jjtree-releases.notes

======================================
Starting with 6.0
======================================
JJTree is extended to support C++ code generation starting with version 6.0. 
The generated code mostly uses the same design as the Java version. 
Note that there are still some kinks being worked on - like the NODE_FACTORY option etc.
 
Implemented all JavaCC options for the C++ version as well.

Implemented the following C++ specific options:
  - NODE_INCLUDES - a common include file for all the node classes.

======================================
Changes in 6.0
======================================

Added support for C++ code generation for the tree builder.

Added the new option:
  -OUTPUT_LANGUAGE - string valued options with "java" and "c++" currently allowed.
  -VISITOR_METHOD_NAME_INCLUDES_TYPE_NAME - include the name of the class being visited in the Visitor

======================================
Changes in 4.2
======================================
Release 4.2 is a maintenance release, incorporating a number of bug
fixes and enhancements. For a complete list, please see the
issue tracker:

https://javacc.dev.java.net/issues/buglist.cgi?component=javacc&field0-0-0=target_milestone&type0-0-0=equals&value0-0-0=4.2

======================================
Changes in 4.1
======================================
Release 4.1 is a maintenance release, incorporating a number of bug
fixes and enhancements. For a complete list, please see the
issue tracker:

https://javacc.dev.java.net/issues/buglist.cgi?component=javacc&field0-0-0=target_milestone&type0-0-0=equals&value0-0-0=4.1

======================================
Changes in 4.0
======================================
New option NODE_EXTENDS to specify a classname that
SimpleNode extends so that boiler plate code can be  put
in that class. See the examples/Interpreter for an example
usage.

======================================
Changes from JJTree 0.3pre5 to 0.3pre6
======================================

----------------------------------------------------------------

Fixed bug where Writers were not being closed correctly.

----------------------------------------------------------------

======================================
Changes from JJTree 0.3pre4 to 0.3pre5
======================================

----------------------------------------------------------------

Fixed a bug where a node annotation #P() caused a null pointer error.

----------------------------------------------------------------

Only generate the jjtCreate() methods if the NODE_FACTORY option is
set.

----------------------------------------------------------------

Fixed a bug where the name of the JJTree state file was being used in
the declaration of the field.

----------------------------------------------------------------

======================================
Changes from JJTree 0.3pre3 to 0.3pre4
======================================

----------------------------------------------------------------

Made the constructors of nodes public.  Also made the node identifier
constants and the associated strings public.

----------------------------------------------------------------

Fixed a misleading error message that was produced when the output
file couldn't be generated for some reason.

----------------------------------------------------------------

Brought the HTML documentation up to date.

----------------------------------------------------------------

Fixed a bug where the file containing the JJTree state class was
ignoring the OUTPUT_DIRECTORY option.

----------------------------------------------------------------

Fixed a bug where a construction like this:

	a=foo() #Foo

was being incorrectly handled and generating bad Java code.

----------------------------------------------------------------

Changed the visitor support from a void function to one which takes a
parameter and returns a result.  This is a non-compatible change, so
you will have to update your code if it uses the visitor support.


[Also, if the string option VISITOR_EXCEPTION is set, its value is
also used in the signatures of the various methods in the pattern.

Please note: this exception support is very provisional and will be
replaced in a following version by a more general solution.  It's only
here because I needed it for one of my own projects and thought it
might be useful to someone else too.  Don't use it unless you're
prepared to change your code again later.  Rob.]

----------------------------------------------------------------

======================================
Changes from JJTree 0.3pre2 to 0.3pre3
======================================

----------------------------------------------------------------

JJTree now uses the same grammar as JavaCC.  This means that Java code
in parser actions is parsed.

----------------------------------------------------------------

Added support for the Visitor design pattern.  If the VISITOR option
is true, JJTree adds an accept method to the node classes that it
generates, and also generates a visitor interface.  This interface is
regenerated every time that JJTree is run, so that new nodes will
cause compilation errors in concrete visitors that have not been
updated for them.

----------------------------------------------------------------

Added a couple of examples to illustrate the Visitor support.
JJTreeExamples/eg4.jjt is yet another version of the expression tree
builder which uses a visitor to dump the expression tree; and
VTransformer is a variation of the Java source code transformer.

VTransformer is also possibly directly useful as a tool that inserts
visitor accept methods into class files that were generated with
earlier versions of JJTree.

----------------------------------------------------------------

Added the BUILD_NODE_FILES option, with a default value of true.  If
set to false, it prevents JJTree from generating SimpleNode.java and
nodes that are usually built in MULTI mode.  Node.java is still
generated, as are the various tree constants, etc.

----------------------------------------------------------------

Code that is inserted into the grammar is now enclosed in the standard
@bgen/@egen pair.

----------------------------------------------------------------

The JJTree state object is now generated into its own file if it
doesn't already exist.  This is to make it easier to modify.

----------------------------------------------------------------

Fixed a couple of bugs in the HTML example grammar where the closing
tags didn't match the opening tags.

----------------------------------------------------------------

Fixed a bug where JJTree was trying to clear the node scope while
handling an exception, even when the node had been closed successfully.

----------------------------------------------------------------

NODE_FACTORY no longer implies NODE_USES_PARSER.

If you have been using NODE_FACTORY, then now you'll also need
NODE_USES_PARSER.  Unless, of course, you were never using the parser
in your node factories.

----------------------------------------------------------------

Removed not very useful debugging stuff from the JJTree state object.
It was causing problems with parsers running in security environments
where access to arbitrary properties is disallowed.

----------------------------------------------------------------

======================================
Changes from JJTree 0.3pre1 to 0.3pre2
======================================

----------------------------------------------------------------

The state that JJTree inserts into the parser class is now guarded by
the formal comments.

----------------------------------------------------------------

The JJTree syntax has been changed so that the node descriptor now
comes after the throws clause, rather than before it.

----------------------------------------------------------------

Fixed a bug where string-valued options did not have their quotes
stripped.

----------------------------------------------------------------

Fixed a bug where nodes were being closed early for actions within
ZeroOrMore etc., expansion units.

----------------------------------------------------------------

The special identifier `jjtThis' was not being translated in parameter
lists or in the BNF declaration section.  Fixed it.

----------------------------------------------------------------

Added the OUTPUT_DIRECTORY option.  The default value is "".

----------------------------------------------------------------

Reinstated node factory methods.  They are enabled by setting the
NODE_FACTORY option to true.  Unlike the original node factory methods
they take two arguments: the node identifier constant and a reference
to the parser.  The reference is null for static parsers.

----------------------------------------------------------------

Added the NODE_USES_PARSER option with a default value of false.  When
set to true, JJTree will call the node constructor with a reference to
the parser object as an extra parameter.  This reference is null for
static parsers.

----------------------------------------------------------------

====================================
Changes from JJTree 0.2.6 to 0.3pre1
====================================

----------------------------------------------------------------

JJTree 0.3pre1 has been bootstrapped with JJTree 0.2.6.

Some aspects of JJTree 0.3pre1 are not backwards-compatible with
0.2.6.  Some users will need to modify their parsers in
straightforward ways in order to work with the new version.  See the
file JJTREE-FIX to find out if you need to change anything, and for
detailed instructions on what to change.

----------------------------------------------------------------

JJTree works with the JavaCC exception handling code.  Any unhandled
exceptions within a node scope are caught, the node stack is cleaned
up a bit, and then the exception is rethrown.

----------------------------------------------------------------

Indefinite and Greater-Than nodes have been generalized into
conditional nodes.

Definite nodes now take any integer expression to indicate now many
children they take.  Conditional nodes take any boolean expression to
indicate whether the node is closed and pushed on to the node stack.

----------------------------------------------------------------

The life cycle of a node is now defined.

----------------------------------------------------------------

User-defined parser methods can be called when a node scope is entered
and exited.

----------------------------------------------------------------

The NODE_STACK_SIZE and CHECK_DEFINITE_NODE options are now ignored.

The NODE_SCOPE_HOOK option has been added.  This boolean option
determines whether calls to certain user-defined parser methods are
generated at the beginning and end of each node scope.  The default
value for this option is false.

----------------------------------------------------------------

The special identifier jjtThis can now be used in the declarations
section of a production, as well as on the left hand side of call to a
nonterminal.

----------------------------------------------------------------

A new method arity() has been added to the JJTree state.  It returns
the number of nodes that have been pushed so far in the current node
scope.

----------------------------------------------------------------

The Node interface has changed.  The method jjtAddChild() now takes an
argument to indicate the index of the child it is adding.

----------------------------------------------------------------

The node factory methods are no longer used.  You can remove all your
jjtCreate() methods.

The node constructor now takes an int parameter instead of a String.
You will have to modify your node implementations to use the new
signature.  The protected field `identifier' no longer exists: you can
use the jjtNodeName[] array to map from the new parameter to the old
String.

----------------------------------------------------------------

The implementation of SimpleNode has changed significantly.  It now
uses an array to hold any child nodes instead of a Vector.  It no
longer implements a node factory, its constructor takes an int instead
of a String, and it uses the jjtNodeName[] mechanism for dumping.  The
setInfo() and getInfo() methods have been removed.

----------------------------------------------------------------

The implementation of the state that JJTree keeps in the parser has
changed.  It is now considerably more lightweight: the auxiliary
classes JJTreeNodeStack and JJTreeNodeStackEnum have been abolished.

----------------------------------------------------------------

The JJTree state method currentNode() has been removed.  Any calls to
the method in an action within a node scope are automatically replaced
by references to the special identifier jjtThis.

----------------------------------------------------------------

==================================
Changes from JJTree 0.2.5 to 0.2.6
==================================

----------------------------------------------------------------

Made appropriate internal modifications for the JavaCC 0.7 parse error
exceptions.

----------------------------------------------------------------

==================================
Changes from JJTree 0.2.4 to 0.2.5
==================================

----------------------------------------------------------------

Fixed a bug where the current node was not being updated in the right
place for final user actions.

----------------------------------------------------------------

==================================
Changes from JJTree 0.2.3 to 0.2.4
==================================

----------------------------------------------------------------

Fixed a bug where bad code was generated for void nodes in MULTI mode.

----------------------------------------------------------------

Fixed a bug where a node decoration directly on an action generated
bad code. For example,

	{} #MyNode

----------------------------------------------------------------

==================================
Changes from JJTree 0.2.2 to 0.2.3
==================================

----------------------------------------------------------------

Added toString() and toString(String) methods to SimpleNode and
modified the dumping code to use them.  Now you can easily customize
how a node appears in the tree dump, without having to reproduce the
tree walking machinery.  See SimpleNode.java for details.

----------------------------------------------------------------

Clarified the concept of node scope.  currentNode() now refers to the
node currently being built for the current scope.  It used to be
incorrectly implemented as referring to the most recently created
node, and was synonymous with peekNode().

This change may break some existing programs.  Those programs should
be changed to use peekNode() where they currently use currentNode().

Added jjtThis to every user action.  It refers to the same node that
currentNode() does, but is already cast to the appropriate node type.

The final user action in a node scope is different from all the
others.  When it is executed the node has been fully created, had its
children added, and has been pushed on the node stack.  By contrast,
other user actions within the scope are called when the children are
still on the stack, and the current node is not.

Added the nodeCreated() method so that final actions within greater
than nodes can tell whether the node was created or not.

----------------------------------------------------------------

Fixed several stupid bugs in the Macintosh main class.

----------------------------------------------------------------

Fixed names of internally used JJTree classes so that they use the JJT
prefix and the name of the parser.  This is to avoid naming conflicts
where there is more than one JJTree parser in the same package.

----------------------------------------------------------------

================================
Changes from JJTree 0.2 to 0.2.2
================================

The main change between Beanstalk 0.2 and JJTree 0.2.2 is the removal
of the factory classes.  Their function is now performed by a static
method in the node classes themselves.

The state maintained in the parser class has been changed from bs to
jjtree.  The prefix on the Node class methods has been changed from bs
to jjt.

A new node method jjtGetNumChildren() returns the number of children
the node has.  You can use this in conjunction with jjtGetChild() to
iterate over the children.

Two new options have been added: OUTPUT_FILE and NODE_DEFAULT_VOID.

----------------------------------------------------------------