Multiply two fixed-precision rational numbers. @param a @param b @return Returns a * b */

NOTE: P_LINEAR values to this routine must be 16-bit, but P_FILE values must * be 8-bit. */

Normalize numerator / denominator

Place a 16-bit number into a buffer in PNG byte order. * The parameter is declared unsigned int, not png_uint_16, * just to avoid potential problems

Fast generic assign (faster than for loop) @param dst Destination pixel @param src Source pixel @param bytesperpixel # of bytes per pixel */

Release the pointer, so that it won't perform deletion in its destruction.

Memory allocators. Modify these to use your own allocator.

Create an error message in allocated memory and set state->err and state->msg accordingly. Free any previous error message already there. Do n

Write the data of a PNG chunk started with png_write_chunk_header(). * Note that multiple calls to this function are allowed, and that the * sum of

Combine an AABB into this one.

Type management for properties

* Return open file's name. */

* Internal version of TIFFFlushData that can be * called by ``encodestrip routines'' w/o concern * for infinite recursion. */

Format Functions */

Divide an integer by a power of 2 and round downwards @return Returns a divided by 2^b */

Get the maximum of two integers @return Returns a if a > b else b */

IDCT stuff **/ reordering should be combined with zigzag scan **/ data order before IDCT **/ 0 8 4 6 **/ 2 10 14 12 **/ 1 11 15 13 **/ 9 3 7 5 *

Write a PNG chunk all at once. The type is an array of ASCII characters * representing the chunk name. The array must be at least 4 bytes in * len

Retrieve function * \return The value of the property. */

Get the length of this vector (the norm).

This is the shared test on whether a gamma value is 'significant' - whether * it is worth doing gamma correction. */

Utility to safely appends strings to a buffer. This never errors out so * error checking is not required in the caller. */

simple and slow implementation of half <-> float conversion