Skip Headers
Oracle® Call Interface Programmer's Guide,
10g Release 2 (10.2)

Go to Documentation Home
Go to Book List
Book List
Go to Table of Contents
Go to Index
Go to Master Index
Master Index
Go to Feedback page
Contact Us

Go to previous page
Go to next page
PDF · Mobi · ePub

10 OCI Object-Relational Programming

This chapter introduces the OCI's facility for working with objects in an Oracle database server. It also discusses the OCI's object navigational function calls.

This chapter contains these topics:

OCI Object Overview

OCI allows applications to access any of the datatypes found in the Oracle database server, including scalar values, collections, and instances of any object type. This includes all of the following:

To take full advantage of Oracle server object capabilities, most applications need to do more than just access objects. After an object has been retrieved, the application must navigate through references from that object to other objects. OCI provides the capability to do this. Through OCI's object navigational calls, an application can perform any of the following functions on objects:

The OCI navigational calls are discussed in more detail later in this chapter.

OCI also provides the ability to access type information stored in an Oracle database. The OCIDescribeAny() function enables an application to access most information relating to types stored in the database, including information about methods, attributes, and type metadata.

See Also:

OCIDescribeAny() is discussed in Chapter 6, "Describing Schema Metadata"

Applications interacting with Oracle objects need a way to represent those objects in a host language format. Oracle provides a utility called the Object Type Translator (OTT), which can convert type definitions in the database to C struct declarations. The declarations are stored in a header file that can be included in an OCI application.

When type definitions are represented in C, the types of attributes are mapped to special C variable types. The OCI includes a set of datatype mapping and manipulation functions that enable an application to manipulate these datatypes, and thus manipulate the attributes of objects.

See Also:

These functions are discussed in more detail in Chapter 11, "Object-Relational Datatypes in OCI"

The terminology for objects can occasionally become confusing. In the remainder of this chapter, the terms object and instance both refer to an object that is either stored in the database or is present in the object cache.

Working with Objects in OCI

Many of the programming principles that govern a relational OCI application are the same for an object-relational application. An object-relational application uses the standard OCI calls to establish database connections and process SQL statements. The difference is that the SQL statements issued retrieve object references, which can then be manipulated with OCI's object functions. An object can also be directly manipulated as a value instance (without using its object reference).

Basic Object Program Structure

The basic structure of an OCI application that uses objects is essentially the same as that for a relational OCI application, as described in the section "OCI Program Structure". That paradigm is reproduced here, with extra information covering basic object functionality.

  1. Initialize the OCI programming environment. You must initialize the environment in object mode.

    Your application will most likely also need to include C struct representations of database objects in a header file.

    See Also:

    These structs can be created by the programmer, or, more easily, they can be generated by the Object Type Translator (OTT), as described in Chapter 14, "Using the Object Type Translator with OCI"
  2. Allocate necessary handles, and establish a connection to a server.

  3. Prepare a SQL statement for execution. This is a local (client-side) step, which may include binding placeholders and defining output variables. In an object-relational application, this SQL statement should return a reference (REF) to an object.


    It is also possible to fetch an entire object, rather than just a reference (REF). If you SELECT a referenceable object, rather than pinning it, you get that object by value. You can also select a non-referenceable object. Fetching the entire object in this way is described in "Fetching Embedded Objects".
  4. Associate the prepared statement with a database server, and execute the statement.

  5. Fetch returned results.

    In an object-relational application, this step entails retrieving the REF, and then pinning the object to which it refers. Once the object is pinned, your application will do some or all of the following:

    • Manipulate the attributes of the object and mark it as dirty

    • Follow a REF to another object or series of objects

    • Access type and attribute information

    • Navigate a complex object retrieval graph

    • Flush modified objects to the server

  6. Commit the transaction. This step implicitly flushes all modified objects to the server and commits the changes.

  7. Free statements and handles not to be reused or re-execute prepared statements again.

All of these steps are discussed in more detail in the remainder of this chapter.

See Also:

Persistent Objects, Transient Objects, and Values

Instances of an Oracle type are categorized into persistent objects and transient objects based on their lifetime. Instances of persistent objects can be further divided into standalone objects and embedded objects depending on whether or not they are referenceable by way of an object identifier.


The terms object and instance are used interchangeably in this manual.

See Also:

For more information about objects, refer to the Oracle Database Application Developer's Guide - Object-Relational Features.

Persistent Object

A persistent object is an object which is stored in an Oracle database. It may be fetched into the object cache and modified by an OCI application. The lifetime of a persistent object can exceed that of the application which is accessing it. Once it is created, it remains in the database until it is explicitly deleted. There are two types of persistent objects:

  • Standalone instances are stored in rows of a object table, and each one has a unique object identifier. An OCI application can retrieve a REF to a standalone instance, pin the object and navigate from the pinned object to other related objects. Standalone object may also be referred to as referenceable objects.

    It is also possible to SELECT a referenceable object, in which case you fetch the object by value instead of fetching its REF.

  • Embedded instances are not stored as rows in a object table. They are embedded within other structures. Examples of embedded objects are objects which are attributes of another object, or instances which exist in an object column of a database table. Embedded instances do not have object identifiers, and OCI applications cannot get REFs to embedded instances.

    Embedded objects may also be referred to as non-referenceable objects or value instances. You may sometimes see them referred to as values, which is not to be confused with scalar data values. The context should make the meaning clear.

The following SQL examples demonstrate the difference between these two types of persistent objects.

Example 1, Standalone Objects

   (name      varchar2(30),
    age       number(3));
CREATE TABLE person_tab OF person_t;

Objects which are stored in the object table person_tab are standalone instances. They have object identifiers and are referenceable. They can be pinned in an OCI application.

Example 2, Embedded Objects

CREATE TABLE department
   (deptno     number,
    deptname   varchar2(30),
    manager    person_t);

Objects which are stored in the manager column of the department table are embedded objects. They do not have object identifiers, and they are not referenceable. This means they cannot be pinned in an OCI application, and they also never need to be unpinned. They are always retrieved into the object cache by value.

Transient Objects

A transient object is a temporary instance whose life does not exceed that of the application, and which cannot be stored or flushed to the server. The application can delete a transient object at any time.

Transient objects are often created by the application using the OCIObjectNew() function to store temporary values for computation. Transient objects cannot be converted to persistent objects. Their role is fixed at the time they are instantiated.

See Also:

"Creating Objects" for more information about using OCIObjectNew().


In the context of this manual, a value refers to either:

  • a scalar value which is stored in a non-object column of a database table. An OCI application can fetch values from a database by issuing SQL statements.

  • an embedded or non-referenceable object.

The context should make it clear which meaning is intended.


It is possible to SELECT a referenceable object into the object cache, rather than pinning it, in which case you fetch the object by value instead of fetching its REF.

Developing an OCI Object Application

This section discusses the steps involved in developing a basic OCI object application. Each step discussed in the section "Basic Object Program Structure" is described here in more detail.

The following figure shows a simple program logic flow for how an application might work with objects. For simplicity, some required steps are omitted. Each step in this diagram is discussed in the following sections.

Figure 10-1 Basic Object Operational Flow

Description of Figure 10-1 follows
Description of "Figure 10-1 Basic Object Operational Flow"

Representing Objects in C Applications

Before an OCI application can work with object types, those types must exist in the database. Typically, you create types with SQL DDL statements, such as CREATE TYPE.

When the Oracle server processes the type definition DDL commands, it stores the type definitions in the data dictionary as type descriptor objects (TDOs).

When your application retrieves instances of object types from the database, it needs to have a client-side representation of the objects. In a C program, the representation of an object type is a struct. In an OCI object application, you may also include a NULL indicator structure corresponding to each object type structure.

See Also:

Application programmers who wish to utilize object representations other than the default structs generated by the object cache should refer to "The Object Cache and Memory Management".

Oracle provides a utility called the Object Type Translator (OTT), which generates C struct representations of database object types for you. For example, if you have a type in your database declared as

( name       VARCHAR2(30),
  empno      NUMBER,
  deptno     NUMBER,
  hiredate   DATE,
  salary     NUMBER);

OTT produces the following C struct and corresponding NULL indicator struct:

struct emp_t
  OCIString    * name;
  OCINumber    empno;
  OCINumber    deptno;
  OCIDate      hiredate;
  OCINumber    salary;
typedef struct emp_t emp_t

struct emp_t_ind
  OCIInd     _atomic;
  OCIInd     name;
  OCIInd     empno;
  OCIInd     deptno;
  OCIInd     hiredate;
  OCIInd     salary;
typedef struct emp_t_ind emp_t_ind;

The variable types used in the struct declarations are special types employed by the OCI object calls. A subset of OCI functions manipulate data of these types.

These struct declarations are automatically written to a header file whose name is determined by the OTT input parameters. You can include this header file in the code files for an application to provide access to objects.

See Also:

These functions are mentioned later in this chapter, and are discussed in more detail in Chapter 11, "Object-Relational Datatypes in OCI".

Initializing Environment and Object Cache

If your OCI application will be accessing and manipulating objects, it is essential that you specify a value of OCI_OBJECT for the mode parameter of the OCIEnvCreate() call, which is the first OCI call in any OCI application. Specifying this value for mode indicates to the OCI libraries that your application will be working with objects. This notification has the following important effects:

  • it establishes the object run-time environment

  • it sets up the object cache

Memory for the object cache is allocated on demand when objects are loaded into the cache.

If the mode parameter of OCIInitialize() is not set to OCI_OBJECT, any attempt to use an object-related function will result in an error.

The client-side object cache is allocated in the program's process space. This cache is the memory for objects that have been retrieved from the server and are available to your application.


If you initialize the OCI environment in object mode, your application allocates memory for the object cache, whether or not the application actually uses object calls.

See Also:

The object cache is mentioned throughout this chapter. For a detailed explanation of the object cache, see Chapter 13, "Object Advanced Topics in OCI".

Making Database Connections

Once the OCI environment has been properly initialized, the application can connect to a server. This is accomplished through the standard OCI connect calls described in "OCI Programming Steps". When using these calls, no additional considerations need to be made because this application will be accessing objects.

There is only one object cache allocated for each OCI environment. All objects retrieved or created through different connections within the environment use the same physical object cache. Each connection has its own logical object cache.

Retrieving an Object Reference from the Server

In order to work with objects, your application must first retrieve one or more objects from the server. You accomplish this by issuing a SQL statement that returns REFs to one or more objects.


It is also possible for a SQL statement to fetch embedded objects, rather than REFs, from a database. See the section "Fetching Embedded Objects" for more information.

In the following example, the application declares a text block that stores a SQL statement designed to retrieve a REF to a single employee object from an object table of employees (emp_tab) in the database, given a particular employee number which is passed as an input variable (:emp_num) at runtime:

text *selemp = (text *) "SELECT REF(e)
                          FROM emp_tab e
                          WHERE empno = :emp_num";

Your application should prepare and process this statement in the same way that it would handle any relational SQL statement, as described in Chapter 2, "OCI Programming Basics":

  • Prepare an application request, using OCIStmtPrepare().

  • Bind the host input variable using the appropriate bind call(s).

  • Declare and prepare an output variable to receive the employee object reference. Here you would use an employee object reference, like the one declared in "Representing Objects in C Applications":

OCIRef  *emp1_ref = (OCIRef *) 0; /* reference to an employee object */

When defining the output variable, set the dty datatype parameter for the define call to SQLT_REF, the datatype constant for REF.

  • Execute the statement with OCIStmtExecute().

  • Fetch the resulting REF into emp1_ref, using OCIStmtFetch().

At this point, you could use the object reference to access and manipulate an object or objects from the database.

See Also:

Pinning an Object

Upon completion of the fetch step, your application has a REF, or pointer, to an object. The actual object is not currently available to work with. Before you can manipulate an object, it must be pinned. Pinning an object loads the object instance into the object cache, and enables you to access and modify the instance's attributes and follow references from that object to other objects, if necessary. Your application also controls when modified objects are written back to the server.


This section deals with a simple pin operation involving a single object at a time. For information about retrieving multiple objects through complex object retrieval, see the section "Complex Object Retrieval".

An application pins an object by calling the function OCIObjectPin(). The parameters for this function allow you to specify the pin option, pin duration, and lock option for the object.

The following sample code illustrates a pin operation for the employee reference we retrieved in the previous section:

if (OCIObjectPin(env, err, emp1_ref, (OCIComplexObject *) 0, 
     OCI_LOCK_X,  &emp1) != OCI_SUCCESS)

In this example, process_error() represents an error-handling function. If the call to OCIObjectPin() returns anything but OCI_SUCCESS, the error-handling function is called. The parameters of the OCIObjectPin() function are as follows:

  • env is the OCI environment handle.

  • err is the OCI error handle.

  • emp1_ref is the reference that was retrieved through SQL.

  • (OCIComplexObject *) 0 indicates that this pin operation is not utilizing complex object retrieval.

  • OCI_PIN_ANY is the pin option. See "Pinning an Object Copy" for more information.

  • OCI_DURATION_TRANS is the pin duration. See "Object Duration" for more information.

  • OCI_LOCK_X is the lock option. See "Locking Objects For Update" for more information.

  • emp1 is an out parameter, which returns a pointer to the pinned object.

Now that the object has been pinned, the OCI application can modify that object. In this simple example, the object contains no references to other objects.

See Also:

For an example of navigation from one instance to another, see the section "Simple Object Navigation".

Array Pin

Given an array of references, an OCI application can pin an array of objects by calling OCIObjectArrayPin(). The references may point to objects of different types. This function provides the ability for fetching objects of different types from different tables in one network round trip.

Manipulating Object Attributes

Once an object has been pinned, an OCI application can modify its attributes. The OCI provides a set of function for working with datatypes of object type structs, known as the OCI datatype mapping and manipulation functions.


Changes made to objects pinned in the object cache affect only those object copies (instances), and not the original object in the database. In order for changes made by the application to reach the database, those changes must be flushed/committed to the server. See "Marking Objects and Flushing Changes" for more information.

For example, assume that the employee object in the previous section was pinned so that the employee's salary could be increased. Assume also that at this company, yearly salary increases are prorated for employees who have been at the company for less than 180 days.

For this example we will need to access the employee's hire date and check whether it is more or less than 180 days prior to the current date. Based on that calculation, the employee's salary is increased by either $5000 (for more than 180 days) or $3000 (for less than 180 days). The sample code on the following page demonstrates this process.

Note that the datatype mapping and manipulation functions work with a specific set of datatypes; you must convert other types, like int, to the appropriate OCI types before using them in calculations.

/* assume that sysdate has been fetched into sys_date, a string. */
/* emp1 and emp1_ref are the same as in previous sections. */
/* err is the OCI error handle. */
/* NOTE: error handling code is not included in this example. */

sb4 num_days;        /* the number of days between today and hiredate */
OCIDate curr_date;          /* holds the current date for calculations */
int raise;   /* holds the employee's raise amount before calculations */
OCINumber raise_num;       /* holds employee's raise for calculations */
OCINumber new_sal;                 /* holds the employee's new salary */

/* convert date string to an OCIDate */
OCIDateFromText(err, (text *) sys_date, (ub4) strlen(sys_date), (text *) 
          NULL, (ub1) 0, (text *) NULL, (ub4) 0, &curr_date);

  /* get number of days between hire date and today */
OCIDateDaysBetween(err, &curr_date, &emp1->hiredate, &num_days);

/* calculate raise based on number of days since hiredate */
if (num_days > 180)
    raise = 5000;
    raise = 3000;

/* convert raise value to an OCINumber */
OCINumberFromInt(err, (dvoid *)&raise, (uword)sizeof(raise),      
                 OCI_NUMBER_SIGNED, &raise_num);

/* add raise amount to salary */
OCINumberAdd(err, &raise_num, &emp1->salary, &new_sal);
OCINumberAssign(err, &new_sal, &emp1->salary);

This example points out how values must be converted to OCI datatypes (for example, OCIDate, OCINumber) before being passed as parameters to the OCI datatype mapping and manipulation functions.

See Also:

For more information about the OCI datatypes and the datatype mapping and manipulation functions, refer to Chapter 11, "Object-Relational Datatypes in OCI".

Marking Objects and Flushing Changes

In the example in the previous section, an attribute of an object instance was changed. At this point, however, that change exists only in the client-side object cache. The application must take specific steps to insure that the change is written in the database.

The first step is to indicate that the object has been modified. This is done with the OCIObjectMarkUpdate() function. This function marks the object as dirty (modified).

Objects that have had their dirty flag set must be flushed to the server for the changes to be recorded in the database. You can do this in three ways:

  • Flush a single dirty object by calling OCIObjectFlush().

  • Flush the entire cache using OCICacheFlush(). In this case the OCI traverses the dirty list maintained by the cache and flushes the dirty objects to the server.

  • Call OCITransCommit() to commit a transaction. Doing so also traverses the dirty list and flushes objects to the server.

The flush operations work only on persistent objects in the cache. Transient objects are never flushed to the server.

Flushing an object to the server can activate triggers in the database. In fact, on some occasions an application may want to explicitly flush objects just to fire triggers on the server side.

See Also:

Fetching Embedded Objects

If your application needs to fetch an embedded object instance—an object stored in a column of a regular table, rather than an object table—you cannot use the REF retrieval mechanism described in the section "Retrieving an Object Reference from the Server". Embedded instances do not have object identifiers, so it is not possible to get a REF to them. This means that they cannot serve as the basis for object navigation. There are still many situations, however, in which an application will want to fetch embedded instances.

For example, assume that an address type has been created.

( street1             varchar2(50),
  street2             varchar2(50),
  city                varchar2(30),
  state               char(2),
  zip                 number(5));

You could then use that type as the datatype of a column in another table:

( name          varchar2(40),
  addr          address);

Your OCI application could then issue the following SQL statement:

SELECT addr FROM clients

This statement would return an embedded address object from the clients table. The application could then use the values in the attributes of this object for other processing.

Your application should prepare and process this statement in the same way that it would handle any relational SQL statement, as described in Chapter 2, "OCI Programming Basics":

  • Prepare an application request, using OCIStmtPrepare().

  • Bind the input variable using the appropriate bind call(s).

  • Define an output variable to receive the address instance. You use a C struct representation of the object type that was generated by OTT, as described in the section "Representing Objects in C Applications":

    addr1      *address; /* variable of the address struct type */

    When defining the output variable, set the dty datatype parameter for the define call to SQLT_NTY, the datatype constant for named datatypes.

  • Execute the statement with OCIStmtExecute()

  • Fetch the resulting instance into addr1, using OCIStmtFetch().

Following this, you can access the attributes of the instance, as described in the section "Manipulating Object Attributes", or pass the instance as an input parameter for another SQL statement.


Changes made to an embedded instance can be made persistent only by executing a SQL UPDATE statement.

See Also:

For more information about preparing and executing SQL statements, see "OCI Programming Steps".

Object Meta-Attributes

An object's meta-attributes serve as flags which can provide information to an application, or to the object cache, about the status of an object. For example, one of the meta-attributes of an object indicates whether or not it has been flushed to the server. These can help an application control the behavior of instances.

Persistent and transient object instances have different sets of meta-attributes. The meta-attributes for persistent objects are further broken down into persistent meta-attributes and transient meta-attributes. Transient meta-attributes exist only when an instance is in memory. Persistent meta-attributes also apply to objects stored in the server.

Persistent Object Meta-Attributes

Table 10-1 shows the meta-attributes for standalone persistent objects.

Table 10-1 Meta-Attributes of Persistent Objects

Meta-Attributes Meaning


does the object exist?


null information of the instance


has the object been locked?


has the object been marked as dirtied?


is the object pinned?

allocation duration

see "Object Duration"

pin duration

see "Object Duration"


Embedded persistent objects only have the nullity and allocation duration attributes, which are transient.

The OCI provides the OCIObjectGetProperty() function, which allows an application to check the status of a variety of attributes of an object. The syntax of the function is:

sword OCIObjectGetProperty ( OCIEnv              *envh, 
                             OCIError            *errh, 
                             CONST dvoid         *obj, 
                             OCIObjectPropId     propertyId,
                             dvoid               *property, 
                             ub4                 *size );

The propertyId and property parameters are used to retrieve information about any of a variety of properties or attributes

The different property ids and the corresponding type of property argument follow.


This identifies whether the given object is a persistent object or a transient object or a value instance. The property argument must be a pointer to a variable of type OCIObjectLifetime. Possible values include:





This returns the schema name of the table in which the object exists. An error is returned if the given object points to a transient instance or a value. If the input buffer is not big enough to hold the schema name an error is returned, the error message will communicate the required size. Upon success, the size of the returned schema name in bytes is returned by size. The property argument must be an array of type text and size should be set to size of array in bytes by the caller.


This returns the table name in which the object exists. An error is returned if the given object points to a transient instance or a value. If the input buffer is not big enough to hold the table name an error is returned, the error message will communicate the required size. Upon success, the size of the returned table name in bytes is returned by size. The property argument must be an array of type text and size should be set to size of array in bytes by the caller.


This returns the pin duration of the object. An error is returned if the given object points to a value instance. The property argument must be a pointer to a variable of type OCIDuration. Valid values include:




This returns the allocation duration of the object. The property argument must be a pointer to a variable of type OCIDuration. Valid values include:



    See Also:

    For more information about durations, see "Object Duration".

This returns the lock status of the object. The possible lock status is enumerated by OCILockOpt. An error is returned if the given object points to a transient or value instance. The property argument must be a pointer to a variable of type OCILockOpt. Note, the lock status of an object can also be retrieved by calling OCIObjectIsLocked().


This returns the dirty status and indicates whether the object is a new object, updated object or deleted object. An error is returned if the given object points to a transient or value instance. The property argument must be of type OCIObjectMarkStatus. Valid values include:




The following macros are available to test the object mark status:



  • OCI_OBJECT_IS_NEW (flag)



This identifies whether the specified object is an object view or not. If the property value returned is TRUE, it indicates the object is a view otherwise it is not. An error is returned if the given object points to a transient or value instance. The property argument must be of type boolean.

Just as a view is a virtual table, an object view is a virtual object table. Each row in the view is an object: you can call its methods, access its attributes using the dot notation, and create a REF that points to it.

Additional Attribute Functions

The OCI also provides functions which allow an application to set or check some of these attributes directly or indirectly, as shown in the following table:

Table 10-2 Set and Check Functions

Meta-Attribute Set With Check With













Transient Object Meta-Attributes

Transient objects have no persistent attributes, and the following transient attributes:

Table 10-3 Transient Meta-Attributes

Transient Meta-Attributes Meaning


does the object exist?


is the object being accessed by the application?


has the object been marked as dirtied?


null information of the instance

allocation duration

see "Object Duration"

pin duration

see "Object Duration"

Complex Object Retrieval

In the examples earlier in this chapter, only a single instance at a time was fetched or pinned. In these cases, each pin operation involved a separate server round trip to retrieve the object.

Object-oriented applications often model their problems as a set of interrelated objects that form graphs of objects. The applications process objects by starting at some initial set of objects, and then using the references in these initial objects to traverse the remaining objects. In a client/server setting, each of these traversals could result in costly network round trips to fetch objects.

Application performance when dealing with objects may be increased through the use of complex object retrieval (COR). This is a prefetching mechanism in which an application specifies a criteria for retrieving a set of linked objects in a single operation.


As described later, this does not mean that these prefetched objects are all pinned. They are fetched into the object cache, so that subsequent pin calls are local operations.

A complex object is a set of logically related objects consisting of a root object, and a set of objects each of which is prefetched based on a given depth level. The root object is explicitly fetched or pinned. The depth level is the shortest number of references that need to be traversed from the root object to a given prefetched object in a complex object.

An application specifies a complex object by describing its content and boundary. The fetching of complex objects is constrained by an environment's prefetch limit, the amount of memory in the object cache that is available for prefetching objects.


The use of COR does not add functionality; it only improves performance so its use is optional.

As an example for this discussion, consider the following type declaration:

CREATE TYPE customer(...);
CREATE TYPE line_item(...);
CREATE TYPE line_item_varray as VARRAY(100) of REF line_item;
CREATE TYPE purchase_order AS OBJECT
( po_number         NUMBER,
  cust              REF customer,
  related_orders    REF purchase_order,
  line_items        line_item_varray);

The purchase_order type contains a scalar value for po_number, a VARRAY of line items, and two references. The first is to a customer type, and the second is to a purchase_order type, indicating that this type may be implemented as a linked list.

When fetching a complex object, an application must specify the following:

  1. a REF to the desired root object.

  2. one or more pairs of type and depth information to specify the boundaries of the complex object. The type information indicates which REF attributes should be followed for COR, and the depth level indicates how many levels deep those links should be followed.

In the case of the purchase order object preceding, the application must specify the following:

  1. the REF to the root purchase order object

  2. one or more pairs of type and depth information for cust, related_orders, or line_items

An application fetching a purchase order will very likely need access to the customer information for that order. Using simple navigation, this would require two server accesses to retrieve the two objects. Through complex object retrieval, the customer can be prefetched when the application pins the purchase order. In this case, the complex object would consist of the purchase order object and the customer object it references.

In the previous example, the application would specify the purchase_order REF, and would indicate that the cust REF attribute should be followed to a depth level of 1:

  1. REF(PO object)

  2. {(customer, 1)}

If the application wanted to prefetch the purchase_order object and all objects in the object graph it contains, the application would specify that both the cust and related_orders should be followed to the maximum depth level possible.

  1. REF(PO object)

  2. {(customer, UB4MAXVAL), (purchase_order, UB4MAXVAL)}

where UB4MAXVAL specifies that all objects of the specified type reachable through references from the root object should be prefetched.

If an application wanted to fetch a PO and all the associated line items, it would specify:

  1. REF(PO object)

  2. {(line_item, 1)}

The application can also choose to fetch all objects reachable from the root object by way of REFs (transitive closure) to a certain depth. To do so, set the level parameter to the depth desired. For the preceding two examples, the application could also specify (PO object REF, UB4MAXVAL) and (PO object REF, 1) respectively to prefetch required objects. Doing so results in many extraneous fetches but is quite simple to specify, and requires only one server round trip.

Prefetching Objects

After specifying and fetching a complex object, subsequent fetches of objects contained in the complex object do not incur the cost of a network round trip, because these objects have already been prefetched and are in the object cache. Keep in mind that excessive prefetching of objects can lead to a flooding of the object cache. This flooding, in turn, may force out other objects that the application had already pinned leading to a performance degradation instead of performance improvement.


If there is insufficient memory in the cache to hold all prefetched objects, some objects may not be prefetched. The application will then incur a network round trip when those objects are accessed later.

The SELECT privilege is needed for all prefetched objects. Objects in the complex object for which the application does not have SELECT privilege will not be prefetched.

Implementing Complex Object Retrieval in OCI

Complex Object Retrieval (COR) allows an application to prefetch a complex object while fetching the root object. The complex object specifications are passed to the same OCIObjectPin() function used for simple objects.

An application specifies the parameters for complex object retrieval using a complex object retrieval handle. This handle is of type OCIComplexObject and is allocated in the same way as other OCI handles.

The complex object retrieval handle contains a list of complex object retrieval descriptors. The descriptors are of type OCIComplexObjectComp, and are allocated in the same way as other OCI descriptors.

Each COR descriptor contains a type REF and a depth level. The type REF specifies a type of reference to be followed while constructing the complex object. The depth level indicates how far a particular type of reference should be followed. Specify an integer value, or the constant UB4MAXVAL for the maximum possible depth level.

The application can also specify the depth level in the COR handle without creating COR descriptors for type and depth parameters. In this case, all REFs are followed to the depth specified in the COR handle. The COR handle can also be used to specify whether a collection attribute should be fetched separately on demand (out-of-line) as opposed to the default case of fetching it along with the containing object (inline).

The application uses OCIAttrSet() to set the attributes of a COR handle. The attributes are:


OCI_ATTR_COMPLEXOBJECT_COLL_OUTOFLINE - fetch collection attribute in an object type out-of-line

The application allocates the COR descriptor using OCIDescriptorAlloc() and then can set the following attributes:


OCI_ATTR_COMPLEXOBJECTCOMP_LEVEL - the depth level for references of the preceding type

Once these attributes are set, the application calls OCIParamSet() to put the descriptor into a complex object retrieval handle. The handle has an OCI_ATTR_PARAM_COUNT attribute which specifies the number of descriptors on the handle. This attribute can be read with OCIAttrGet().

Once the handle has been populated, it can be passed to the OCIObjectPin() call to pin the root object and prefetch the remainder of the complex object.

The complex object retrieval handles and descriptors must be freed explicitly when they are no longer needed.

See Also:

For more information about handles and descriptors, see "Handles" and "OCI Descriptors".

COR Prefetching

The application specifies a complex object while fetching the root object. The prefetched objects are obtained by doing a breadth-first traversal of the graph(s) of objects rooted at a given root object(s). The traversal stops when all required objects have been prefetched, or when the total size of all the prefetched objects exceeds the prefetch limit.

COR interface

The interface for fetching complex objects is the OCI pin interface. The application can pass an initialized COR handle to OCIObjectPin() (or an array of handles to OCIObjectArrayPin()) to fetch the root object and the prefetched objects specified in the COR handle.

sword OCIObjectPin ( OCIEnv              *env, 
                     OCIError            *err, 
                     OCIRef              *object_ref,
                     OCIComplexObject    *corhdl,
                     OCIPinOpt           pin_option, 
                     OCIDuration         pin_duration, 
                     OCILockOpt          lock_option,
                     dvoid               **object );

sword OCIObjectArrayPin ( OCIEnv            *env, 
                          OCIError          *err, 
                          OCIRef            **ref_array, 
                          ub4               array_size,
                          OCIComplexObject  **cor_array,
                          ub4               cor_array_size, 
                          OCIPinOpt         pin_option, 
                          OCIDuration       pin_duration,
                          OCILockOpt        lock, 
                          dvoid             **obj_array,
                          ub4               *pos );

Note the following points when using COR:

  1. A null COR handle argument defaults to pinning just the root object.

  2. A COR handle with type of the root object and a depth level of 0 fetches only the root object and is thus equivalent to a null COR handle.

  3. The lock options apply only to the root object.


    In order to specify lock options for prefetched objects, the application can visit all the objects in a complex object, create an array of REFs, and lock the entire complex object in another round trip using the array interface (OCIObjectArrayPin()).

Example of COR

The following example illustrates how an application program can be modified to use complex object retrieval.

Consider an application that displays a purchase order and the line items associated with it. The code in boldface accomplishes this. The rest of the code uses complex object retrieval for prefetching and thus enhances the application's performance.

OCIEnv *envhp;
OCIError *errhp;
OCIRef **liref;
OCIRef *poref;
OCIIter *itr;
boolean  eoc;
purchase_order *po = (purchase_order *)0;
line_item *li = (line_item *)0;
OCISvcCtx *svchp;
OCIComplexObject *corhp;
OCIComplexObjectComp *cordp;
OCIType *litdo;
ub4 level = 0;

/* get COR Handle */
OCIHandleAlloc((dvoid *) envhp, (dvoid **) &corhp, (ub4)
                OCI_HTYPE_COMPLEXOBJECT, 0, (dvoid **)0);

/* get COR descriptor for type line_item */
OCIDescriptorAlloc((dvoid *) envhp, (dvoid **) &cordp, (ub4)
                OCI_DTYPE_COMPLEXOBJECTCOMP, 0, (dvoid **) 0);

/* get type of line_item to set in COR descriptor */
OCITypeByName(envhp, errhp, svchp, (const text *) 0, (ub4) 0,
              (const text *) "LINE_ITEM",
              (ub4) strlen((const char *) "LINE_ITEM"), (text *) 0,
              (ub4) 0, OCI_DURATION_SESSION,
              OCI_TYPEGET_HEADER, &litdo);

/* set line_item type in COR descriptor */
OCIAttrSet( (dvoid *) cordp, (ub4) OCI_DTYPE_COMPLEXOBJECTCOMP,
            (dvoid *) litdo, (ub4) sizeof(dvoid *), (ub4)
            OCI_ATTR_COMPLEXOBJECTCOMP_TYPE, (OCIError *) errhp);
level = 1;

/* set depth level for line_item_varray in COR descriptor */
OCIAttrSet( (dvoid *) cordp, (ub4) OCI_DTYPE_COMPLEXOBJECTCOMP,
            (dvoid *) &level, (ub4) sizeof(ub4), (ub4)

/* put COR descriptor in COR handle */
OCIParamSet(corhp, OCI_HTYPE_COMPLEXOBJECT, errhp, cordp,

/* pin the purchase order */
OCIObjectPin(envhp, errhp, poref, corhp, OCI_PIN_LATEST,

/* free COR descriptor and COR handle */
OCIDescriptorFree((dvoid *) cordp, (ub4) OCI_DTYPE_COMPLEXOBJECTCOMP);
OCIHandleFree((dvoid *) corhp, (ub4) OCI_HTYPE_COMPLEXOBJECT);

/* iterate and print line items for this purchase order */
OCIIterCreate(envhp, errhp, po->line_items, &itr);

/* get first line item */
OCIIterNext(envhp, errhp, itr, (dvoid **)&liref, (dvoid **)0, &eoc);
while (!eoc)        /* not end of collection */
/* pin line item */
   OCIObjectPin(envhp, errhp, *liref, (dvoid *)0, OCI_PIN_RECENT,
       OCI_LOCK_NONE, (dvoid **)&li));

/* get next line item */
OCIIterNext(envhp, errhp, itr, (dvoid **)&liref, (dvoid **)0, &eoc);

OCI Versus SQL Access to Objects

If an application needs to manipulate a graph of objects (inter-related by object references) then it is more effective to use the OCI interface rather than the SQL interface for accessing objects. Retrieving a graph of objects using the SQL interface may require executing multiple SELECT statements which would mean multiple network round trips. Using the complex object retrieval capability provided by the OCI, the application can retrieve the graph of objects in one OCIObjectPin() call.

Consider the update case where the application retrieves a graph of objects and modifies it based upon user interaction and then wishes to make the modifications persistent in the database. Using the SQL interface, the application would have to execute multiple UPDATE statements to update the graph of objects. If the modifications involved creation of new objects and deletion of existing objects then corresponding INSERT and DELETE statements would also need to be executed. In addition, the application would have to do more bookkeeping, such as keeping track of table names, because this information is required for executing the INSERT/UPDATE/DELETE statements.

Using the OCI's OCICacheFlush() function, the application can flush all modifications (insertion, deletion and update of objects) in a single operation. The OCI does all the bookkeeping, thereby requiring less coding on the part of the application. So for manipulating graph of objects the OCI is not only efficient but also provides an easy to use interface.

Consider a different case in which the application needs to fetch an object given its REF. In the OCI this is achieved by pinning the object using the OCIObjectPin() call. In the SQL interface this can be achieved by dereferencing the REF in a SELECT statement (for example, SELECT DEREF(ref) from tbl;). Consider situations where the same REF (reference to the same object) is being dereferenced multiple times in a transaction. By calling OCIObjectPin() with the OCI_PIN_RECENT option, the object will be fetched from the server only once for the transaction and repeated pins on the same REF result in returning a pointer to the already-pinned object in the cache. In the case of the SQL interface, each execution of the SELECT DEREF... statement would result in fetching the object from the server and hence would result in multiple round trips to the server and multiple copies of the same object.

Finally, consider the case in which the application needs to fetch a non-referenceable object. For example,

CREATE TABLE department 
deptno number, 
deptname varchar2(30), 
manager employee_t 

employee_t instances stored in the manager column are non-referenceable. Only the SQL interface can be used to fetch manager column instances. But if employee_t has any REF attributes, OCI calls can then be used to navigate the REF.

Pin Count and Unpinning

Each object in the object cache has a pin count associated with it. The pin count essentially indicates the number of code modules that are concurrently accessing the object. The pin count is set to 1 when an object is pinned into the cache for the first time. Objects prefetched with complex object retrieval enter the object cache with a pin count of zero.

It is possible to pin an already-pinned object. Doing so increases the pin count by one. When a process finishes using an object, it should unpin it, using OCIObjectUnpin(). This call decrements the pin count by one.

When the pin count of an object reaches zero, that object is eligible to be aged out of the cache if necessary, freeing up the memory space occupied by the object.

The pin count of an object can be set to zero explicitly by calling OCIObjectPinCountReset().

An application can unpin all objects in the cache related to a specific connection, by calling OCICacheUnpin().

See Also:

NULL Indicator Structure

If a column in a row of a database table has no value, then that column is said to be NULL, or to contain a NULL. Two different types of NULLs can apply to objects:

  • Any attribute of an object can have a NULL value. This indicates that the value of that attribute of the object is not known.

  • An object instance may be atomically NULL. This means that the value of the entire object is unknown.

Atomic nullity is not the same thing as nonexistence. An atomically NULL instance still exists, its value is just not known. It may be thought of as an existing object with no data.

When working with objects in the OCI, an application can define a NULL indicator structure for each object type used by the application. In most cases, doing so simply requires including the NULL indicator structure generated by OTT along with the struct declaration. When the OTT output header file is included, the NULL indicator struct becomes available to your application.

For each type, the NULL indicator structure includes an atomic NULL indicator (whose type is OCIInd), and a NULL indicator for each attribute of the instance. If the type has an object attribute, the NULL indicator structure includes that attribute's NULL indicator structure. The following example shows the C representations of types with their corresponding NULL indicator structures.

struct address
   OCINumber    no;
   OCIString    *street; 
   OCIString    *state;
   OCIString    *zip;
typedef struct address address;

struct address_ind
  OCIInd    _atomic;
  OCIInd    no;
  OCIInd    street;
  OCIInd    state;
  OCIInd    zip;
typedef struct address_ind address_ind;

struct person 
   OCIString      *fname;
   OCIString      *lname;
   OCINumber      age;
   OCIDate        birthday;
   OCIArray       *dependentsAge;
   OCITable       *prevAddr;
   OCIRaw         *comment1;
   OCILobLocator  *comment2;
   address        addr;
   OCIRef         *spouse;
typedef struct person person;
struct person_ind
  OCIInd        _atomic;
  OCIInd        fname;
  OCIInd        lname;
  OCIInd        age;
  OCIInd        birthday;
  OCIInd        dependentsAge;
  OCIInd        prevAddr;
  OCIInd        comment1;
  OCIInd        comment2;
  address_ind   addr;
  OCIInd        spouse;
typedef struct person_ind person_ind;


The dependentsAge field of person_ind indicates whether the entire varray (dependentsAge field of person) is atomically NULL or not. NULL information of individual elements of dependentsAge can be retrieved through the elemind parameter of a call to OCICollGetElem(). Similarly, the prevAddr field of person_ind indicates whether the entire nested table (prevAddr field of person) is atomically NULL or not. NULL information of individual elements of prevAddr can be retrieved through the elemind parameter of a call to OCICollGetElem().

For an object type instance, the first field of the NULL indicator structure is the atomic NULL indicator, and the remaining fields are the attribute NULL indicators whose layout resembles the layout of the object type instance's attributes.

Checking the value of the atomic NULL indicator allows an application to test whether an instance is atomically NULL. Checking any of the others allows an application to test the NULL status of that attribute, as in the following code sample:

person_ind *my_person_ind
if( my_person_ind -> _atomic == OCI_IND_NULL)
   printf ("instance is atomically NULL\n");
if( my_person_ind -> fname == OCI_IND_NULL)
   printf ("fname attribute is NULL\n");

In the preceding example, the value of the atomic NULL indicator, or one of the attribute NULL indicators, is compared to the predefined value OCI_IND_NULL to test if it is NULL. The following predefined values are available for such a comparison:

  • OCI_IND_NOTNULL, indicating that the value is not NULL

  • OCI_IND_NULL, indicating that the value is NULL

  • OCI_IND_BADNULL, indicates that an enclosing object (or parent object) is NULL. This is used by PL/SQL, and may also be referred to as an INVALID_NULL. For example if a type instance is NULL, then its attributes are INVALID_NULLs.

Use the function OCIObjectGetInd() to retrieve the NULL indicator structure of an object.

If you update an attribute in its C structure, you must also set the NULL indicator for that attribute:

obj->attr1 = string1;
OCIObjectGetInd(envhp, errhp, obj, &ind);
ind->attr1 = OCI_IND_NOTNULL;

See Also:

For more information about OTT-generated NULL indicator structures, refer to Chapter 14, "Using the Object Type Translator with OCI".

Creating Objects

An OCI application can create any object using OCIObjectNew(). To create a persistent object, the application must specify the object table where the new object will reside. This value can be retrieved by calling OCIObjectPinTable(), and it is passed in the table parameter. To create a transient object, the application needs to pass only the type descriptor object (retrieved by calling OCIDescribeAny()) for the type of object being created.

OCIObjectNew() can also be used to create instances of scalars (for example, REF, LOB, string, raw, number, and date) and collections (for example, varray and nested table) by passing the appropriate value for the typecode parameter.

Attribute Values of New Objects

By default, all attributes of a newly created objects have NULL values. After initializing attribute data, the user must change the corresponding NULL status of each attribute to non-NULL.

It is possible to have attributes set to non-NULL values when an object is created. This is accomplished by setting the OCI_OBJECT_NEWNOTNULL attribute of the environment handle to TRUE using OCIAttrSet(). This mode can later be turned off by setting the attribute to FALSE.

If OCI_OBJECT_NEWNOTNULL is set to TRUE, then OCIObjectNew() creates a non-NULL object. The attributes of the object have the default values described in the following table, and the corresponding NULL indicators are set to NOT NULL.

Table 10-4 Attribute Values for New Objects

Attribute Type Default Value

If an object has a REF attribute, the user must set it to a valid REF before flushing the object or an error is returned.


The earliest possible date Oracle allows, which is midnight, 01-JAN-4712 BCE (equivalent to Julian day 1).


The earliest possible date Oracle allows, 01-JAN-4712 BCE (equivalent to Julian day 1).


The earliest possible date and time Oracle allows, which is midnight, 01-JAN-4712 BCE (equivalent to Julian day 1).


The earliest possible date and time Oracle allows, which is midnight, 01-JAN-4712 BCE (equivalent to Julian day 1) at UTC (0:0) time zone.


The earliest possible date and time Oracle allows, which is midnight, 01-JAN-4712 BCE (equivalent to Julian day 1) at UTC (0:0) time zone.












Raw data with length set to 0. Note: the default value for a RAW attribute is the same as that for a NULL RAW attribute.


OCIString with 0 length and first char set to NULL. The default value is the same as that of a NULL string attribute.


OCIString with 0 length and first char set to NULL. The default value is the same as that of a null string attribute.


OCIString with 0 length and first char set to NULL. The default value is the same as that of a null string attribute.


collection with 0 elements.


table with 0 elements.


empty CLOB.


empty BLOB.


The user must initialize the BFILE to a valid value by setting the directory object and filename.

Freeing and Copying Objects

Use OCIObjectFree() to free memory allocated by OCIObjectNew(). An object instance can have attributes that are pointers to additional memory (secondary memory chunks).

Freeing an object deallocates all the memory allocated for the object, including the associated NULL indicator structure and any secondary memory chunks. You must neither explicitly free the secondary memory chunks nor reassign the pointers. Doing so can result in memory leaks as well as memory corruption. This procedure deletes a transient, but not a persistent, object before its lifetime expires. An application should use OCIObjectMarkDelete() to delete a persistent object.

An application can copy one instance to another instance of the same type using OCIObjectCopy().

Object Reference and Type Reference

The object extensions to the OCI provide the application with the flexibility to access the contents of objects using their pointers or their references. The OCI provides the function OCIObjectGetObjectRef() to return a reference to an object given the object's pointer.

For applications that also want to access the type information of objects, the OCI provides the function OCIObjectGetProperty() to return a reference to an object's type descriptor object (TDO), given a pointer to the object.

Creating Objects Based on Object Views or User-Defined OIDs

Applications can use the OCIObjectNew() call to create objects which are based on object views, or on tables with user-defined OIDs (Oracle Internet Directories). If OCIObjectNew() receives a handle to an object view or a table with a user-defined Oracle Internet Directory, then the reference it returns is a pseudo-reference. This pseudo-reference cannot be saved into any other object, but it can be used to fill in the object's attributes so that a primary-key-based reference can be obtained with OCIObjectGetObjectRef().

This process involves the following steps:

  1. Pin the object view or object table on which the new object will be based.

  2. Create a new object using OCIObjectNew(), passing in the handle to the table or view obtained by the pin operation in step 1.

  3. Fill in the necessary values for the object. These include those attributes which make up the user-defined Oracle Internet Directory for the object table or object view.

  4. Use OCIObjectGetObjectRef() to obtain the primary-key-based reference to the object, if necessary. If desired, return to step 2 to create more objects.

  5. Flush the newly created object(s) to the server.

The following sample code shows how this process might be implemented to create a new object for the emp_view object view in the HR schema:

void object_view_new ()
dvoid    *table;
OCIRef   *pkref;
dvoid    *object;
OCIType  *emptdo;
/* Set up the service context, error handle and so on.. */
/* Pin the object view */
OCIObjectPinTable(envp,errorp,svctx, "HR", strlen("HR"), "EMP_VIEW",
    strlen("EMP_VIEW"),(dvoid *) 0, OCI_DURATION_SESSION, (dvoid **) &table);

/* Create a new object instance  */
OCIObjectNew(envp, errorp, svctx, OCI_TYPECODE_OBJECT,(OCIType *)emptdo, table,

/* Populate the attributes of "object" */
/* Allocate an object reference */
OCIObjectNew(envp, errorp, svctx, OCI_TYPECODE_REF, (OCIType *)0, (dvoid *)0,

/* Get the reference using OCIObjectGetObjectRef */
/* Flush new object(s) to server */
} /* end function */

Error Handling in Object Applications

Error handling in OCI applications is the same, whether or not the application uses objects.

See Also:

For more information about function return codes and error messages, see "Error Handling in OCI".

Type Inheritance

Type inheritance of objects has many similarities to inheritance in C++ and Java. You can create an object type as a subtype of an existing object type. The subtype is said to inherit all the attributes and methods (member functions and procedures) of the supertype, which is the original type. Only single inheritance is supported; an object cannot have more than one supertype. The subtype can add new attributes and methods to the ones it inherits. It can also override (redefine the implementation) of any of its inherited methods. A subtype is said to extend (that is, inherit from) its supertype.

As an example, a type Person_t can have a subtype Student_t and a subtype Employee_t. In turn, Student_t can have its own subtype, PartTimeStudent_t. A type declaration must have the flag NOT FINAL so that it can have subtypes. The default is FINAL, which means that the type can have no subtypes.

All types discussed so far in this chapter are FINAL. All types in applications developed before release 9.0 are FINAL. A type that is FINAL can be altered to be NOT FINAL. A NOT FINAL type with no subtypes can be altered to be FINAL. Person_t is declared as NOT FINAL for our example:

( ssn NUMBER,
  name VARCHAR2(30),
  address VARCHAR2(100)) NOT FINAL;

A subtype inherits all the attributes and methods declared in its supertype. It can also declare new attributes and methods, which must have different names than those of the supertype.The keyword UNDER identifies the supertype, like this:

CREATE TYPE Student_t UNDER Person_t
( deptid NUMBER,
  major  VARCHAR2(30)) NOT FINAL;

The newly declared attributes deptid and major belong to the subtype Student_t. The subtype Employee_t is declared as, for example:

CREATE TYPE Employee_t UNDER Person_t
( empid NUMBER,
  mgr   VARCHAR2(30));

The resulting structs generated by OTT for this example are shown at:

This subtype Student_t, can have its own subtype, such as PartTimeStudent_t:

CREATE TYPE PartTimeStudent_t UNDER Student_t
( numhours NUMBER) ;


The benefits of polymorphism derive partially from the property substitutability. Substitutability allows a value of some subtype to be used by code originally written for the supertype, without any specific knowledge of the subtype being needed in advance. The subtype value behaves to the surrounding code just like a value of the supertype would, even if it perhaps uses different mechanisms within its specializations of methods.

Instance substitutability refers to the ability to use an object value of a subtype in a context declared in terms of a supertype. REF substitutability refers to the ability to use a REF to a subtype in a context declared in terms of a REF to a supertype.

REF type attributes are substitutable, that is, an attribute defined as REF T can hold a REF to an instance of T or any of its subtypes.

Object type attributes are substitutable: an attribute defined to be of (an object) type T can hold an instance of T or any of its subtypes.

Collection element types are substitutable: if we define a collection of elements of type T, it can hold instances of type T and any of its subtypes. Here is an example of object attribute substitutability:

( title VARCHAR2(30),
  author Person_t     /* substitutable */);

Thus, a Book_t instance can be created by specifying a title string and a Person_t (or any subtype of Person_t) instance:

Book_t('My Oracle Experience',
        Employee_t(12345, 'Joe', 'SF', 1111, NULL))

NOT INSTANTIABLE Types and Methods

A type can be declared to be NOT INSTANTIABLE, which means that there is no constructor (default or user defined) for the type. Thus, it will not be possible to construct instances of this type. The typical usage would be define instantiable subtypes for such a type. Here is how this property is used:

CREATE TYPE USAddress_t UNDER Address_t(...);
CREATE TYPE IntlAddress_t UNDER Address_t(...);

A method of a type can be declared to be NOT INSTANTIABLE. Declaring a method as NOT INSTANTIABLE means that the type is not providing an implementation for that method. Further, a type that contains any NOT INSTANTIABLE methods must necessarily be declared as NOT INSTANTIABLE. For example:


A subtype of a NOT INSTANTIABLE type can override any of the NOT INSTANTIABLE methods of the supertype and provide concrete implementations. If there are any NOT INSTANTIABLE methods remaining, the subtype must also necessarily be declared as NOT INSTANTIABLE.

A NOT INSTANTIABLE subtype can be defined under an instantiable supertype. Declaring a NOT INSTANTIABLE type to be FINAL is not useful and is not allowed.

OCI Support for Type Inheritance

The following calls support type inheritance.


This function has been enhanced to provide information specific to inherited types. Additional attributes have been added for the properties of inherited types. For example, you can get the supertype of a type.

See Also:

Table 6-7, "Attributes of Types" and Table 6-9, "Attributes of Type Methods" for attributes that OCIDescribeAny() can be used to read

Bind and Define Functions

OCI bind functions support REF, instance and collection element substitutability (subtype instances can be passed in where supertype is expected). There are no changes to the OCI bind interface, since all type checking and conversions are done at the server side.

OCI define functions also support substitutability (subtype instances can be fetched into define variables declared to hold the supertype). However, note that this might require the system to resize the memory to hold the subtype instance.


The client program must use objects that are allocated out of the object cache (and are thus re-sizable) in such scenarios.

The client should not use a struct (allocated on the stack) as the define variable if the value is potentially polymorphic.

See Also:

Chapter 11, "Object-Relational Datatypes in OCI" for details of the bind and define processes


This function will return the REF of the TDO of the most specific type of the input object. This operation will return an error if the user does not have privileges on the most specific type.


This function copies the contents of the source instance to the target instance. The source and target instances must be of the same type. It is not possible to copy between a supertype and a subtype.

Similarly, the tdo argument must describe the same object type as the source and target objects, and must not refer to a subtype or supertype of the source and target objects.


The input element can be a instance of the subtype of the declared type. If the collection is of type Person_t, this function can be used to assign an Employee_t instance as an element of the collection.


The input element can be a instance of the subtype of the declared type; if the collection is of type Person_t, this function can be used to append an Employee_t instance to the collection.


The collection element returned could be an instance of the subtype of the declared type.

OTT Support for Type Inheritance

The Object Type Translator (OTT) supports type inheritance of objects by declaring first the inherited attributes in an encapsulated struct called '_super', followed by the new declared attributes. This is done because C does not support type inheritance.

See Also:

"OTT Support for Type Inheritance" for an example and discussion

Type Evolution

Adding, dropping and modifying type attributes are supported. This concept is known as type evolution. It is discussed in:

OCIDescribeAny() will return information about the latest version of the requested type if the type of the input object is OCI_OTYPE_NAME, and the type of the described object is OCI_PTYPE_TYPE, that is, the name input to OCIDescribeAny() is a type name.

See Also:

OCITypeArrayByName() and OCITypeByName(). To access type information use these functions, as well as OCIDescribeAny().

See Also:

"Type Evolution and the Object Cache" for a discussion of the impact of type evolution on the object cache