1 terminatorX - ChangeLog
2 All changes were done by me (Alexander König) if not stated otherwise.
3 This is not a GNU-style ChangeLog but you sort of get the idea what was
7 - Fixed a bug in XML serialization that dropped strings' contents.
8 - You can now hit "F11" to turn fullscreen mode on off - this will help using
9 terminatorX on small DJ laptops. Note that the key will have no effect in
10 scratch - hit it before you get that engine running.
11 - terminatorX now stores set files as XML documents, too - so terminatorX has
12 become terminatorXML finally. Note: for sets with a large number of events
13 recorded into the sequencer - the XML based set files can grow significally
14 larger than the old ones.
15 - The old code to support the obsolete binary tX file formats has now been
16 scheduled for deletion. It's been moved into the tX_legacy* source files and
17 it will be built only when "--enable-legacy" was used on configuring
18 terminatorX (which is the default). Expect this to become disabled and then
19 totally removed from the sources in future.
20 - Enhanced audio data display - you can now zoom into the audio signal, which
21 should allow much more precise scratching with larger audio files (ie
22 complete tracks). On playback terminatorX will scroll the displayed area
23 to ensure you always see the area of the sample in which you're currently
25 - I have not backported the new code to the old gtk+ 1.2 - so this release no
26 longer builds with gtk+ 1.2 - you have to upgrade to gtk+ Version 2 to be
27 able to build terminatorX.
28 - fixed a gcc 3.2 compilation bug - thanks to
29 Adam Duck <duck@informatik.uni-frankfurt.de>
32 - Fixed a compilation bug ocurring on big endian systems only.
33 - Fixed the built-in wav loading routines for big endian systems (again).
34 - Finally traced down and fixed a long outstanding bug in the dial widget,
35 the background of the dials is now drawn correctly.
36 - Ported terminatorX to gtk+ Version 2. The code for gtk2 will be enabled
37 when gtk2 is detected by configure. To force terminatorX to use the good
38 old gtk 1.2.x (x>=5) run configure with the '--disable-gtk2' switch.
39 WARNING: gdkxft-users: gtk2 applications segfault when preloading
40 libgdk_xft.so! So unset LD_PRELOAD and set GDK_USE_XFT=1 instead.
41 - Applied a patch from Adrian that brings lots of new stuff:
42 - Debian package support (removed - Mike Furr now maintains an official
43 Debain package for terminatorX)
44 - enhanced GNOME integration
45 - a manpage for terminatorX
46 - Included another gcc3 fix thanks to Gregor Riepl <seto-kun@freesurf.ch>
47 - Improved configure's compatibilty - it should detect pthreads more reliably
48 and it should detect ladspa.h when located in /usr/local/include without
49 /usr/local/include being contained in the include path.
50 - This release should compile and run on FreeBSD systems and possibly
51 OpenBSD, too. Thanks for help and testing this goes to
52 Tilman Linneweh <tilman@arved.de>.
53 - Sequencer now refuses to play when no events available - and displays
54 a dialog that suggests running the audio engine.
55 - De-uglified the terminatorX note dialog.
56 - Fixed gtk+2 signals for load file dialog
57 - Pre-listen support for mpg321
58 - Added a new auto-hiding "panel-bar" to which turntable control and audio
59 panels can be minimized to.
60 - Cleaned up the turntables' audio panel for better looks and easier usage
61 - Rewrote the engine thread completely - unlike before the engine thread
62 is created on startup and kept alive until termination.
63 - terminatorX can now be run suid root - and it might even be safe. After
64 creating the engine thread both terminatorX threads drop root privileges
65 before evaluating any parameters or reading/writing files, etc. The problem
66 is the pthread "manager thread" is out of control of terminatorX code - so
67 I cannot tell that thread to drop privileges.
68 - Fixed Drag'n'Drop for GNOME 2
69 - To enhance the loading of audiofiles terminatorX now contains code to use
70 mad (MPEG Audio Decoder), libvorbis and libaudiofile directly instead of
71 pipeing audio data through mpg123, ogg123 or sox. As a result the new routines
72 bring following advantages:
73 - tX loads audiofiles significantly faster
74 - tX now honours the samplerate of an audiofile
75 (by adjusting the pitch of the turntable)
76 - tX provides much better feedback on the loading progress
77 Although the old loading-methods are still available (and piped loading has
78 been improved, too) I highly recommend installing the libraries above as the
79 new routines simply work better.
80 - Deleted tons of old outdated code in tX_engine.cc and tX_mastergui.cc
83 - Applied a patch from Matthew Evans <activesx@hotmail.com> that introduces
84 support for Ogg Vorbis audiofiles. It requires ogg123 > RC2 and sox to work.
85 Autodetection is supported via configure (--enable-ogg123).
86 - Experimental support for input devices other than the core-pointer
87 under XFree86 > 4. This uses the xsetpointer executable that comes with
88 XFree86. configure should autodetect this correctly, in case it doesn't
89 use the --enable-xetpointer switch.
90 - Adrian sent me a patch introducing two new buttons: mute and solo. As it
91 didn't fully harmonize with the sequencer I rewrote parts of it and added new
92 functionality to the turntables: They can now be switched to mix_solo or
93 mix_mute - these settings affect current mixing only - their values will
94 neither be saved nor will changes to those control be recorded in the
95 sequencer - if you want to record muting events either use the mute button
96 from the turntable's trigger menu, the keyboard short-cuts (alt/strg), or
97 your middle mouse button.
98 - fixed a bug that produced corrupt wavfiles on big endian machines
99 - some minor fixes to make terminatorX compile with newer gcc releases
100 - Well, I am allowed to change my mind, aren't I? Anyway, 3DNow! is back.
101 Nevertheless newer gcc releases seem to have a bug in handling movq and
102 the current 3DNow! stereo implementation does not seem to be very efficient.
103 So use --enable-3dnow only if you want to help me improve perfomance.
104 - Modified the mpg123 loader in order to make it work with mpg321, too.
105 mpg321 lacks support for mpg123's -m switch ("singlemix") therefor
106 mpg?2?'s output is now piped into sox to convert the signal to mono.
107 For this reason mpg?2? support now requires sox to be installed, too.
108 - added a workaround for a gtk+ bug that causes empty labels when strings
109 contain umlauts - unluckily my name features one of these so I had to write
113 - applied a big patch from Adrian Reber (adrian@lisas.de). This patch includes:
114 o compile fixes for the newer gcc releases
115 o enhancements to the commandline switches (try --help for more details)
116 o bug fixes for tX_global.c
117 - the bigNewThang (TM) is STEREO output. The Pan-Buttons now actually do what
118 you'd expect them to. Therefore your soundcard must have stereo support of
119 course but I guess in 2001 this really shouldn't be a problem, right?
120 The recorded wavfiles will now be stereo wavfiles of course.
121 Note that stereo output means double audio-bandwith, so if you tuned your
122 old terminatorX setup to smallest latency you might get clicks now. You can
123 easily fix this by increasing the buffer size in the options dialog by one -
124 this will result in the same latency as before btw.
125 - The echo effect has undergone some enhancements to profit from the new stereo
126 output. It's signal will now be mixed seperately so you can choose a different
127 stereo panorama position for the echo signal. So the new echo effect now
128 features it's own PAN-knob. This does have some sideeffects though, for more
129 details on this check the manual.
130 Additionally the echo effect now has it's own volume setting to influence
131 the dry/wet ratio of this effect.
132 - Some LADSPA effects tend to create NAN (Not A Number) float values under
133 certain conditions. This is bad as all following operations will result in
134 NANs, so there will be no audio output no more. Now I will not introduce code
135 to check for NANs as this is a waste of CPU-power, producing NANs is IMHO
136 a faulty behaviour for LADSPA plugins and it should be fixed in there.
137 Now a problem with the previous release was that the echo and lowpass effects
138 "ate" those NANs resulting in NANs in those effects' buffers. So once a NAN
139 got in there these effects caused the whole audio output to mute (forever).
140 Which is why I now introduced some code to clear the buffers in echo and
141 lowpass whenever you dis- and reenable them.
142 - The GUI does have two master vu-meters now. The turntable's still have one
143 vu-meter only. Note that the echo-signal is *not* included in the turntable's
144 vu-meter now, as it's mixed seperately.
145 - The 3DNow!-Optimizations are gone. And they'll proabably never come back as
146 I don't have time to maintain the same algorithm twice.
147 - The documentation has been updated to meet the new features. A lot of typos
148 have been fixed as well.
151 - this list may be incomplete
152 - This release brings support for LADSPA plugins. These plugins can be inserted
153 into any turntable's effect chain at any position. You can edit the the
154 effect sequence with the two tiny blue arrows. The builtin effects reside in
155 the same effect chain as LADSPA plugins, but unlike LADSPA plugins they cannot
156 be removed (nor multi-instanciated).
157 The LADSPA plugins have to "fit" into the effect chain, which is why
158 terminatorX supports *mono* plugins (plugins that have one audio-in and one
159 audio-out port) only. For the plugin's control ports, termintatorX supports
160 all HINTS except for LADSPA_HINT_LOGARITHMIC. Some control ports might be
161 hard to handle as LAPDSA control ports can have very wide value ranges. It is
162 possible to enter precise values though.
163 Get LADSPA and the CMT plugins from http://www.ladspa.org
164 All parameters for LADSPA-plugins are "sequencable parameters", so you can
165 record parameter-change events for the plugin's knobs just as for the
167 - GUI re-design to support dynamic plugin GUIs. The audiofile-display is now
168 seperated from the turntable's controls. The groups are split in to panels
169 which can be minimized if not required (orange arrow). The panels'
170 minimized-status will be saved within the set files.
171 To add a plugin effect into the turntables effect chain click on the "FX"-
172 button and choose the plugin of your choice. To get more info on the
173 plugin effects click on their panel's label.
174 NOTE: The panning-control is not working yet. It'll work as soon as stereo
176 - As always when introducing something new, the set-file file-format has
178 - The parameter-to-mouse-axis mapping has been moved from hardcoded to
179 dynamic. You can now map any parameter that's considered "mappable" to the
180 mouse-axis of your choice (for each turntable separately). This is true
181 for the control ports of LADSPA-plugins, too.
182 - Josh Steiner <joschi@eds.org> has joined terminatorX development. For this
183 release he has added some startup parameters for terminatorX, some of them
184 aiming for future MIDI support, and therefor not functional yet. Run
185 "terminatorX --help" to get more details.
186 - I know write the "manual" in SGML with the DocBook DTD. The distribution
187 contains the HTML and the postscript version. These transformation have been
188 created using James Clark's jade (DSSSL-processor) and Norman Walsh's DSSSL
189 stylesheets for DocBook.
190 - Removed obsolete "number of vtts"-option from the options dialog.
191 - Applied a patch from Josh that fixes running the soundfile editor for files
192 including a blank in their filename.
193 - New tooltips for the complete GUI.
196 - applied a tiny patch from Atsushi Yamagata <yamagata@plathome.co.jp>
197 to set locale on startup.
200 - terminatorX has a sequencer now. Warning: with the current design the
201 sequencer will correctly play back recorded events only when the buffer size
202 (of the audio device) is set to the same size as it was when recording the
203 events. I know this is annoying and I will try to find a solution to this.
204 Please take a look at the README to find out how the sequencer works.
205 - fixed a bug causing the lowpass filter produce a constant signal
206 - fixed the load file / drag'n'drop segfault bug
207 - fixed the prelisten zombie processes bug
208 - fixed the segfault on termination bug (with the audio engine running)
209 - hacked the mixer code to use 3DNow! too. This improves 3DNow! performance gain
210 from previously ~6% to more than 20% (depending on compiler optimizations)
211 - extended the set-file-format to allow storing of sequencer events. This
212 release allows loading old tX-sets too.
213 - some gui icons. Choose Text/Icon / Icon only / Text only display of these
214 icons in the Options dialog. You'll have to restart tX to see the changes
216 - "soundfile-editor support": Enter the soundfile-editor of your choice in the
217 tX-Options dialog (Still looking for an editor? Checkout Dave's Linux
218 Sound and MIDI Pages: http://www.bright.net/~dlphilp/linuxsound/).
219 Click the edit button in the Main-Panel to start your editor with the
220 turntable's soundfile - edit ;) - save and click the reload button.
221 - big endian bug fixes by Scott C. Knight <scknight@indiana.edu>
222 - --enable-benchmark will now produce a benchmark binary. This takes a set file
223 as a parameter (that set file should autotrigger all tables, else the
224 benchmark will measure how fast tX can generate zeros ;) WARNING: the
225 benchmark binary is not a usable terminatorX binary.
226 - a new dial-widget. I created this for the upcoming GUI-redesign - and decided
227 I let you debug it right now ;) If you don't want it run configure with the
228 --disable-dial switch.
229 - some gui redesign. The volume scales are now located near their corresponding
230 "flash"-widget. This release fits into very small windows - so it should be
231 usable on smaller displays as well.
232 - Adrian told me configure doesn´t have to check for glib after checking for
233 gtk+ - I guess he´s right so the glib-test has gone.
237 - implemented "sequenceable parameters" (see tX_seqpar.*). This means the
238 groundwork for the sequencer is done. Although this release does not yet
239 feature a sequencer - there's still a profit: you can now see the GUI (e.g.
240 the lowpass-freq slider) update while you change that value with your mouse in
241 dga grab mode (e.g. up/down with the default X/Y-Control settings).
242 - fixed (libmmx) mmx.h to work with gcc 2.95 and above
243 - fixed compilation problems for big endian machines. still runtime probs
245 - fixed configure-script as the new bash complains about == in string compares
246 - fixed some gui segfaults
247 - added an ugly GNOME-mimetype installer. This installs the ".tX"-mime type, an
248 terminatorX-application icon and a ".tX"-icon. I could use some hints on how
249 I could make automake install that instead of an extra script.
251 - added Drag'n Drop support. Works with the great gmc. Drop an audio file over
252 the filename-button or the wavdisplay of a turntable to load it into that
253 table. Drop an audiofile over the "Add Turntable" Button to create a new
254 turntable and load that file into that turntable. Drop a tX-set file over the
255 "Load Set" button to load the set.
258 - fixes the load_piped() compilation error if you use neither sox nor mpg123.
259 Note that you have no "pre-listening" without mpg123 for mp3s and no
260 "pre-listening" for any other file without sox.
263 - Yet another redesign ;) Oh well this project moved from c++/multithreaded
264 to c++/singlethreaded, to c/multithreaded and now, finally, back to
265 c++/multithreaded. This shows quite some foresight, I'd say ;)
266 - Due to the redesign I had to change a lot (everything?) again so this document
267 can impossibly hold all changes. but let's see:
268 - tX now uses floats instead of doubles for a lot of parameters. Therefore the
269 old .terminatorXrc.bin is now called .terminatorX3rc.bin to allow parallel
270 use of older releases. But you'll have to configure the new release again
271 the first time you run it.
272 - tX now supports an unlimited no of turntables. The only limit is your display
274 - Each of these turntables has their own pitch / volume settings and a
275 dedicated lowpass and echo effect section.
276 - You can now map mouse motion to a function of your choice (scratching/
277 volume/cutoff/echo-feedback and nothing)
278 - Recording will now happen straight to disk. You'll have to specify a filename
279 before you start recording.
280 - You can sync the turntables to one master turntable.
281 - There are two master settings for pitch and volume available to modify all
282 turntables. The turntable pitch is relative to the master pitch.
283 - You can save a complete set (of turntables and settings) to a .tX set file.
284 The audiofile data will *not* be saved within that file, but it'll contain
285 the absolut filename to these files. I know this makes the act of exchanging
286 sets over the net more complex so if somebody wants to code set files with
287 integrated audio data... ;)
288 - The way audiofiles are loaded has been redesigned too: the builtin wav
289 routines, the sox support and mpg123 are no longer connected. This means
290 you can use mpg123 support without sox support and still have the builtin
291 wav routines enabled. If you have the builtin wav reader and sox enabled, tX
292 will try try to use the builtin routines first, and if they fail sox will be
293 used as a fallback which is of course slower as you know... There's some
294 support for Benno Senoner's brilliant 3.2-mmap() patch but the patch still
296 - You can now select which parameter to map to a mouse axis.
297 - experimental 3DNow! support (works with gcc < 2.95 only)
298 - tX-windows now have an icon, thanks Michael!
301 - implemented 2D-Scratching ;) You can now use the other axis of your mouse
302 two modify the cutoff frequency of the lowpass filter (if enabled). The
303 resonance factor can be set via the slider underneath "LP Filter". Seems
304 to eat quite some CPU power though...
305 - Updated THANKS, AUTHORS
306 - as most people seem to use sox/mpg123 anyway and as I still get "I want mp3"
307 emails I changed the configure script slightly: It now automatically enables
308 the use of sox and mpg123 if these are available. If you still want 'em good
309 old 16Bit/44.1Khz/Mono builtin routines you have to specify --enable-wavonly
310 with your configure cmdline.
311 - Prepared the tarball for the Win32 port. You can now compile tX with cygwin
312 and the Makefile.cygwin in the src directory if you have all the required
313 libs (gtk, pthreads etc...). The resulting exe will run BUT is completely
314 unusable for now... Stou Sandalski <tangui@cell2000.net> and I are planning
315 to make it usable in the future.
316 - The About-Dialog now contains some info on how the binary was compiled
319 - Optimized audio-rendering code for speed (might run on some machines
320 with smaller buffsize (-> lower latency) than pre-3.1) - Hey, tX now uses
321 less than 10% CPU (says kpm) with mixing enabled on my i586-200MMX machine!!
322 - "fixed" stdout-output. Due to the pipe-buffering realtime schratching can
323 not be considered usable for now :(
324 - Added config option --enable-scheduler, which will cause the terminatorX
325 engine-thread to run with FIFO scheduling policy and highest priority. This
326 might enable some machines to use smaller buffsizes. It requries root
327 privileges though - either run terminatorX as root - or make terminatorX
328 suid root WHICH IS A BAD IDEA FOR MULTI-USER SYSTEMS IF OTHER USERS CAN
329 ACCESS THE BINARY ;) You only have to think about this when you experience
330 buffer underruns with your desired latency. And it might not even help...
331 - Rehacked the de-clicking (fading) code. The old one tried to be much more
332 intelligent - and failed fading out the signal sometimes. The new code
333 is much simpler (faster) and is much more reliable.
334 - added common sense to the warp feature (f-key in scratch mode). The warping
335 speed is still connected to the mouse speed but it's now adjusted to the
336 sample size as well: you can warp through any sample completely with the same
338 - added debug configure switch. Enables (very few) debug messages for now. I
339 decided not to enable the old V2.X messages as they are sent to stdout, which
340 would sound pretty strange to people who use the stdout-option.
343 - Fixed Adrian's name in the THANKS file. Sorry, Adrian.
344 - Corrected Makefile.am for license.c that did not build correctly on some
345 systems - Thanks to Michael Pruett <mikep@ugcs.caltech.edu>
348 - fixed a bug that caused terminatorX to crash without --enable-sox or
352 - Changed everything ;)
353 - switched from C++ to C for gtk's sake
354 - switched back two multithreading
355 - implemented new gtk+ gui and dialogs
356 - new mouse-sampling code now uses DGA's DirectMouse
357 - implemented "initalizing"-support for XInput devices.
358 - new (alpha) "pre-listening" feature in scratch/loop dialogs
359 - hacked vtt for click-free audio. Might be problematic with big
360 kernel level buffersizes but scratching with big buffsize is no
362 - options are now stored in ~/.terminatorXrc.bin automatically (on exit)
363 as well as windowsize and the last used filenames
364 - terminatorX now comes with a configure script - no more Makefile hacking
365 required (nor a good idea ;) )
366 - new wavdisplay-widget update frequency is now configureable
367 - new position-display is much more visible which allows more precise
369 - new mute-feature allows you to silence the scratch-playback completely
370 while the loop keeps playing (press either <m>-key or right mouse-button.
371 - support for mouse buttons:
372 o 1. (left) button: same as <SPACE>, do scratch
373 o 2. (right) button: mute scratch
374 o 3. (middle) button: stop
375 - wrote support for sox and mpg123. This enables terminatorX to read close
376 to any audiofile. You need to have sox and mp3 and this feature (as it is
377 still alpha) needs to be enabled via configure-switches see INSTALL
378 - licmak now is an awk script. Thanks Christian! I was also received two perl
379 scripts (Thanks, guys!) but I now use the awk script as it is the smallest.
380 - introduced <F>-key to "warp" through scratch-sample (just check it out)
381 - dropped commandline arguments
382 - dropped old X-Code + Mouse warping madness
383 - bug fixes (less than bug introduces though ;) )
387 - implemented a time counter
388 - licmak now finally is what it should be: a perl script (licmak.pl)
389 it was contributed by: Adam Sjøgren <asjo@diku.dk>. Thanks!
392 - X-Window background fixed
393 - wavfile overreading fixed by Andrew C. Bul+hac?k (acb@zikzak.net)
394 - Using portable type definitions now where required. If you don't
395 have a <sys/types.h> on your system and your machines type sizes
396 equal those of an X86-machine you can specify -DUSE_X86_TYPES in
397 Makefile to compile without <sys/types.h>
398 - Slight change in audiodevice-handling that fixes improper buffsize
399 settings on some soundcards (e.g. GUS)
400 - Implemented "keep device open"-feature. Needs to be enabled via the
401 -DKEEP_DEV_OPEN switch in Makefile
402 - Added some comments to the Makefile for easier editing
403 - Added Big Endian support. Use -DBIG_ENDIAN_MACHINE for big endian hosts
404 like ultrapenguin suns or linux ppc. There's also -DBIG_ENDIAN_AUDIO
405 for big endian audio hardware (rare?) available, although proabably
406 nobody will need this.
407 - Changed mix behaviour: Mix is now automatically turned on on startup if
408 the user has specified a proper loop file. (with the --loop or -l switch)
409 - New mixing routine. You can now set the loop to scratch volume ratio with
410 the cursor keys (left/right). If you modify the volumes while recording
411 these action's will not be recorded. Saving happens with the currently
413 - Fixed README-Bug said <m> instead of <x>
416 - first released version
417 - added colors. more standard use of X11.
418 - added loop support.
419 - added statusbar with mode/status display
420 - added commandline options and help (on usage and keys)
421 - added raw/mixed save scratch feature
422 - added licmak to include license into binary
423 - new handling of XAutoRepeat, only turned off now when SPACE pressed.
424 - modified wave display (faster + nicer + incorrect (who cares anyway))
425 - Makefile fixed by Adrian Reber (adreit00@fht-esslingen.de)
431 - added recording, playback of recording functionality
434 - moved from multithreaded to singlethreaded
440 - no more position display due to pthread/X probs
446 - accelerating vtt via sinus