Skip to main content

SA Engine Java Interfaces

Stream Analyze Sweden AB
Version 3.0

This document describes the external interfaces between SA Engine and the programming language Java. There are mainly two ways to interface SA Engine with Java programs: In the client interface Java programs call SA Engine, while in the plug-in interface foreign OSQL functions are implemented as Java methods. The combination is also possible where foreign functions in Java call SA Engine back through the client interface.


There are two main kinds of external interfaces, the client and the plug-in interfaces:

  • With the client interface a program in Java calls SA Engine. The client interface allows OSQL queries and function calls to be shipped from application programs to remote SA Engine servers or to an embedded SA Engine system running in the same process as the application.
  • With the plug-in interface OSQL functions are implemented as public methods in Java. The foreign Java functions are executed in the same process as SA Engine. The client interface can be used also in foreign Java function implementations.

The result of an OSQL query or function call is an object stream, which is a possibly infinite stream of objects. The client interface provides primitives to map over the elements in such object streams. Analogously the plug-in interface provides API primitives to produce object streams.

This documentation describes the SA Engine API for Java (version 8 and higher) where the new "lambda-functions" enable a very elegant object stream interface where lambda functions in Java are used as callback functions. It is possible to use SA Engine also for older Java versions by using lower level interface primitives not described here.

This documentation introduces the Java API of SA Engine through a number of example programs whose source codes are in the folders sa.engine/demo/*/Java of an installed SA Engine system. In that folder you will find a number of examples for how to use the Java API. You are assumed to be familiar with OSQL.

To compile an example Java program using the Java API of SA Engine you must make sure that your CLASSPATH includes the file sa.engine/bin/sa_Java.jar that implements the API. If SA Engine is installed and CLASSPATH set correctly you can compile the Java program in folder sa.engine/demo/Hello/Java on your PC with:

javac -cp "../../../bin/sa_Java.jar" *.java

The client interface

With the client interface there are two ways to call SA Engine from Java:

  • In the query interface strings containing OSQL statements are sent to SA Engine for dynamic evaluation. The result from a query is an object stream. The forEach interface in Java 8 and beyond provides a powerful mechanism where the system iterates over object streams and applies user lambda-functions (callback methods) for each element in the received object stream. The embedded query interface is relatively slow since the OSQL statements must be parsed and compiled at run time.
  • In the function interface SA Engine functions are directly called from Java, without the overhead of dynamically parsing and executing OSQL statements. The result of a function call is an object stream. The function interface is significantly faster than the query interface. It is therefore recommended to always define SA Engine functions stored in the local database for the various SA Engine operations performed by the application and then use the function interface to invoke them directly.

When calling SA Engine from application programs, the application usually often runs as a client to an SA Engine server (SAS) running on some other computer. This is called the remote connection. With the remote connection several applications running in different locations can remotely access the same SAS concurrently. The Java applications and the SAS run as separate programs so that the server will survive client crashes and vice versa.

It is also possible to run the system as an embedded SA Engine system in the same process and memory address space as the client application. Several client threads can thereby concurrently access the embedded SA Engine thread. This is called the local connection.


A Java object of class Connection represents connections to an SA Engine server. The connection is established when the connection object is created using the constructors:

new Connection(String p)

new Connection()

The constructor Connection(p) will establish a remote connection to an SA Engine peer named p that can be one of:

  • an SA Engine client running on an edge device,
  • an SA Engine stream server (SAS) coordinating, communication with edge devices, or
  • a name server, which is a SAS that keeps track of all other peers in a federation of SA Engine peers.

The format of the string p can be:


If just a peer name is specified the peer must be a local peer known by the name server running on the same computer as the client. The local name server can be reached using the peer name nameserver. If peer@host is specified a connection is established to a peer running on the name server of the specified host. Specifying peer@localhost is equivalent to just peer. The name server by default listens on port 35021; the format peer@host:portno is used when the name server on that host uses some other port.

When the connection constructor has no argument, a local connection is established to an embedded SA Engine system running in the same main memory process as the caller.

The query interface

In the query interface, strings being OSQL statements are sent to SA Engine stream servers or edges for execution. The following is a Java application that prints the five first natural number in file sa.engine/demo/clent/Java/

public class QueryRange {
public static void main(String argv[]) {
Connection c = new Connection("p");
ObjectStream s = c.query("range(1,5)");
s.mapAll(e -> System.out.println(e.getIntElem(0)));

Example 1. Mapping over the result from a query

The constructor Connection("p") first opens connection c to a peer named "p" on the local computer.

The method call c.query(q) sends a query string q to be executed by the peer to which c connects. The query will there be compiled and optimized by SA Engine and then an object stream s of class ObjectStream is constructed by SA Engine to represent the result stream of the query. The application utilizes the lambda expression of Java ( to map over object streams using the mapAll method. In the example a lambda expression is applied on all elements e in the result object stream from the OSQL query range(1,5).

An object stream element e is a tuple of one or several objects. In the example each e is a tuple containing a single integer. To access the object in position pos of a tuple e the method e.GetIntElem(pos) is called. The tuple positions are enumerated from 0 and up.

The example shows how execute a single OSQL query where the system will block the current thread to wait for tuples to arrive on the object stream. Several OSQL queries can execute in parallel in separate Java threads.

A common case is that one or several OSQL statements are sent to the server for synchronous evaluation and then immediately waiting for them to finish. This is made by applying the method run() on the result stream from the method query(). For example:

Tuple res = c.query("create function add1(Number x) -> Number as x + 1").run();

Here, the method run() iterates over the entire result object stream res to execute the create function OSQL statement. run() returns the last element of the object stream when the query is completed. In the example the object stream contains only the object representing the definition of the created OSQL function add1, which will be returned from run(). Using the method run() for queries returning infinite streams will cause an infinite loop.

The function interface

The time to dynamically compile and optimize a query by SA Engine can be rather long, so a better way is to directly call OSQL functions through the function interface. The following Java code in file sa.engine/demo/client/Java/ calls the OSQL function range(1,5):

public class CallRange {
public static void main(String argv[]) {
Connection c = new Connection("p");
Tuple argl = Tuple.make(1, 5);
ObjectStream s ="range", argl);
s.mapAll (e -> System.out.println(e.getIntElem(0)));

Example 2. Calling a function and mapping over the result

Most of the code is the same as in The only difference is that an argument tuple (1,5) bound to argl is first constructed by calling the variadic Java method Tuple.make(1,5) and that the function named "range" with arguments in tuple argl is called by method"range", argl). The argument tuple argl represents arguments of OSQL function calls from Java. An object stream is returned by method on which a lambda expression to print all result elements e is applied.

Mapping over infinite streams

The results of queries and function calls returning streams are also represented as object streams. There is no upper limit on how many elements can be retrieved from such an object stream. The system is able to handle object streams s containing indefinite numbers of elements, in which case the mapping will never terminate. In such cases the mapping will have to be done with the method map() where the lambda expression returns a Boolean continuation flag that terminates the mapping when false. Alternatively, the mapping can be terminated by throwing an exception.

For example, the following query in sa.engine/demo/client/Java/ prints the numbers 0.0 and 0.1 by iterating over the two first elements of the indefinite stream returned by the OSQL call heartbeat(0.1):

public class CallHeartbeat {
public static void main(String argv[]) {
Connection c = new Connection("p");
Tuple argl = Tuple.make(0.01);
ObjectStream s ="heartbeat", argl);
int[] cnt = {1}; -> {
return cnt[0]++ < 2;

Example 3. Example on how to stop an infinite object stream

When the lambda expression returns the continuation flag true the mapping continues, while false will terminate it. In the example it is terminated after printing the two first elements e of the object stream. Notice that lambda expressions in Java require the counter cnt to be called by reference by defining it as an array.

Implementing foreign Java functions

Foreign OSQL functions in Java are defined as methods in public classes in Java class files that are dynamically loaded into SA Engine. This chapter describes through examples the different kinds of foreign functions definable in Java.

A Hello World foreign function

As a first very simple example, we implement a foreign Java function hello() -> Charstring that returns the string "Hello World". The file sa.engine/demo/Hello/Java/ has the following contents:

public class Hello {
public void helloF(CallContext cxt, Tuple tpl) {
tpl.setElem(0,"Hello World");

Example 4. Simplest foreign function in Java returning the string "Hello World"

The parameter cxt is a context object of class CallContext for managing the call by SA Engine, and tpl is a parameter tuple of class Tuple representing both the arguments (inputs) and results (outputs) of the foreign function call.

In the example the function binds the unbound parameter position 0 of the parameter tuple tpl to the desired result string "Hello World" by calling the method tpl.setElem(0, "Hello World"). Positions of tuples are enumerated 0 and up; in this case there is no argument so the result is in position 0.

When the result has been filled in using tpl.setElem() the entire tuple tpl is sent to SA Engine using the method cxt.emit(tpl).

Compile with the command:

javac -cp "../../../bin/sa_Java.jar" *.java

After the compilation, start the SA Engine console REPL with the shell command:


In the console REPL, register the external Java function implementation by executing the OSQL statement:

create function hello() -> Charstring
as foreign 'JAVA:Hello/helloF';

The function can then be called immediately in the REPL:


You will notice that, if needed, SA Engine first dynamically loads a Java Virtual Machine (JVM) to execute the call. The call will fail if no JVM is found. If it fails to load the JVM, make sure that a suitable JVM can be reached from the folder where the console REPL was started. If the JVM cannot be found, set the variable JAVA_HOME to the Java home folder.

The OSQL script sa.engine/demo/Hello/validate.osql tests that Hello World works correctly. It is recommended that you make such test scripts for all new foreign functions. You can run the test script with the OS command:

sa.engine -O test.osql

The foreign Java function definitions are permanently saved in the database image when saved on disk so that they will be reloaded when SA Engine is started again with the saved image.

Once Hello World works you know that you have set up the environment for compiling and plugging in Java code to SA Engine correctly. You are ready to define any foreign Java function.

Foreign function with arguments and result

The arguments of a foreign OSQL function in Java are stored in the parameter tuple as well as the results. For example, the function myconcat(Charstring x, Charstring y) -> Charstring concatenates strings x and y. It has the following implementation as method myconcatBBF in sa.engine/demo/Basic/Java/

public void myconcatBBF (CallContext cxt, Tuple tpl) { 
String x = tpl.getStringElem(0); // Pick up 1st argument x
String y = tpl.getStringElem(1); // Pick up 2nd argument x
tpl.setElem(2, x+y);

Example 5. Foreign Java function implementation taking two parameters

In this case the arguments x and y of myconcat(x,y) are in positions zero and one of the parameter tuple tpl and the computed result is bound in position two.

Compile the Java code in folder sa.engine/demo/Basic/Java/ with

javac -cp "../../../bin/sa_Java.jar" *.java

The OSQL code of myconcat() in file sa.engine/demo/Basic/Java/definitions.osql is defined as:

create function myconcat(Charstring x, Charstring y) -> Charstring
as foreign 'JAVA:Simple/myconcatBBF';

The OSQL script sa.engine/demo/Basic/validate.osql includes validation of myconcat(). The code in the Java folder can be tested with:

sa.engine -O test.osql

Foreign function with several results

Foreign functions can also return more than one value in a simple result tuple. For example, the foreign function sqrt2(Number x)->(Number pos,Number neg) returns both the positive and negative square roots of number x. It is implemented as method sqrt2BFF in file sa.engine/demo/Basic/Java/

public void sqrt2BFF(CallContext cxt, Tuple tpl) { 
double root = Math.sqrt(tpl.getDoubleElem(0));
tpl.setElem(1, root);
tpl.setElem(2, -root);

Example 6. Foreign Java function implementation returning tuple of several values

In this case the foreign function has a single input argument in position 0 of tuple tpl. The method call tpl.getDoubleElem(0)converts the input argument into a floating point number. An exception is raised if the argument is not a number. The method tpl.setElem is called twice to set the two result elements in the parameter tuple tpl.

The OSQL code of sqrt2() is:

create function sqrt2(Number x) -> (Number neg, Number pos)
as foreign 'JAVA:Simple/sqrt2BFF';

Foreign function returning a vector

The OSQL type Vector is represented in Java as arrays. The function vsqrt2(Number x) -> Vector of Number returns the negative and positive square root of number x. It is implemented as method vsqrt2BF in file sa.engine/demo/Basic/Java/

public void vsqrt2BF (CallContext cxt, Tuple tpl) { 
double x = tpl.getDoubleElem(0); // Pick up argument x
if(x >= 0) {
double root = Math.sqrt(x);
double[] roots = {-root, root};
tpl.setElem(1, roots);

Example 7. Foreign function method returning a vector

In this case the foreign function has a single input argument in position 0 of the parameter tuple tpl. The method call tpl.getDoubleElem(0) converts the input argument into a floating point number. An exception is raised if the argument is not a number. The method tpl.setElem converts the Java array root to an OSQL object of type Vector.

The OSQL definition of vsqrt2() is:

create function vsqrt2(Number x) -> Vector of Number
as foreign 'JAVA:Simple/vsqrt2BF';

Foreign function taking vectors as arguments

As an example of a function taking vectors as arguments, the function dotprod(Vector v, Vector w) ->Number returns the Cartesian product of vectors v and w. It is implemented as method dotprodBBF in file sa.engine/demo/Basic/Java/ as:

public void dotprodBBF(CallContext cxt, Tuple tpl)
double[] v = tpl.getDoubleArrayElem(0); // First argument
double[] w = tpl.getDoubleArrayElem(1); // Second argument
double prod = 0;
if (v.length! = w.length)
throw new sa_Exception("Array index out of bounds");
for(int i=0; i < v.length; i++) {
prod = prod + v[i]*w[i];

Example 8. Foreign Java function implementation taking vectors as arguments

The method call tpl.getDoubleArrayElem(pos) returns the element at position pos of tuple tpl as a Java array.

The expression throw new sa_Exception(msg) raises an SA Engine exception.

The OSQL definition of dotprod() is:

create function dotprod(Vector v, Vector w) -> Number
as foreign 'JAVA:Collections/dotprodBBF';

Foreign function generating a bag

Foreign functions can return bags of values. For example, the foreign function natural(Numberm, Number n)->Bag of Number returns a bag of the integers from m to n. It is implemented as method naturalBBF in file sa.engine/demo/Basic/Java/

    public void naturalBBF(CallContext cxt, Tuple tpl)
int m = tpl.getIntElem(0); // Pick up integer m in 1st argument
int n = tpl.getIntElem(1); // Pick up integer n in 2nd argument
for(int i = m; i<=n; i++) {
tpl.setElem(2, i); // Bind result bag element
cxt.emit(tpl); // Emit result bag element

Example 9. Foreign Java function implementation returning a bag of numbers from m to n

The method call tpl.getIntElem(p) converts the input argument in position p of tpl to an integer. The method cxt.emit(tpl) is called several times to iteratively emit each element of the result bag of numbers.

The OSQL definition of natural(m,n) is:

create function natural(Number m, Number n) -> Bag of Number
as foreign 'JAVA:Collections/naturalBBF';

Foreign aggregate function

To implement foreign aggregate functions in Java that iterate over collections (bags, streams or vectors) the method is used. For example, the function sqsum(Bag of Number b)->Number will sum up the square all numbers in bag b. It is implemented as method sqsumBF in file sa.engine/demo/Aggregation/Java/

public void sqsumBF(CallContext cxt, Tuple tpl) {
double[] sum = {0};
cxt.mapAll(tpl.getOidElem(0), e -> {
double x = e.getDoubleElem(0);
sum[0] += x*x;
tpl.setElem(1, sum[0]);

Example 10. Implementation of an aggregate function sqsum over a bag of numbers

If you call sqsum(range(1,10)) from the console REPL you will get the expected result 55.

The method CallContext.mapAll(coll, l) takes as arguments a collection coll to map over and a lambda expression that is applied on each element e in the result object stream.

The OSQL definition of sqsum() is:

create function sqsum(Bag b) -> Number
as foreign 'JAVA:Aggregation/sqsumBF';

Aggregation over vectors

A foreign aggregate function implementation over bags can also be used for aggregating over vectors. For example, the method sqsumBF above can also be used for computing the sum of the square of numbers in vector by defining the function:

create function sqsum(Vector v) -> Number
as foreign 'JAVA:Aggregation/sqsumBF';

Aggregation over finite streams

A foreign aggregate function implementation over bags can also be used for aggregating over finite streams. For example, the method sqsumBF above can also be used for computing the sum of the square of numbers in a finite stream s by defining the function:

create function sqsum(Stream s) -\> Number
as foreign 'JAVA:Aggregation/sqsumBF';

Foreign function returning a record

Records in SA Engine are represented as class Record in Java. For example, the function rsqrt2(Number x)->Record returns the square root of x as a record {"neg": -r "pos": r}. It is implemented as method rsqrt2BF in file sa.engine/demo/Basic/Java/

public void rsqrt2BF(CallContext cxt, Tuple tpl) {
double x = tpl.getDoubleElem(0);
if(x >= 0) {
double root = Math.sqrt(x);
// Construct record {"neg":-root,"pos":root}:
Record r = new Record();
r.put("neg", -root);
r.put("pos", root);
cxt.emit(tpl); // Emit record r

Example 11. Foreign function returning a record.

The method call r.put(a,v) sets the value for attribute a in in record r to v.

The OSQL definition of rsqrt2() is:

create function rsqrt2(Number x) -> Record
as foreign 'JAVA:Collections/rsqrt2BF';

Foreign function accessing a record

The function getnum(Record r,Charstring a)->Number returns attribute a in record r as a number. It is implemented as method getnumBBF in file sa.engine/demo/Basic/Java/

public void getnumBBF(CallContext cxt, Tuple tpl) {
Record r = tpl.getRecordElem(0);
String field = tpl.getStringElem(1);
Tuple val = r.get(field);
if(val == null) return; // Returns nil
double x = val.getDoubleElem(0);
tpl.setElem(2, x);

Example 12. Foreign function accessing a record.

The OSQL definition of getnum() is:

create function getnum(Record r, Charstring field) -> Number
as foreign 'JAVA:Collections/getnumBBF';

Foreign function returning an infinite stream

The same mechanism as for bags is used for returning (possibly infinite) streams of values. For example, the function natural_numbers()->Stream of Number returns an infinite stream of the natural numbers (integers from one to infinity). It is implemented as method natural_numbersF in file sa.engine/demo/Streams/Java/

public void natural_numbersF(CallContext cxt, Tuple tpl) {
int i=0;
while(true) {
tpl.setElem(0, i++);

Example 13. Foreign Java function implementation returning an infinite stream of all natural numbers

If you call natural_numbers() from the console REPL an infinite stream of number is returned and the system will print natural numbers until you interrupt it with CTRL-C. The call section(natural_numbers(), 10, 20) will return a finite stream.

Foreign stream transformation function

A stream transformation function takes a stream as argument and produces a new transformed stream as result. For example, the function power_stream(Stream s,Number n) -> Stream of Number generates a stream of the numbers xx in stream s. It has the following implementation in sa.engine/demo/Streams/Java/

public void power_streamBBF(CallContext cxt, Tuple tpl) {
double exp = tpl.getDoubleElem(1);
cxt.mapAll(tpl.getOidElem(0), e -> {
double x = e.getDoubleElem(0)
tpl.setElem(2,Math.pow(x, exp));

Example 14. Implementation of a stream transformation function

The important thing here is that emit() is called for each stream element in the lambda expression.

The OSQL definition of power_stream() is:

create function power_stream(Stream of Number s, Number n) -> Stream of Number
as foreign 'JAVA:Streams/power_streamBBF';

Exception handling

SA Engine can raise two kinds of runtime Java exceptions: NoMoreData and sa_Exception. The user need not declare or catch these runtime exceptions, but sometimes a try ... finally ... construct is needed in order to guarantee that resources allocated by the foreign function are always freed. The error message for an SA Engine exception can be obtained by calling the Java system exception method getMessage().

The exception NoMoreData is raised when an application indicates that no more data is needed.

The exception sa_Exception is raised by the system when some error is detected. To raise a new exception with a new error message, sa_Exception has a constructor that takes an error string as parameter.

The cause of an sa_Exception can be investigated through the following sa_Exception attributes:

errstr: Attribute holding the error message string.

errform: Attribute holding the object causing the error.

Multi-directional foreign functions

Foreign OSQL functions can be made invertible. For example, assume a foreign square root function:

create function sqroot(Number x) -> Bag of (Number r)
as foreign 'JAVA:Multi/sqrootBF';

Its definition if file sa.engine/demo/Basic/Java/ is:

public class Multi {
public void sqrootBF(CallContext cxt, Tuple tpl) {
double x = tpl.getDoubleElem(0);
if(x==0) {
tpl.setElem(1, 0);
else if(x > 0) {
double root = Math.sqrt(x);
tpl.setElem(1, root);
tpl.setElem(1, -root);

Example 15. Implementation of the foreign function sqroot(number x) -> Bag of (Number r)

The bag of numbers 2.0 and -2.0 will be returned by the OSQL query sqroot(4).

If sqroot(x) were invertible we could also make a query calling its inverse:

select x from Number x where 2 in sqroot(x); /* Result is 4.0 */

Since the definition above is not multidirectional the system will raise an error that the query is not executable because variable x is unknown.

We now define sqroot as an invertible foreign Java function by redefining it as:

create function sqroot(Number x) -> Bag of (Number r)
as multidirectional ('bf' foreign 'JAVA:Sqrt/sqrootBF')
('fb' foreign 'JAVA:Sqrt/sqrootFB');

after first having added the method sqrootFB in

    public void positive_sqrtFB(CallContext cxt, Tuple tpl) { 
double x = tpl.getDoubleElem(1);
if(x > 0)
tpl.setElem(0, x*x);

Example 16. The Java implementation of the inverse to the sqroot foreign function

The method sqrootFB implements the inverse of sqroot. In this case position one in tpl is bound and position zero is filled in.

As in the example, multi-directional foreign functions are functions that can be executed also when arguments are unknown. The benefit of multi-directional foreign functions is that a larger class of queries calling the function is executable (safe), and that the system can make better query optimization.

A multi-directional foreign function has several implementations depending on the binding pattern of its arguments and results. The binding pattern is a string of 'b':s and 'f':s, indicating which arguments or results in a given implementation are known or unknown, respectively.

A simple foreign OSQL function is a special case where all arguments are known and all results are unknown.

To implement a multi-directional function you first need to think of for which binding patterns implementations are needed. In the sqroot case one implementation handles the two square roots and the other one handles the inverse square.

The file sa.engine/demo/Basic/Java/ contains the code to implement multi-directional sqroots.

Non-blocking parallelism

The system is normally locked by the SA Engine kernel while a foreign function is running so that the programmer can assume that SA Engine can be called from foreign Java functions without explicitly locking the Java thread. Therefore, foreign functions may block while waiting for some resource, making the system perform badly or even hang when foreign functions in separate threads wait for some resource. This is not acceptable if a foreign function waits for some event to occur or when long-running or parallel computations are made, e.g. by GPU co-processors.

To allow for non-blocking kernel computations, SA Engine provides the ability for Java-code kernels in foreign functions to run in parallel in the background. The typical code pattern is:

    public void backgroundbf(CallContext cxt, Tuple tpl)
throws sa_Exception
Oid bg = cxt.getBG(); // Pick up object holding current background thread
... // Initialize
... // Do something not calling SA Engine here
cxt.emit(..); // Emit result

Three methods of class CallContext are provided for paralell processing of kernel code:

Oid CallContext.getBg()
CallContext enterBG(Oid bg)
leaveBG(Oid bj)
  1. First the method getBG() is called to obtain a background object bg of type Oid representing the (current) thread where the parallelism is going to take place.

  2. Then the background object is passed as argument to the method enterBG(bg) to indicate that the code section following the call is executed in the background. Here, for example, long running computations or GPU kernel computations can run.

  3. Finally, when the background code has finished a call to leaveBG(bg) indicates that the background computation has finished.

The code executed in the background is not allowed to call SA Engine, except for the call to enterBG.

You can call enterBG/leaveBG several times in the same foreign function, but, since the calls are relatively slow, it is recommended not to call them too often.

Data objects

The Java API uses a number of Java classes and methods documented in this section.


The class Tuple is a commonly used class in the SA Engine Java interface. A tuple represents an ordered finite sequence of SA Engine objects.

Tuples are used for many purposes in the interfaces:

Tuple creation

A new tuple with size s is created with the constructor Tuple(s).

If all elements of the new tuple are known in advance, you can also construct it using the variadic constructor Tuple.make(...), for example Tuple.make(1,2,3).

The method t.getArity() returns the number of elements in tuple t.

The elements of a tuple are enumerated starting at zero and can be accessed through a number of tuple access functions specific for each element class, as described next.

Integer elements

To access an integer stored in position pos of tuple t, call the method t.getIntElem(pos). Floating point numbers are rounded to the closest integer. An error is generated if there is no number in the specified position of the tuple.

To store integer i in element pos of tuple t, call the method t.setElem(pos, i).

Floating point elements

To get a double precision floating point number stored in position pos of tuple t, call the method t.getDoubleElem(pos). Integers are converted to floating point numbers. An error is generated if there is no number in the specified position of the tuple.

To store floating point number x in element pos of tuple t, call the method t.setElem(pos,x).

String elements

To get a string stored in position pos of tuple t, call the method t.getStringElemt(pos). An error is generated if the element is not a string.

To store a string str in element pos of tuple t, call the method t.setElem(pos,str).

Data type of element

The data type of element pos of tuple t can be tested with the following Boolean methods:









OSQL vectors (type Vector) are represented as Java arrays. To get an array stored in position pos of tuple t, call the method t.getArrayElem(pos). An error is generated if there is no vector in the specified position of the tuple.

To store a copy of array a as an OSQL vector in element pos of tuple t, call the method t.setElem(pos,a).


OSQL records (type Record) are represented in Java as objects of class Record. To access a record stored in position pos of tuple t, call the method t.getRecordElem(pos). An error is generated if there is no record in position pos of tuple t.

As for tuples there are a number of methods for accessing the element of a record depending on their type:


Values are internally stored as single element tuples in a record. To retrieve the tuple t holding the value of key k in record r call Tuple t = r.get(k). The call r.getInt("k") is thus equivalent to r.get("k").getIntElem(0).

To store a copy of record r as an OSQL record in element pos of tuple t, call the method t.setElem(pos,r).

To set value v to key k for record r call the method r.put(k,v).

Object proxies

An object proxy is a Java object representing a corresponding referenced OSQL object accessed through a connection. Object proxies can reference any kind of data stored in SA Engine, including numbers, strings, surrogate objects, arrays, and internal SA Engine data structures. Object proxies are represented using the Java class Oid.

To get an object proxy representing the OSQL object stored in position pos of tuple t, call the method t.getElem(pos). If there is no element in position pos, Java's null is returned.

To store an OSQL object proxy o in element pos of tuple t, call the method t.setElem(pos,t).

The name of the SA Engine data type of an object proxy o can be retrieved as a string by calling the method o.getTypename().

The proxy object representing the data type of proxy object o can be retrieved by calling the method o.getType().

The connection to the SA Engine system owning an object proxy o can be retrieved by calling the method o.getConnection().