Known issues:
	- MIDI support is still in pre-alpha state so it's not included when compiling fJACK.
	  You can include it by uncommenting the INCLUDE fjack/midi.frt line in fjack/main.frt

1.01	2010-05-13 17:22:53
	- There were some jack-state related bugs with transport mode and client zombie state - probably fixed after
	  lots of testing
	- the finite impulse optimized code for vfx is fixed
	- fir filters for iforth interrupt safety ...
	- importing fftw3 wisom works now

1.00	2010-03-15 13:48:06
	- a lot of avarage -> average typos have been corrected!
	- after a lot of work and bug fixing this is the first major release.

0.130	2010-03-03 21:49:42
	- check for correct modes and channels to make playing wav files safer

0.129	2010-02-21 03:02:49
	- added REPOSITION-PLAY-FILE	( position relative-flag -- )
	- added PLAY-FILE-POSITION	( -- position|-1 )
	- BUGFIX: fixed problem with forgetting plans without also forgetting fjack

0.128	2010-02-06 01:10:58
	- Worked on forgetting-fJACK, the ANS MARKER -fjack is supported now on both systems.
	  *** STILL TO FIX *** in vfx you can only remove the complete fjack system
	- Also AtExit is an easier go now
	- Memory leak in object creation fixed; also uses jack strings
	- better testing because all throws after allocation and file handling are done via ?FILE ?ALLOCATE
	- some R/W OPEN-FILE are R/O where possible

0.127	2010-02-04 09:45:47
	- tests in plan freeing - first step to really forgetting plans
	- corrected windows dynlibs
	- more precise error messages with starting jack demons

0.126	2010-02-01 23:41:11
	- improved overall stability in iForth environment with forgetting, probably the fftplans will
	  be created/forgettable objects in one of the next releases and it will be possible to use
	  less parts of fjack
		- the plugin objects in applications were not looked after while forgetting, this
		  prevented safe removal like -gui/grafdemo because the objects were still in old dictionary
		  space and were overwritten later with 'painfull' data.
		- #FJACK-SAMPLE correct when no plugins are installed
		- snippets have FORGET> field for iFORTH now
		- seperated wav handling code in START/STOP-JACK-CLIENT
	- added GRAPH-REORDER-HOOK and the reordering callback
	- added PORT-CONNECT-HOOK  and the port connect callback

0.125	2010-01-21 19:03:05
	- frequencies are floats now on request for higher precision
	- added SET-OBJ-FLOAT-FREQUENCY

0.124	2009-12-30 13:23:26
	- phaser effect plugin added to examples
	- added adaptive filter freezing
	- less dummy words for doc generation
	- threadsafe dummyvars for vfx fir sse2 kernels - might be necessary for future jack releases
	- setting plug_xt_quality plug_xt_tune are both set to FALSE by default.
	- the taps/size of a filter is defined in client as plug_size
	- INIT-PLUGINS in JACKAUDIO, .JACK-EXTENDED included
	- Two changes in the user interface for the latest effects!
		SET-PLUGIN-BYPASS instead of switching
		added EFFECTS-FIR; the effects filter is changed in the user interface

0.123	2009-12-26 14:38:50
	- slightly faster adp code due to marcels asm32 fixes
	- added alpha STREAM-EXTERN-FRAMES in client

0.122	2009-12-22 21:47:46
	- BUGFIX: i didn't observe the optimized adp filter had a problem in (calc-adp-muls-sse2) since 0.116
	- As the vfx assembler doesn't include sse2 instructions the assembler-optimized
	  filter routines and the cpu-feature detection are compiled in iforth and saved/loaded to
	  relocatable binary code files. In optv32.frt the CODE words are defined as dummies
	  and the compiled binaries are copied to there.
	  In fjack.frt a definition of save-sse2-overlay? will switch to compile-binaries mode
	- The assembler is only included when the overlays are generated.
	- $CONNECTIONS replaced by $FJACKCONNECIONS and $FJACKSYSFILE to seperate between
	  system files like binaries and the connections.
	- client info slightly changed
	- OPTIMIZED-FILTER-CODE? is a config switch to allow optimized filter kernels - debugging aid
	- even more debugging info for fir filter generation
	- phase delay support started

0.121	2009-12-14 23:40:16
	- added object checking in FIR and adaptive filters
	- FIR filters also have an equalizer type now; uses the new plug_static element
	  the equalizers have functions to control a frequency-point or range of level-setting
	  and are fully supported by the showfft plugin for iforth to change settings interactively.
	- the fft modules allows shifted access in the history now for proper pre&after filter display and analysis
	- port creating checks implemented
	- for debugging sessions there is more memory information available via the switch EXTENDED-JACK-INFO

0.120	2009-12-09 22:29:52
	- a little version jump. The redesign of the processing chain
	  was done after a discussion with Elliott Chapin that brought some new ideas and made
	  me go much deeper in fjack internals that i thought. BUT these changes seem to be
	  rather important for future develpments. They offer lots of effects with much easier
	  plugin design, first examples what can be done are the mixer and vibrato plugins.

	- the processing chain order has changed a lot, every plugin object a private history,
	  the plugins are 'connected' now via input->output history pairs, no hard wiring
	  via the plugin-order any more. Se details in the manuals and graphics.
          This chaining needs 'a lot' of memory, each history buffer now has $20000 SFLOATS
	  to allow more than 2 seconds of delay at 48000/sec sample rate, so for 32 #PLUGINS
	  you will use 16MB.
	  history-keeping needs more stores&fetches and first tests showed decreased performance.
	  BUT the reordering gives slightly better performance except on heavy-loaded systems because
	  of better cache lining. Tests confirmed this on dualcore intels, pentium M and northwood p4
	  machines. So no performance penalty in most situations with much greater flexibility.

	- other Changes:
		- PRE/POST-IO-HOOKS are not there any more, not used so far anyway.
		- the fft plugin uses history
		- options in fftw3lib for creating plans
		- the fir2 and adaptive filters have the address corrections for the data in forth to
		  allow more generic usage in the sse2 code.

	- News & additions:
		- a better leveler plugin
		- a primitive vibrato plugin that uses the interpolated history access
		- config value AUTO-HISTORY-CLEAR?
		- mixer plugin is redesigned and moved to the main fjack system because of importance, .MIXER-INFO
		- a generic effects filter, this waits for real world applications like echo & more

		- the public #FJACK-SAMPLE for history access and your plugins so you don't need plugin counters.
		  This is used in all plugin modules instead of private counters
		- @HISTORY-DATA @INTERPOLATED-HISTORY-DATA MOVE-HISTORY-DATA OBJECT-HISTORY @IDX-HISTORY-DATA SLOT/CHANNEL-HISTORY
		  to support plugin objects development
		- RECONNECT-PLUGIN added.
		- SWITCH-PLUGIN-BYPASS added. Not used so far but possibly good for applications with lots
		  of interactive changing plugins. jackdemo and radio/swlapp will use this in the near future
		  for even less audio dropping.
	- bugfixes
		- corrected fftw3.so.3 naming
		- missing + DF! in CALC-FIR-FILTER added

0.115-0.119 internal checkouts

0.114	2009-11-18 19:46:35 ( deprecated, functionality replaced in 0.115 )
	- changed the mixproc user interface for seperate mixer-level and action setting
	- added graphs for mixer
	- added synchronized IO processing
	- the client has direct access to audio input data, intended to be used for tapped delays

0.113	2009-11-13 12:56:24
	- added the mixproc plugin
	- made STOP-JACK-TRANSPORT safer, it waits until it has really stopped

0.112	2009-10-17 21:43:00
	- integrated MIDI ports and MIDI library calls have been tested to begin MIDI development
	- MIDI-HOOK redefined; the PROCESS-CALLBACK presents MIDI events
	  MIDI support is not in a stable state yet, hoping for help/suggestions

0.111	2009-08-26 01:38:31
	- decoder has a 3pt moving avarage in front
	- logo sound files for starting/stopping the client are added
	- the FIR modules is gone into one source checking for the library
	- error - probable a "don't care" in $CONNECTIONS fixed
	- more work on safer forgetting
	- a generic FIR filter is implemented that allows setting the coefficients
	  calculating function.
	- the tests and rarely used plugins or of mediocre quality are moved to examples
	- The filter kernels allow all filter sizes between 3 and maximum
	- filter kernels isolated and sse2 code available for 64/32 bit iforth versions
	- This is the first release with automatic switching to sse2 code when a cpu is detected
	  On my 64bit system the jack-load in jack-benchmark dropped from 33% to ~28% just by the
	  better forth-filter algorithm, with sse2 it's about 15%.
	  The 32bit version also dropped from ~30%, 26%, 18%
	  As vfx forth has no sse2 assembler at the moment, i will have to write some sort of
	  loaded binary code -> delayed

0.110	- intermediate version with first 32bit sse2 code

0.109	2009-08-11 00:40:48
	- fixed problems with iforth forgetting

0.108	2009-08-09 18:31:32
	- the fir and adaptive filters have again been improved they use 4 input data buffers
	  for 16b aligned access, also 32bit sse2 support has started.
	- OBJ-BENCHMARK has been added for measuring performance of filter objects, it shows
	  larger performance improvements than expected from the jack load measurements.

0.107	2009-08-07 00:06:56
	- Major work on FIR and Adaptive filters has been done, the overall performance for
	  all cpus has been improved
	- There is a new FIR filter module fir2.frt offering these features
		- Calculating the coefficients is done via a reverse FFT so you need the fftw3 library
		  In case you don't have this you may switch to the 'old' module fir.frt
		- BAND-STOP AND BAND-PASS filters have been added in fir2.frt
		- windows function can be changed in fir2.frt
		- filter response can be modified with a user defined function in fir2.frt
		  This means you can wholly redefine the filter response
		- for 64bit machines there are SSE2 assembler optimized filter kernels
	- The real performance gain for fir&adaptive filters is difficult to measure as it's
	  dependent on the system memory load. The larger the load - the better is improvement.
	  On my ~2GHz core2duo the load in jack-benchmark dropped from ~30% to ~20% !
	- added inverse fft for libfftw3 to implement the above
	- frequency2 added to all plugins; so band-stop/pass filters can be displayed and checked

0.106	2009-07-26 15:53:51
	- more precise error messages for client
	- clearing connections checks for a running client
	- default is NOT creating documentation

0.104	2009-06-15 22:12:00
	- fixes on fft & fftw3 plugins with AtCold

0.103	2009-06-03 01:57:11
	- DSP-DVALUE added
	- fftw3 real-complex fft analysis implemented
	- fft analyser plugin
	- dynlibs.frt for iforth implemented as a general extern: interface
		- available as a general module
		- dependencies removed
	- all 'default' floats changed to 64bit floats so no speed dependency for vfx forth
	- all modules and wav files are working with turnkeys now
	- iforth also has the AtCold and AtExit hooks now - dependencies removed
	- prototype file added
	- for iforth there is a gui/showfft module and a little gui/jackdemo application.
	  Both have no vfx port until there is a fast graphical library available.
	- corrected decoder plugin
	- ALL modules and the client don't use JACK-ALLOCATE and the fixed memory limit any more,
	  this is achieved by auto-allocate/deallocate of ALL plugins


0.102	2009-05-16 00:24:27
	- counted string to zero terminated rewritten without allocate to allow use in callbacks
	- rewrote the Library: and Extern: implementations for iforth to allow turnkeys
	- osdeps for iforth using Library: mechanism
	- iforth fix for userpool in turnkeys
	- turnkeys start the client with known hurray
	- verbose switch is a value
	- added library handling for linux version until a stable module exists


0.101	2009-05-13 07:25:45
	- all plugins use the plug_xt_initfilter mechanism to allocate buffers
	- redefined snippets; include linked list to reallocate memory and content for turnkeys
	- added the turnkey.frt defining the turnkey hooks
	- notches work up, works with channels but only 8 filters
	- docs for leveler
	- renamed TEST to PLAY-SNIPPETS
	- FIR and ADAPIVE filters can chance the number of taps with  SET-OBJ-QUALITY
	- explicit connections in START-CLIENT


0.100	2009-05-09 13:37:31
	- first steps to turnkeys; the client struct is now in dictionary memory
		- no *client and clientsize any more
	- all non-permanent data is allocated at boot by INIT-JACK-TEMP-MEM
	- BYE more complete
	- FIXME: have to think about objects memory; auto allocation ?
	- io_procs code revisited, have missed some stupid decisions
	- added link filed for all defined plugin objects and added LIST-ALL-PLUGINS
	- the types had a bug; COMP-FCHANNEL, had to be used
		- slightly faster code with less good optimizers
	- docs for notches and tests added
	- this history.txt added


0.99 first b-release sent


