)
• Primitive clipping (e.g. line vs box, triangle vs box, etc.)
• Splines
• Spherical harmonics
• Perlin noise
• Quadratic equation solver
• Convex hull solver
• Matrix diagonalization and decomposition?
• Complex numbers?
• Projective math?
• Grassmann algebra?
• Dual numbers?
• Filename/path manipulation, à la Python's os.path
• Custom memory allocators with logging, leak detection, etc.
• Threading-related stuff? Or does std::async provide enough support?
• Cross-platform, cross-compiler support
• Command-line parameter parsing
• Basic UTF-8 string processing and conversion to/from UTF-16 for Windows
• rigid transformation class, represented by quat and translation vector?
• Infinite-far-plane and z-reversed projection matrices
• Containers (arrays, hashtables, intrusive linked lists)
• Allocators (linear allocator, stack allocator, tracked heap allocator?)