- Each row of the table becomes an object.
- The column attributes of the table become the attributes of the objects.
- The entries in the row become the values of those attributes.
- Any NULL (undefined) entries are omitted from the object.
For example, suppose our table is
|
|
DIVISION TYPE QUANTITY
-------- ------------ ----------
1 armor 20
2 mechanized infantry 100
3 infantry 500
4 armor 30
|
Then the schema will consist of
- schema:
- the name of the schema (generated by Ogre, but modifiable)
- id:
- the logical identifier or name for the object
- division, type, quantity:
- data values derived from the original relations
and the corresponding set of objects will be:
id: ob1
division: 1
type: armor
quantity: 20
id: ob2
division: 2
type: mechanized infantry
quantity: 100
id: ob3
division: 3
type: infantry
quantity: 500
id: ob4
division: 3
type: armor
quantity: 30
Note that each object has been given an id for the system, which
uniquely identifies that object.
Each object also has a schema attribute which is not shown above. The name of the schema is system-generated, although it can be set by the Ogre user.
One or more attributes in a row of a table may have unspecified values, formally known as NULL values. By design, Ogre does not create attribute-value pairs for such attributes, since part of the motivation is to allow condensed representation of data in which the tables are fairly sparse. For example, in the Location table presented earlier,
|
|
DIVISION LOCATION
-------- ---------
1 peninsula
2 foothills
3 plains
4
|
the location of division 4 is NULL, representing unknown. If this relation were converted to objects, only three of the objects would have a value for attribute LOCATION. The fourth object would be created, but would not have an attribute LOCATION.
Clearly our model is only one possible model which could be used. For example, another model would preserve the attribute in all objects, but create some kind of NULL value for the attribute in the object. As part of our objective is to reduce the space required for sparse relations (ones in which NULL's are frequent), we chose the current model.
Mapping Multiple Tables to Objects
A degenerate form of mapping for multiple tables
would involve simply repeating
the above single-table relation mapping once for each
relation. However, this may not capture the intent of the original
relational database. Typically relational databases
distribute the description of
single objects over multiple tables (through a design process known as
"normalization"), and the attribute values of one or
more tables would then be linked. We would like to recapture those
linkages in creating objects.
As the exact nature of the linkage cannot be inferred from the tables alone,
it is up to the user to specify the linkage, using Ogre's GUI.
Suppose there are two tables R and S with a common column attribute A and
we intend that rows of R and S having the same value of A really designate
the same object. We can indicate this intention in Ogre's GUI by making
a connection between nodes representing these attributes.
In relational database terminology, this is conceptually a
join of
the two tables.
For example, consider the two tables which have a common attribute DIVISION:
Identifying the attributes DIVISION in the two tables, Ogre could create the
following more comprehensive objects:
Ogre makes it possible to link any number of relations in this way. When
the linkages describe the user's desired
interpretation of the relations as objects,
Ogre will present a query to the relational database to extract that
information. Ogre will then generate the corresponding objects.
The user also can inspect both the original relations and the resulting objects
using the GUI.
DIVISION CO
-------- --------
1 Jones
2 Smith
3 Thomas
4 Williams
DIVISION TYPE QUANTITY
-------- ------------ ----------
1 armor 20
2 mechanized infantry 100
3 infantry 500
4 armor 30
id: ob1
co: Jones
division: 1
type: armor
quantity: 20
id: ob2
co: Smith
division: 2
type: mechanized infantry
quantity: 100
id: ob3
co: Thomas
division: 3
type: infantry
quantity: 500
id: ob4
co: Williams
division: 3
type: armor
quantity: 30
Using Ogre
Currently Ogre runs on the Sparcstation jake.jpl.nasa.gov, and is configured
to access a sample
database derived as an excerpt of the ASAS database.
- Connect to directory ~keller/ogre.os (.os means the ObjectStore version of
Ogre) by typing
cd ~keller/ogre.os - Type
go
