CIF 3

Java code generation

The CIF 3 code generator can generate Java code from a CIF specification. It is assumed the reader of this page is familiar with the general information of the CIF 3 code generator tool. This page describes specific information applicable only to Java code generation.

Supported specifications

The CIF 3 code generator supports a subset of CIF specifications. Generation of Java code does not impose additional restrictions. See the Supported specifications section of the CIF 3 code generator page for more information.

Options

Besides the common options that apply to all target languages/platforms of the CIF 3 code generator, the Java code generation can be influenced by the following additional options:

  • Java package: The Java package in which to generate Java code. If not specified, the code prefix is used. The specified package should be a valid Java package name, and must not be empty.

The code prefix that can be configured using a common option, is used in the generated Java code as prefix for class names.

Compilation

The generated Java code can be compiled with a Java 7 compiler or newer.

The generated code is complete, and can be used as is, without any CIF related library or runtime. No additional dependencies or libraries are required. Only the standard Java libraries, part of any Java installation, are needed.

Generated files

Java code generation leads to two Java files being generated:

  • <prefix>.java
  • <prefix>Test.java

where <prefix> is replaced by the code prefix configured using the Code prefix option.

The first file contains the code for all the features of the CIF model in an abstract class named <prefix>. The second file contains a class named <prefix>Test, which inherits from the <prefix> class, and implements various abstract methods. This test class can be used to test the code, and can serve as a starting point for actually implementing the coupling between the code generated for the CIF model, and the environment in which it runs.

Executing the code

The code can be executed in multiple ways:

public void execOnce(double newTime)

Invoke this method to execute the code once. Inputs are read, transitions are executed until none are possible, output are written, etc.

The newTime is the time in seconds, since the start of the first execution of this method. For each invocation, the time that has elapsed so far should be provided. This becomes the new value of variable time.

Using this method, you decide when you invoke this method, and with what values to use for variable time. This approach allows full control.

public void exec(long frequency)

Invoke this method to execute the code repeatedly, without ever stopping. This repeatedly invokes the execOnce method.

The frequency can be used to configure how often, in number of times per second, the code should be executed. By given a frequency of 100, the method attempts to execute the code every 10 milliseconds. If a single execution takes less than 10 milliseconds, the method will sleep for the remainder of what is left of the 10 milliseconds. However, if a single execution takes more than 10 milliseconds, the next execution is immediately started.

It is also possible to execute the code repeatedly, as fast as possible, without any delays. As soon as a single execution has finished, another execution is started. This kind of behavior can be obtained by providing a negative or zero value for frequency.

Environment interface

The code generated for the CIF model does not do anything useful by itself. Only once it is coupled to the environment, can the CIF model become useful. The following methods are provided in the interface of the generated class as abstract methods, and need to be implemented in a derived class:

protected void updateInputs()

In this method, you should update the values of the input variables of the CIF model. Note that you should obtain these from the environment. You should not access the state of the CIF model from this method, as it may not be initialized or up-to-date yet.

An example of how to update the input variables, can be found in the generated <prefix>Test class. There, each input variable is assigned the default value for its data type.

If you want to update the variable, you should always assign completely new fresh values. Don’t modify arrays and tuples in-place, as that may lead to other variables being changed as well.

protected void preExec()

This method is invoked each time, just before the code for the CIF model is executed. This method is invoked before the updateInputs method is invoked.

You should not access the state of the CIF model from this method, as it may not be initialized or up-to-date yet.

protected void postExec()

This method is invoked each time, just after the code for the CIF model is executed.

In this method, you should write the output variables of the CIF model to the environment. Since the code for the CIF model was just executed, these variables may have new values.

All discrete and continuous variables of the CIF model are available, as are variable time and the current locations of all of the automata of the CIF model.

The code to write the output values to the environment is the dual to reading the inputs from the environment, as is done in the updateInputs method.

protected void infoExec(long duration, long cycleTime)

This method informs about the duration of a single execution of the code generated for the CIF model.

The duration is the total number of nanoseconds spent executing the code. The cycleTime is the desired maximum duration of the execution, in nanoseconds, or -1 if not available.

You can use this method to detect when the code runs longer than the desired maximum duration, and thus the desired execution frequency can not be achieved. You can also use this method to figure out the variability of the duration of execution.

The doInfoExec field of the class can be used to configure whether this method is invoked during execution by the exec method. This is enabled (true) by default.

protected void infoEvent(int idx, boolean pre)

This method informs about events that are about to be executed or have just been executed.

The idx is the 0-based index of the event. You can feed this index to the getEventName method to obtain the absolute name of the event. The pre is true if the event is about to be executed, and false if the event has just been executed.

You can use this method to be informed about what the code does during execution. You could for instance log that for debugging.

The doInfoEvent field of the class can be used to configure whether this method is invoked during execution. This is disabled (false) by default.

protected void infoPrintOutput(String text, String target)

This method informs that new print output has been generated during execution.

The text is the text that is to be printed. The target indicates the file or special target to which text is to be printed. If printed to a file, an absolute or relative local file system path is given. Paths may contain both / and \ as path separators. Supply the path to the <prefix>Utils.normalizePrintTarget method to normalize the path to use path separators for the current platform on which the Java code is executed. There are two special targets: :stdout to print to the standard output stream, and :stderr to print to the standard error stream.

You should use this method to actually print the text to standard output, standard error, or files. An example of how to do this, can be found in the generated <prefix>Test class.

The doInfoPrintOutput field of the class can be used to configure whether this method is invoked during execution. This is enabled (true) by default.

Data access and manipulation

For each CIF constant, a field is generated in the Java class. Similar fields are generated for the state variables (discrete and continuous variables), and the input variables. A time field is always present and contains the current model time.

For each internal user-defined function of the CIF model, a method is generated in the Java class. Similarly, a method is generated for each algebraic variable of the CIF model, as well as for the derivative of each continuous variable (except variable time).

The standard library functions and operators that are supported by the code generation, are also available, in the <prefix>`Utils inner class. Not all operators and standard library functions are available. Only those that have no corresponding Java operator or method, or behave differently with respect to runtime errors, are available in this inner class.

For more information on the names of the generated field and methods, see the section on naming below.

Data types

The following table lists CIF types supported by the code generator, and their equivalent in the generated Java code:

CIF type Java type Tuple postfix [4]
bool boolean / java.lang.Boolean [1] B
int (with or without range) int / java.lang.Integer [1] I
real double / java.lang.Double [1] R
string java.lang.String S
list[...] t java.util.List<t> [2] Lt, with t a postfix for the element type
enum e = ... <prefix>Enum inner enumeration (class) [3] E
tuple(...) CifTuple_T... inner class [4] TnFFF, with n the number of fields, and FFF a postfix for the type of each field
[1](1, 2, 3) The primitive Java type is used if possible, and the class variant is used instead if necessary, for instance for element types of a java.util.List.
[2]The code generator does not support CIF lists in general. Only CIF arrays are supported.
[3]Due to linearization, all enumerations are merged together to a single enumeration. As such, always exactly one Java enum is generated for a CIF model.
[4](1, 2) For each unique tuple type (ignoring integer and array ranges of the types of the fields, as well as field names) an inner Java class is generated. Their names all start with CifTuple_, and end in a postfix that describes the type. The third column in the table lists the texts used for each type in the postfix. For instance, for a tuple(tuple(int x; real y; string z) a, list[3] int b) type, the tuple class would be named CifTuple_T2T3IRSLI.

Runtime errors

Some CIF models that are syntactically valid, may lead to runtime errors when simulated or executed. For instance, there may be a division by zero, an out of bounds projection, or an assignment may lead to the range of the assigned variable being violated. The generated Java code detects all such problems and throws a <prefix>Exception in such cases. This exception class is an inner class of the <prefix> class. The exception indicates what caused the runtime error, using an informative end-user readable message.

Naming

The generated code will contain names for variables, functions, etc. The names in the generated code are based on the absolute names of the objects in the original CIF model. The names are further influenced by the linearization algorithm. Essentially, the absolute names are used, where each . is replaced by a _. Furthermore, a _ is added at the end, to avoid conflicts with other non-generated names, Java keywords, etc. For instance, for an automaton a with a discrete variable b in it, the absolute name is a.b. In the Java code, the variable will be named a_b_.

The code generator ensures unique names in the generated Java code. If the same Java name results from two different CIF objects, one of them is renamed, by adding a 2, or 3, or 4, etc to make it unique. For instance, if the CIF model has a constant named a_b in the top level scope, and a constant named b in the a automaton, both would be named a_b_. One of them is renamed to a_b_2. If such renaming takes place, a warning is printed to the console.

CIF model Java code Example CIF name Example Java name
Constant Field a.b a_b_
Discrete variable Field a.b a_b_
Continuous variable Field a.b a_b_
Derivative of a continuous variable Method a.b' a_b_deriv()
Algebraic variable Method a.b a_b_()
Input variable Field a.b a_b_
User-defined function Method a.b a_b_()
Parameter of a user-defined function Method parameter a.b a_b_
Local variable of an internal user-defined function Local variable of a method a.b a_b_
Enumeration Enumeration a.b <prefix>Enum, [5]
Enumeration literal Enumeration constant a.b.c <prefix>Enum._c [5]
Tuple type Inner class n/a CifTuple_T... [6]
[5](1, 2) Due to linearization, all enumerations are merged together to a single enumeration. As such, always exactly one Java enum is generated for a CIF model.
[6]For details on tuple types and the classes that are generated for them, see the data types section above.