List space

From Organic Design wiki
Revision as of 21:33, 9 January 2011 by Infomaniac (talk | contribs) (List-space methods: gr tweak)

The Nodes are relatively high-level structures and at runtime within the peer-nodes they sit upon a more fundamental layer called list space in which the actual nodal change takes place in a peer.

List items

The project uses a very fundamental memory model based on binary sequences. The entire memory can be divided into groups of smaller sequences which can themselves each be referred to by a locally unique binary sequence. For example, a 16 megabyte section of RAM can be divided into two million 64-bit sequences. List-space uses blocks of binary as its local memory resource and divides the block up into smaller binary sequences called list items. Every list-item holds the same structure of content which is a sequence of three list-item-keys.

List item keys

Since binary can be treated as numbers, each of the two million sequences in the example above can itself be referred to, or addressed, by a unique binary sequence called a list-item-key. In the case of two million items, the list-item-keys are 21 bit sequences. In other words, a 16MB list-space is divided into two million list-items each formed from three 21-bit list-item-keys.

  • The slight inefficiency of wasting one out of every 64 bits is done because of increased efficiency gained from working with sizes which are powers of two.
  • Even an "empty" list-item containing only zero's is still considered to be composed of three list-item-keys all referring to the very first list-item in the list-space (the one that has a sequence of 21 zero's for its address).

Three keys

There are a number of ways that list-items can be linked together to form lists and binary-trees which are all constructed from this general "three-slot" data structure, such as stacks, queues and axes. List-space uses a generic loop linking system for handling what's in use and what's free, but the main functionality of list-space that is the foundation for Nodal Reduction is binary traversal.

Within each of the three keys common to all list-items, the first and second are used for binary traversal which allows list-item-keys to be used as associations (see node references for more details on this concept). The last key is context-specific and represents a value at the end of an association, but both the association-key and the association-value are list-item-keys -- references to other list-items.

List-space methods

List-space is an environment which offers a generic set of methods that can manipulate a space of list-items, each addressable by a binary list-item-key. Each list-item is composed of three list-item-keys. This simple model is rich enough to support many kinds of higher data structures such as stacks, queues, threads, loops and trees.

  • listInsert()
  • listRemove(subject)
  • listTraverse(subject, object)
  • listGetValue(subject)
  • listSetValue(subject, value)
  • listGetKeys(subject)

List space definition

To allow automated replication of list-space functionality in different environments, nodal organisations are defined to integrate with the program compilation resources. To make use of these nodal organisations, the list-space must be completely defined using the concepts available from within a node-space.

The list-space functions are designed to use only very generic operations. The highest level operations a list-space needs are as follows:

  • malloc - the ability to access an area of memory in the form of a sequence of items which each have a fixed ordered range of possible states. The most practical being 24 or 32 bit binary words, or in a high level language, an array of integers.
  • reference - each item in the sequence can be referred to by exactly one of the possible states - i.e. a state can be interpreted as an item-reference.
  • change - states can be copied from one address to another.
  • binary operations - the ability to represent a state as a binary sequence and iterate through it. Interpretation of a state as an address is also be based on representing state as binary. In both high level languages, or machine code level this can be achieved with the standard binary operations such as AND, OR, XOR, <<, >>.

See also