Internal data representation: data iterators

nmad internal data representation is an iterator, ie the user essentially has to write a nm_data_traversal_t function which applies a function to every contiguous chunk of the data type. More...

Modules
	Data descriptors
	Core types to define data descriptors.
	Built-in data types
	A collection of pre-defined data iterators for common data types such as:
	Data helper functions
	A collection of helper function to help manipulating nm_data.
	Data slicing primitives.
	Data slicer is a collection of primitives to get a series of blocks with bounds controled by the caller from a given nm_data iterator.

Detailed Description

nmad internal data representation is an iterator, ie the user essentially has to write a nm_data_traversal_t function which applies a function to every contiguous chunk of the data type.

The iterator may use an internal state of size up to _NM_DATA_CONTENT_SIZE. The types must then be declared using macro NM_DATA_TYPE which associate the iterator function with the type for its internal state, the given name, and generates some typed accessor functions.

The main patterns to access data are:

• traversal: apply a given function to every chunk of data. The API is function nm_data_traversal_apply(), using a user-supplied nm_data_apply_t.
• generator: iterator (init/next/destroy) that returns a pointer to every chunk of data. The API is nm_data_generator_init(), nm_data_generator_next(), nm_data_generator_destroy().
• slicer: copy blocks of data with user-defined bounds for blocks, not necessary following in-memory data layout. The API is: nm_data_slicer_init(), nm_data_slicer_forward(), nm_data_slicer_copy_from(), nm_data_slicer_copy_to(), nm_data_slicer_destroy().

For a user-defined data type descriptor, user must define a traversal function nm_data_traversal_t, and optionnaly an optimized generator with nm_data_generator_init_t, nm_data_generator_next_t, nm_data_generator_destroy_t.

There is a collection of Built-in data types for common cases. Manipulation of such data representation is usually done through Data helper functions in higher layers, and Data slicing primitives. in drivers.