The following changes (change numbers refer to perforce) were
made from version 3.1.1 to 3.1.2

Runtime
-------

Change 5641 on 2009/02/20 by jimi@jimi.jimi.antlr3

	Release version 3.1.2 of the ANTLR C runtime.
	
	Updated documents and release notes will have to follow later.

Change 5639 on 2009/02/20 by jimi@jimi.jimi.antlr3

	Fixed: ANTLR-356
	
	Ensure that code generation for C++ does not require casts

Change 5577 on 2009/02/12 by jimi@jimi.jimi.antlr3

	C Runtime - Bug fixes.
	
	 o Having moved to use an extract directly from a vector for returning 
	   tokens, it exposed a
	   bug whereby the EOF boudary calculation in tokLT was incorrectly 
	   checking > rather than >=. 
	 o Changing to API initialization of tokens rather than memcmp() 
	   incorrectly forgot to set teh input stream pointer for the 
	   manufactured tokens in the token factory;
	 o Rewrite streams for rewriting tree parsers did not check whether the 
	   rewrite stream was ever assigned before trying to free it, it is now 
	   in line with the ordinary parser code.

Change 5576 on 2009/02/11 by jimi@jimi.jimi.antlr3

	C Runtime: Ensure that when we manufacture a new token for a missing 
	token, that the user suplied custom information (if any) is copied 
	from the current token.

Change 5575 on 2009/02/08 by jimi@jimi.jimi.antlr3

	C Runtime - Vastly improve the reuse of allocated memory for nodes in 
	  tree rewriting.
	
	A problem for all targets at the moment si that the rewrite logic 
	generated by ANTLR makes no attempt
	to reuse any resources, it merely gurantees that the tree shape at the 
	end is correct. To some extent this is mitigated by the garbage 
	collection systems of Java and .Net, even thoguh it is still an overhead to
	keep creating so many modes.
	
	This change implements the first of two C runtime changes that make 
	best efforst to track when a node has become orphaned and will never 
	be reused, based on inherent knowledge of the rewrite logic (which in
	the long term is not a great soloution). 
	
	Much of the rewrite logic consists of creating a niilnode into which 
	child nodes are appended. At: rulePost processing time; when a rewrite 
	stream is closed; and when becomeRoot is called, there are many situations
	where the root of the tree that will be manipulted, or is finished with 
	(in the case of rewrtie streams), where the nilNode was just a temporary 
	creation for the sake of the rewrite itself. 
	
	In these cases we can see that the nilNode would just be left ot rot in 
	the node factory that tracks all the tree nodes.
	Rather than leave these in the factory to rot, we now keep a resuse 
	stck and always reuse any node on this
	stack before claimin a new node from the factory pool.
	
	This single change alone reduces memory usage in the test case (20,604 
	line C program and a GNU C parser) 
	from nearly a GB, to 276MB. This is still way more memory than we 
	shoudl need to do this operation, even on such a large input file,
	but the reduction results in a huge performance increase and greatly 
	reduced system time spent on allocations.
	
	After this optimizatoin, comparison with gcc yeilds:
	
	time gcc -S a.c
	a.c:1026: warning: conflicting types for built-in function ‘vsprintf’
	a.c:1030: warning: conflicting types for built-in function ‘vsnprintf’
	a.c:1041: warning: conflicting types for built-in function ‘vsscanf’
	0.21user 0.01system 0:00.22elapsed 97%CPU (0avgtext+0avgdata 0maxresident)k
	0inputs+240outputs (0major+8345minor)pagefaults 0swaps
	
	and
	
	time ./jimi
	Reading a.c
	0.28user 0.11system 0:00.39elapsed 98%CPU (0avgtext+0avgdata 0maxresident)k
	0inputs+0outputs (0major+66609minor)pagefaults 0swaps
	
	And we can now interpolate the fact that the only major differnce is 
	now the huge disparity in memory allocations. A
	future optimization of vector pooling, to sepate node resue from vector 
	reuse, currently looks promising for further reuse of memory.
	
	Finally, a static analysis of the rewrte code, plus a realtime analysis 
	of the heap at runtime, may well give us a reasonable memory usage 
	pattern. In reality though, it is the generated rewrite logic
	that must becom optional at not continuously rewriting things that it 
	need not, as it ascends the rule chain.

Change 5563 on 2009/01/28 by jimi@jimi.jimi.antlr3

	Allow rewrite streams to use the base adaptors vector factory and not 
	try to malloc new vectors themselves.

Change 5562 on 2009/01/28 by jimi@jimi.jimi.antlr3

	Don't use CALLOC to allocate tree pools, use malloc as there is no need 
	for calloc.

Change 5561 on 2009/01/28 by jimi@jimi.jimi.antlr3

	Prevent warnigsn about retval.stop not being initialized when a rule 
	returns eraly because it is in backtracking mode

Change 5558 on 2009/01/28 by jimi@jimi.jimi.antlr3

	Lots of optimizations (though the next one to be checked in is the huge 
	win) for AST building and vector factories.
	
	A large part of tree rewriting was the creation of vectors to hold AST 
	nodes. Although I had created a vector factory, for some reason I never got 
	around to creating a proper one, that pre-allocated the vectors in chunks and 
	so on. I guess I just forgot to. Hence a big win here is prevention of calling 
	malloc lots and lots of times to create vectors.
	
	A second inprovement was to change teh vector definition such that it 
	holds a certain number of elements wihtin the vector structure itself, rather 
	than malloc and freeing these. Currently this is set to 8, but may increase. 
	For AST construction, this is generally a big win because AST nodes don't often 
	have many individual children unless there has not been any shaping going on in 
	the parser. But if you are not shaping, then you don't really need a tree.
	
	Other perforamnce inprovements here include not calling functions 
	indirectly within token stream and common token stream. Hence tokens are 
	claimed directly from the vectors. Users can override these funcitons of course 
	and all this means is that if you override tokenstreams then you pretty much 
	have to provide all the mehtods, but then I think you woudl have to anyway (and 
	I don't know of anyone that has wanted to do this as you can carry your own 
	structure around with the tokens anyway and that is much easier).

Change 5555 on 2009/01/26 by jimi@jimi.jimi.antlr3

	Fixed: ANTLR-288
	Correct the interpretation of the skip token such that channel, start 
	index, char pos in lie, start line and text are correctly reset to the start of 
	the new token when the one that we just traversed was marked as being skipped. 
	
	This correctly excludes the text that was matched as part of the 
	SKIP()ed token from the next token in the token stream and so has the side 
	effect that asking for $text of a rule no longer includes the text that shuodl 
	be skipped, but DOES include the text of tokens that were merely placed off the 
	default channel.

Change 5551 on 2009/01/25 by jimi@jimi.jimi.antlr3

	Fixed: ANTLR-287
	Most of the source files did not include the BSD license. THis might
	not be that big a deal given that I don't care what people do with it
	other than take my name off it, but having the license reproduced 
	everywhere
	at least makes things perfectly clear. Hence this mass change of 
	sources and templates
	to include the license.

Change 5550 on 2009/01/25 by jimi@jimi.jimi.antlr3

	Fixed: ANTLR-365
	Ensure that as soon as we known about an input stream on the lexer that
	we borrow its string factroy adn use it in our EOF token in case
	anyone tries to make it a string, such as in error messages for 
	instance.

Change 5548 on 2009/01/25 by jimi@jimi.jimi.antlr3

	Fixed: ANTLR-363
        At some point the Java runtime default changed from discarding offchannel
        tokens to preserving them. The fix is to make the C runtime also 
	default to preserving off-channel tokens.

Change 5544 on 2009/01/24 by jimi@jimi.jimi.antlr3

	Fixed: ANTLR-360
	Ensure that the fillBuffer funtiion does not call any methods
	that require the cached buffer size to be recorded before we
	have actually recorded it.

Change 5543 on 2009/01/24 by jimi@jimi.jimi.antlr3

	Fixed: ANTLR-362
	Some users have started using string factories themselves and
	exposed a flaw in the destroy method, that is intended to remove
	a strng htat was created by the factory and is no longer needed.
	The string was correctly removed from the vector that tracks them
	but after the first one, all the remaining strings are then numbered
	incorrectly. Hence the destroy method has been recoded to reindex
	the strings in the factory after one is removed and everythig is once
	more hunky dory.
	User suggested fix rejected.

Change 5542 on 2009/01/24 by jimi@jimi.jimi.antlr3

	Fixed ANTLR-366
	The recognizer state now ensures that all fields are set to NULL upon 
creation
	and the reset does not overwrite the tokenname array

Change 5527 on 2009/01/15 by jimi@jimi.jimi.antlr3

	Add the C runtime for 3.1.2 beta2 to perforce

Change 5526 on 2009/01/15 by jimi@jimi.jimivista.antlr3

	Correctly define the MEMMOVE macro which was inadvertently left to be 
	memcpy.

Change 5503 on 2008/12/12 by jimi@jimi.jimi.antlr3

	Change C runtime release number to 3.1.2 beta

Change 5473 on 2008/12/01 by jimi@jimi.jimivista.antlr3

	Fixed: ANTLR-350 - C runtime use of memcpy
	Prior change to use memcpy instead of memmove in all cases missed the 
	fact that the string factory can be in a situation where overlaps occur. We now 
	have ANTLR3_MEMCPY and ANTLR3_MEMMOVE and use the two appropriately.

Change 5471 on 2008/12/01 by jimi@jimi.jimivista.antlr3

	Fixed ANTLR-361
	 - Ensure that ANTLR3_BOOLEAN is typedef'ed correctly when building for 
	   MingW

Templates
---------

Change 5637 on 2009/02/20 by jimi@jimi.jimi.antlr3

	C rtunime - make sure that ADAPTOR results are cast to the tree type on 
	a rewrite

Change 5620 on 2009/02/18 by jimi@jimi.jimi.antlr3

	Rename/Move:
	From: //depot/code/antlr/main/src/org/antlr/codegen/templates/...
	To: //depot/code/antlr/main/src/main/resources/org/antlr/codegen/templates/...
	
	Relocate the code generating templates to exist in the directory set 
	that maven expects.
	
	When checking in your templates, you may find it easiest to make a copy 
	of what you have, revert the change in perforce, then just check out the 
	template in the new location, and copy the changes back over. Nobody has oore 
	than two files open at the moment.

Change 5578 on 2009/02/12 by jimi@jimi.jimi.antlr3

	Correct the string template escape sequences for generating scope
	code in the C templates.

Change 5577 on 2009/02/12 by jimi@jimi.jimi.antlr3

	C Runtime - Bug fixes.
	
	 o Having moved to use an extract directly from a vector for returning 
	    tokens, it exposed a
	    bug whereby the EOF boudary calculation in tokLT was incorrectly 
	    checking > rather than
	    >=. 
	 o Changing to API initialization of tokens rather than memcmp() 
	    incorrectly forgot to 
	    set teh input stream pointer for the manufactured tokens in the 
	    token factory;
	 o Rewrite streams for rewriting tree parsers did not check whether the 
	    rewrite stream
	    was ever assigned before trying to free it, it is now in line with 
	    the ordinary parser code.

Change 5567 on 2009/01/29 by jimi@jimi.jimi.antlr3

	C Runtime - Further Optimizations
	
	Within grammars that used scopes and were intended to parse large 
	inputs with many rule nests,
	the creation anf deletion of the scopes themselves became significant. 
	Careful analysis shows that
	for most grammars, while a parse could create and delete 20,000 scopes, 
	the maxium depth of
	any scope was only 8. 
	
	This change therefore changes the scope implementation so that it does 
	not free scope memory when
	it is popped but just tracks it in a C runtime stack, eventually 
	freeing it when the stack is freed. This change
	caused the allocation of only 12 scope structures instead of 20,000 for 
	the extreme example case.
	
	This change means that scope users must be carefule (as ever in C) to 
	initializae their scope elements
	correctly as:
	
	1) If not you may inherit values from a prior use of the scope 
	    structure;
	2) SCope structure are now allocated with malloc and not calloc;
	
	Also, when using a custom free function to clean a scope when it is 
	popped, it is probably a good idea
	to set any free'd pointers to NULL (this is generally good C programmig 
	practice in any case)

Change 5566 on 2009/01/29 by jimi@jimi.jimi.antlr3

	Remove redundant BACKTRACK checking so that MSVC9 does not get confused 
	about possibly uninitialized variables

Change 5565 on 2009/01/28 by jimi@jimi.jimi.antlr3

	Use malloc rather than calloc to allocate memory for new scopes. Note 
	that this means users will have to be careful to initialize any values in their 
	scopes that they expect to be 0 or NULL and I must document this.

Change 5564 on 2009/01/28 by jimi@jimi.jimi.antlr3

	Use malloc rather than calloc for copying list lable tokens for 
	rewrites.

Change 5561 on 2009/01/28 by jimi@jimi.jimi.antlr3

	Prevent warnigsn about retval.stop not being initialized when a rule 
	returns eraly because it is in backtracking mode

Change 5560 on 2009/01/28 by jimi@jimi.jimi.antlr3

	Add a NULL check before freeing rewrite streams used in AST rewrites 
	rather than auto-rewrites.
	
	While the NULL check is redundant as the free cannot be called unless 
	it is assigned, Visual Studio C 2008
	gets it wrong and thinks that there is a PATH than can arrive at the 
	free wihtout it being assigned and that is too annoying to ignore.

Change 5559 on 2009/01/28 by jimi@jimi.jimi.antlr3

	C target Tree rewrite optimization
	
	There is only one optimization in this change, but it is a huge one.
	
	The code generation templates were set up so that at the start of a rule,
	any rewrite streams mentioned in the rule wer pre-created. However, this
	is a massive overhead for rules where only one or two of the streams are
	actually used, as we create them then free them without ever using them.
	This was copied from the Java templates basically.
	This caused literally millions of extra calls and vector allocations
	in the case of the GNU C parser given to me for testing with a 20,000 
	line program.
	
	After this change, the following comparison is avaiable against the gcc 
	compiler:
	
	Before (different machines here so use the relative difference for 
	comparison):
	
	gcc:
	
	real    0m0.425s
	user    0m0.384s
	sys     0m0.036s
	
	ANTLR C
	real    0m1.958s
	user    0m1.284s
	sys     0m0.656s
	
	After the previous optimizations for vector pooling via a factory,
	plus this huge win in removing redundant code, we have the following
	(different machine to the one above):
	
	gcc:
	0.21user 0.01system 0:00.23elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k
	0inputs+328outputs (0major+9922minor)pagefaults 0swaps
	
	ANTLR C:
	
	0.37user 0.26system 0:00.64elapsed 99%CPU (0avgtext+0avgdata 0maxresident)k
	0inputs+0outputs (0major+130944minor)pagefaults 0swaps
	
	The extra system time coming from the fact that although the tree 
	rewriting is now optimal in terms of not allocating things it does 
	not need, there is still a lot more overhead in a parser that is generated 
	for generic use, including much more use of structures for tokens and extra 
	copying and so on. I will
	continue to work on improviing things where I can, but the next big 
	improvement will come from Ter's optimization of the actual code structures we 
	generate including not doing things with rewrite streams that we do not need to 
	do at all.
	
	The second machine I used is about twice as fast CPU wise as the system 
	that was used originally by the user that asked about this performance.

Change 5558 on 2009/01/28 by jimi@jimi.jimi.antlr3

	Lots of optimizations (though the next one to be checked in is the huge 
	win) for AST building and vector factories.
	
	A large part of tree rewriting was the creation of vectors to hold AST 
	nodes. Although I had created a vector factory, for some reason I never got 
	around to creating a proper one, that pre-allocated the vectors in chunks and 
	so on. I guess I just forgot to. Hence a big win here is prevention of calling 
	malloc lots and lots of times to create vectors.
	
	A second inprovement was to change teh vector definition such that it 
	holds a certain number of elements wihtin the vector structure itself, rather 
	than malloc and freeing these. Currently this is set to 8, but may increase. 
	For AST construction, this is generally a big win because AST nodes don't often 
	have many individual children unless there has not been any shaping going on in 
	the parser. But if you are not shaping, then you don't really need a tree.
	
	Other perforamnce inprovements here include not calling functions 
	indirectly within token stream and common token stream. Hence tokens are 
	claimed directly from the vectors. Users can override these funcitons of course 
	and all this means is that if you override tokenstreams then you pretty much 
	have to provide all the mehtods, but then I think you woudl have to anyway (and 
	I don't know of anyone that has wanted to do this as you can carry your own 
	structure around with the tokens anyway and that is much easier).

Change 5554 on 2009/01/26 by jimi@jimi.jimi.antlr3

	Fixed: ANTLR-379
	For some reason in the past, the ruleMemozation() template had required 
	that the name parameter be set to the rule name. This does not seem to be a 
	requirement any more. The name=xxx override when invoking the template was 
	causing all the scope names derived when cleaning up in memoization to be 
	called after the rule name, which was not correct. Howver, this only affected 
	the output when in output=AST mode.
	
	This template invocation is now corrected.

Change 5553 on 2009/01/26 by jimi@jimi.jimi.antlr3

	Fixed: ANTLR-330
	Managed to get the one rule that could not see the ASTLabelType to call 
	back in to the super template C.stg and ask it to construct hte name. I am not 
	100% sure that this fixes all cases, but I cannot find any that fail. PLease 
	let me know if you find any exampoles of being unable to default the 
	ASTLabelType option in the C target.

Change 5552 on 2009/01/25 by jimi@jimi.jimi.antlr3

	Progress: ANTLR-327
	Fix debug code generation templates when output=AST such that code
	can at least be generated and I can debug the output code correctly.
	Note that this checkin does not implement the debugging requirements
	for tree generating parsers.

Change 5551 on 2009/01/25 by jimi@jimi.jimi.antlr3

	Fixed: ANTLR-287
	Most of the source files did not include the BSD license. THis might
	not be that big a deal given that I don't care what people do with it
	other than take my name off it, but having the license reproduced 
	everywhere at least makes things perfectly clear. Hence this mass change of 
	sources and templates to include the license.

Change 5549 on 2009/01/25 by jimi@jimi.jimi.antlr3

	Fixed: ANTLR-354
	Using 0.0D as the default initialize value for a double caused
	VS 2003 C compiler to bomb out. There seesm to be no reason other
	than force of habit to set this to 0.0D so I have dropped the D so
	that older compilers do not complain.

Change 5547 on 2009/01/25 by jimi@jimi.jimi.antlr3

	Fixed: ANTLR-282
	All references are now unadorned with any type of NULL check for the 
	following reasons:
	
		1) A NULL reference means that there is a problem with the 
		   grammar and we need the program to fail immediately so 
		   that the programmer can work out where the problem occured;
		2) Most of the time, the only sensible value that can be 
		   returned is NULL or 0 which
		   obviates the NULL check in the first place;
		3) If we replace a NULL reference with some value such as 0, 
		   then the program may blithely continue but just do something 
		   logically wrong, which will be very difficult for the 
		   grammar programmer to detect and correct.

Change 5545 on 2009/01/24 by jimi@jimi.jimi.antlr3

	Fixed: ANTLR-357
	The bug report was correct in that the types of references to things
	like $start were being incorrectly cast as they wer not changed from 
	Java style casts (and the casts are unneccessary). this is now fixed 
	and references are referencing the correct, uncast, types.
	However, the bug report was wrong in that the reference in the bok to 
	$start.pos will only work for Java and really, it is incorrect in the 
	book because it shoudl not access the .pos member directly but shudl 
	be using $start.getCharPositionInLine().
	Because there is no access qualification in C, one could use 
	$start.charPosition, however
	really this should be $start->getCharPositionInLine($start);

Change 5541 on 2009/01/24 by jimi@jimi.jimi.antlr3

	Fixed - ANTLR-367
	The code generation for the free method of a recognizer was not 
	distinguishing tree parsers from parsers when it came to calling delegate free 
	functions.
	This is now corrected.

Change 5540 on 2009/01/24 by jimi@jimi.jimi.antlr3

	Fixed ANTLR-355
	Ensure that we do not attempt to free any memory that we did not
	actually allocate because the parser rule was being executed in
	backtracking mode.

Change 5539 on 2009/01/24 by jimi@jimi.jimivista.antlr3

	Fixed: ANTLR-355
	When a C targetted parser is producing in backtracking mode, then the 
	creation of new stream rewrite structures shoudl not happen if the rule is 
	currently backtracking

Change 5502 on 2008/12/11 by jimi@jimi.jimi.antlr3

	Fixed: ANTLR-349 Ensure that all marker labels in the lexer are 64 bit 
	compatible

Change 5473 on 2008/12/01 by jimi@jimi.jimivista.antlr3

	Fixed: ANTLR-350 - C runtime use of memcpy
	Prior change to use memcpy instead of memmove in all cases missed the 
	fact that the string factory can be in a situation where overlaps occur. We now 
	have ANTLR3_MEMCPY and ANTLR3_MEMMOVE and use the two appropriately.

Change 5387 on 2008/11/05 by parrt@parrt.spork

	Fixed x+=. issue with tree grammars; added unit test

Change 5325 on 2008/10/23 by parrt@parrt.spork

	We were all ref'ing backtracking==0 hardcoded instead checking the 
	@synpredgate action.