1. Data Definition Language

	Used to define relations and domains.
	Table driven.


1.1 Domains

    Domains are used to save storage, format output, and verify input, as well
as to document the structure of a database.  DBIO does not use domain
information to verify the legality of predicates.


	attributes of a domain:

		name		domain name
		type		atomic type
		default		default value (none, indef, actual)
		minimum		minimum value permitted
		maximum		maximum value permitted
		enumval		list of legal values
		units		units label
		format		default output format


	predefined (atomic) domains:

		bool
		byte*N
		char*N
		int*N
		real*N

The precision of an atomic domain is specified by N, the number of bytes of
storage to be reserved for the value.  N may be any integer value greater
than or equal to N=1 for byte, char, and int, or N=2 for real.  The byte
datatype is an unsigned (positive) integer.  The floating point datatype
has a one byte (8 bit) base 2 exponent.  For example, char*1 is a signed
byte, byte*2 is an unsigned 16 bit integer, and real*2 is a 16 bit floating
point number.


1.2 Groups

    A group is an aggregate of two or more domains or other groups.  Groups
as well as domains may be used to define the attributes of a relation.
Repeating groups, i.e., arrays of groups, are not allowed (a finite number
of named instances of a group may however be declared within a single relation).


	attributes of a group:

		name		group name as used in relation declarations
		nelements	number of elements (attributes) in group
		elements	set of elements (see below)

	
	attributes of each group element:

		name		attribute name
		domain		domain on which attribute is defined
		naxis		number of axes if array valued
		naxisN		length of each axis if array valued
		label		column label for output tables


1.3 Relations

    A relation declaration consists of a list of the attributes forming the
relation.  An attribute is a named instance of an atomic domain, user defined
domain, or group.  Any group, including nested groups, may be decomposed
into a set of named instances of domains, each of which is defined upon an
atomic datatype, hence a relation declaration is decomposable into a linear
list of atomic fields.  The relation is the logical unit of storage in a
database.  A base table is an named instance of some relation.


	attributes of a relation:

		name		name of the relation
		nattributes	number of attributes
		atr_list	list of attributes (see below)
		primary_key
		title

	
	attributes of each attribute of a relation:

		name		attribute name
		domain		domain on which attribute is defined
		naxis		number of axes if array valued
		naxisN		length of each axis if array valued
		label		column label for output tables


The atomic attributes of a relation may be either scalar or array valued.
The array valued attributes may be either static (the amount of storage is
set in the relation declaration) or dynamic (a variable amount of storage
is allocated at runtime).  Array valued attributes may not be used as
predicates in queries.


1.4 Views

    A view is a logical relation defined upon one or more base tables, i.e.,
instances of named relations.  The role views perform in a database is similar
to that performed by base tables, but views do not in themselves occupy any
storage.  The purpose of a view is to permit the appearance of the database
to be changed to suit the needs of a variety of applications, without having
to physically change the database itself.  As a trivial example, a view may
be used to provide aliases for the names of the attributes of a relation.


	attributes of a view:

		name		name of the view
		nattributes	number of attributes
		atr_list	list of attributes (see below)

	
	attributes of each attribute of a view:

		name		attribute name
		mapping		name of the table and attribute to which this
				    view attribute is mapped