/[svn]/linuxsampler/trunk/ChangeLog
ViewVC logotype

Diff of /linuxsampler/trunk/ChangeLog

Parent Directory Parent Directory | Revision Log Revision Log | View Patch Patch

revision 660 by schoenebeck, Fri Jun 17 19:49:30 2005 UTC revision 1651 by persson, Sun Jan 27 15:07:11 2008 UTC
# Line 1  Line 1 
1  Version CVS HEAD (?)  Version CVS HEAD (?)
2    
3    * packaging changes:    * packaging changes:
4        - autoconf bugfix: the PKG_CONFIG variable wasn't initialized properly,
5          causing e.g. the libgig test to fail when
6          "./configure --disable-jack-driver" was used
7          (patch by Alexis Ballier)
8    
9      * general changes:
10        - bugfix: on some POSIX systems instrument editor plugins refused to
11          load as we used a non-portable Linux specific struct field
12          (fixes bug #70, patch by Ronald Baljeu)
13        - fixed endless loop which occured when loading an instrument editor
14          plugin DLL on Windows systems failed
15        - fixed memory leaks that occurred when liblinuxsampler was unloaded
16        - optimized the SynchronizedConfig class so it doesn't wait
17          unnecessarily long after an update
18    
19      * audio driver:
20        - removed the nonsense audio channel constraint (which was hard coded to
21          max. 100 audio channels) for most audio drivers
22        - JACK audio driver did not offer a device parameter "SAMPLERATE" as
23          opposed to the LSCP specs
24        - bugfix: the SAMPLERATE parameter of some drivers (e.g. JACK)
25          reflected the wrong value
26        - fixed a memory management bug in ASIO driver
27    
28      * MIDI driver:
29        - added JACK MIDI driver
30    
31      * instruments database:
32        - avoid time consuming samples scanning when adding instruments
33          to the instruments database
34    
35      * Gigasampler format engine:
36        - fixed a memory leak that could happen when a channel was deleted
37          while notes were playing
38        - made it possible to load an instrument even if the audio thread
39          isn't running
40    
41    Version 0.5.1 (6 December 2007)
42    
43      * packaging changes:
44        - added autoconf checks for pthread library
45        - added autoconf check for pthread bug found on certain NPTL-enabled
46          glibc versions (see Gentoo bug report #194076)
47        - added autoconf checks for MS Windows
48    
49      * general changes:
50        - bugfix: the thread used by an editor plugin didn't die when the
51          editor closed
52        - bugfix: Ringbuffer.h: fill_write_space_with_null() did not zero
53          out all the space. operator--() did not apply size_mask after
54          decrementing the read_ptr. DEFAULT_WRAP_ELEMENTS set to 0 in
55          order to avoid problems with the _NonVolatileReader functions.
56        - bugfix: Stream.h: added a missing divide by BytesPerSample in
57          GetWriteSpace(). Since this function is currently only used in
58          the stream's qsort() compare function, it didn't trigger any
59          bugs.
60        - Resampler.h, Synthesizer.h: cubic interpolation now works in
61          24bit mode too. Faster method to read 24bit words on little
62          endian machines (x86): replaced 3 byte reads + shifts with a 1
63          unaligned 32bit read and shift
64        - experimental support for MS Windows (MIDI input via MME, AUDIO
65          output via ASIO)
66        - made handling of SIGINT signal (Ctrl-C) a bit more robust
67        - support for monitoring the total number of active disk streams
68          (new LSCP commands: GET TOTAL_STREAM_COUNT,
69          SUBSCRIBE TOTAL_STREAM_COUNT, UNSUBSCRIBE TOTAL_STREAM_COUNT)
70    
71      * AUDIO driver:
72        - added Windows ASIO low latency audio driver
73    
74      * MIDI driver:
75        - added MME Windows MIDI driver
76    
77      * LSCP server:
78        - added support for Windows style path / filenames, however with
79          forward slash path separators instead of backslash
80          (i.e. "C:/foo/bar.gig")
81        - allow naughty liblscp to send non-string device parameters within
82          apostrophes as well
83        - added new LSCP commands: "GET FILE INSTRUMENTS <file>",
84          "LIST FILE INSTRUMENTS <file>" and
85          "GET FILE INSTRUMENT INFO <file> <index>" for retrieving informations
86          about an arbitrary instrument file on the system where the sampler is
87          running on
88    
89    Version 0.5.0 (15 October 2007)
90    
91      * packaging changes:
92        - config.h is not going to be installed along with liblinuxsampler's
93          API header files anymore
94        - only the API relevant header (and source) files will be exported to
95          the Doxygen API documentation (explicitly listed in Doxyfile.in)
96        - added completely new XCode project files for Mac OSX which is now
97          capable to execute our autoconf environment, thus no need anymore
98          to maintain the compile time configuration file (osx/version.h) for
99          OSX manually (patch by Toshi Nagata)
100        - fixed buggy boolean --enable-foo / --disable-foo configure script
101          parameters
102        - global.h now only covers global definitions that are needed for the
103          C++ API header files, all implementation internal global definitions
104          are now in global_private.h
105        - atomic.h is not exposed to the C++ API anymore
106        - no need to include config.h anymore for using LS's API header files
107        - fixed warnings in API doc generation
108    
109      * general changes:
110        - replaced the old, confusing MIDI program change mechanism by a
111          flexible MIDI instrument mapper which allows to map arbitrary
112          (MIDI bank, MIDI program) pairs with arbitrary
113          (engine type, instrument file, file index) triplets which will be
114          loaded on the respective channel when such MIDI program change
115          messages arrive, beside that, each entry allows to define a life-time
116          strategy for the instrument, so the user can define whether the
117          instrument should i.e. be permanently loaded (to be able to switch
118          quickly among sounds) or i.e. loaded just on demand when the
119          respective program change arrives (to safe RAM space), as well as a
120          global volume factor for each entry, so the user can adjust the volume
121          dynamic of his mapped instrument collection without having to modify
122          the instrument files, also one can manage arbitrary amount of such
123          MIDI instrument maps and assign each sampler channel individually
124          a certain map, so that i.e. separation of normal instruments and
125          drumkits is possible
126        - new notification events for tracking changes to audio/MIDI devices,
127          MIDI instrument maps, MIDI instruments, FX sends, global volume.
128        - sampler was limited to load max. 200 instruments in the background
129          due to a constant size RingBuffer FIFO which is now replaced by a
130          dynamic (unlimited) size std::list FIFO
131        - added FX sends, these allow to route audio signals to arbitrary audio
132          output channels for being processed by external effect processors
133          (i.e. jack-rack), the send levels are controllable via arbitrary MIDI
134          controllers
135        - global (sampler wide) volume can now be controlled at runtime
136        - Implemented new, improved notification system
137        - fixed compilation errors regarding OSX
138          (patch by Toshi Nagata)
139        - implemented instruments database
140        - added support for escape sequences to the instruments database
141        - added highly experimental support for on-the-fly instrument editing
142          within the sampler's process (by using instrument editor plugins),
143          you'll notice the new "Registered instrument editors:" message on
144          startup, added a new LSCP command:
145          "EDIT CHANNEL INSTRUMENT <sampler-channel>"
146          to spawn a matching instrument editor for the instrument on the
147          given sampler channel, the plugin path can be overridden at compile
148          time with ./configure --enable-plugin-dir=/some/dir
149        - added experimental code for synchronizing instrument editors hosted
150          in the sampler's process to safely edit instruments while playing
151          without a crash (hopefully) by either suspending single regions
152          wherever possible or if unavoidable whole engine(s)
153        - fixed several issues in fundamental "Thread" class: set scheduling
154          policy and priority on thread level, set a minimum stack size for
155          thread (TODO: a reasonable value yet to be tested), bugfix: non-RT
156          threads simply inherited properties of starting thread instead of
157          setting their own policy and priority
158        - minor fix in our automatic stack trace mechanism on crashes, the
159          main process did not wait for the stack trace process to finish
160          its output
161        - fixed some minor memory leaks
162        - reenabled assembly features support, at the moment only for
163          enabling a fast denormal FPU mode (x86 platforms supporting SSE2)
164        - minor assembly fix in x86 features detection (don't use the PIC
165          register, to avoid relocations in the text segment at runtime)
166        - POSIX callback functions of Thread.h are hidden
167    
168      * MIDI driver:
169        - dispatch bank select (MSB and LSB) messages
170    
171      * audio driver:
172        - the ALSA audio output driver parameters now reflect the correct
173          parameter value ranges for the respective selected sound card
174          (patch by Till Wimmer, a bit fixed and extended)
175    
176      * Gigasampler format engine:
177        - if a filter is used and EG2 finishes before EG1, let the voice
178          die when EG2 has finished (this fixes a problem with clicks and
179          voice starvation for some gigs)
180        - playback is no longer disabled during instrument loading
181        - all notes playing on a channel that changes its instrument keep
182          playing with the old instrument until they get a note off
183          command
184        - EG fix: a release value of zero could cause noises or crash
185        - handle MIDI coarse tuning messages (MIDI RPN #0 MSB #2 LSB)
186        - EG fine tuning: when attack is zero the EG starts at a level
187          above max sustain level, which means that there is a decay phase
188          even if sustain is 100%
189        - more EG fixes: the level could sometimes go below zero and cause
190          noises or crashes
191        - minor fix of EGDecay (patch by Toshi Nagata)
192        - fixed compiler error when --enable-override-filter-type was
193          supplied to the configure script (fixes #46)
194        - disk thread: queue sizes are now proportional to CONFIG_MAX_STREAMS
195          instead of a fix value
196        - behavior fix: on MIDI CC# 65 (portamento on / off), 126 (mono mode),
197          127 (solo mode) only kill voices if the respective mode really
198          changed
199    
200      * LSCP server:
201        - fixed compile time error for old Bison versions
202          (i.e. v1.28 found on MacOS 10.4, patch by Toshi Nagata)
203        - parser now supports extended ASCII character set
204          (up to ASCII code 255, i.e. includes now umlauts and accents)
205        - filename arguments in LSCP commands now allow to use escape
206          sequences, that is directly literal as one of: \', \", \\, \n, \r,
207          \f, \t, \v, or as octal ASCII code value like \132, or as hex ASCII
208          code value like \xf2) (fixes bug #24)
209        - the following LSCP commands now also support escape sequences for at
210          least one of their text-based parameters (i.e. name, description):
211          "ADD MIDI_INSTRUMENT_MAP", "MAP MIDI_INSTRUMENT",
212          "SET MIDI_INSTRUMENT_MAP NAME", "SET FX_SEND NAME", "CREATE FX_SEND",
213          "SET DB_INSTRUMENT_DIRECTORY NAME",
214          "SET DB_INSTRUMENT_DIRECTORY DESCRIPTION", "SET DB_INSTRUMENT NAME",
215          "SET DB_INSTRUMENT DESCRIPTION", "FIND DB_INSTRUMENTS",
216          "FIND DB_INSTRUMENT_DIRECTORIES"
217        - returns verbose syntax errors (line and column where syntax error
218          occured, the unexpected character and the actually expected, possible
219          character(s), the latter only if less than 5 possibilities)
220        - made sure that LSCP syntax is not affected by gigedit locale
221          settings
222        - bugfix regarding strings parameter lists: all comma separated lists
223          of strings were treated as being one string containing commas
224          (fixes #57)
225    
226      * Bug fixes:
227        - fixed crash occurring on certain LSCP scripts (Bug 39)
228        - another thread safety fix for lscp "load engine" and "set
229          channel audio output device"
230        - fixed a crash which occurs when reassigning the same engine
231          on a sampler channel with connected MIDI device
232        - fixed a crash which occurs when changing the number of ports of a MIDI
233          device connected to a sampler channel to number less then or equal
234          to the index of the port to which the sampler channel is connected.
235        - The previous bindings were not been disconnected when altering
236          the ALSA_SEQ_BINDINGS parameter. Introduced a NONE keyword for
237          unsubscribing from all bindings (e.g. ALSA_SEQ_BINDINGS=NONE).
238        - The active stream/voice count statistic was incorrect.
239        - notification events were not been sent for some sampler
240          channel changes
241        - added default min and max values to restrict the number of allowed
242          audio output channels and MIDI input ports
243        - the connection to the PCM interface is now closed when destroying
244          an audio output device
245        - files with slash in their path or filename could not be loaded
246    
247      * test cases:
248        - updated and fixed (haven't been touched in a while)
249    
250    Version 0.4.0 (24 November 2006)
251    
252      * packaging changes:
253        - changed deprecated copyright attribute to license; added ldconfig
254          to post-(un)install steps; added devel package for liblinuxsampler;
255          to linuxsampler.spec (RPM)
256        - install necessary development header files for allowing 3rd party
257          applications to link against liblinuxsampler
258        - liblinuxsampler's API documentation can be generated with 'make docs'
259          (Doxygen required)
260        - added benchmark to automatically detect the best triangle LFO
261          implementation (currently either an integer math solution or a
262          di-harmonic approximation), automatic detection can be overriden
263          with --enable-signed-triang-algo=x and --enable-unsigned-triang-algo=x
264          configure script argument though (mandatory for cross-compilation)
265        - do not automatically pick optimized gcc flags if the user already
266          provided some on his own (as CXXFLAGS)
267        - added compile time option to disable processing of All-Notes-Off MIDI
268          messages
269        - added compile time options to allow disabling the various audio and
270          MIDI drivers
271        - fixed automatic GCC CPU switch detection on PPC
272          (patch by Ebrahim Mayat)
273    
274      * Gigasampler format engine:
275        - extensive synthesis optimization
276          (reimplementation of EGs and LFO(s), removed synthesis parameter
277          prerendering and the synthesis parameter matrix in general, splitting
278          each audio fragment into subfragments now where each subfragment uses
279          constant pitch and filter coefficients. The volume coefficient is
280          linearly interpolated inside a subfragment, unless
281          --disable-interpolate-volume is set.)
282        - fine tuning of the EG modulation parameters
283        - improved filter cutoff calculation by adding support for the
284          following gig parameters: Cutoff freq (used when no cutoff
285          controller is defined), Control invert, Minimum cutoff, Velocity
286          curve and Velocity range. The keyboard tracking now scales
287          cutoff frequency instead of resonance.
288        - added support for gig parameter Resonance.
289        - fixed bug in sysex handling (patch by Juan Linietsky)
290        - added global volume attenuation of -9 dB (0.35f) to prevent clipping
291          which can be overridden with --enable-global-attenuation
292        - EG fixes: made the length of "attack hold" stage more
293          accurate. Release stage can now start before attack stage
294          ends. Cancel release didn't work when sustain was zero. Attack
295          time now has a minimal value to prevent clicks.
296        - fixed pitch changes larger than one octave
297        - fixed EG3 (pitch envelope) synthesis which was neutral all the time
298        - implemented portamento mode and solo mode (a.k.a 'mono mode'):
299          all modes can be altered via standard GM messages, that is CC5 for
300          altering portamento time, CC65 for enabling / disabling portamento
301          mode, CC126 for enabling solo mode and CC127 for disabling solo mode
302        - fine tuning of the curves for volume (CC7), pan (CC10 and gig
303          parameter) and crossfade
304        - added support for the "attenuation controller threshold" gig
305          parameter
306        - added smoothing of volume changes caused by control change
307          messages
308        - sample loop parameters are now taken from the DimensionRegion
309          instead of the wave chunk
310        - fixed keyswitching for v3 gigs with a number of keyswitch splits
311          not equal to a power of two
312        - reimplementation of the filter algorithm. The new filter is more
313          accurate and supports all gig filter types, including bandreject
314          and lowpass turbo.
315        - real support for 24 bit samples - samples are not truncated to
316          16 bits anymore
317        - support for aftertouch (channel pressure, not polyphonic
318          aftertouch)
319        - LFO1 behaviour fixed (dampening from max volume instead of
320          amplifying from 0)
321    
322      * LSCP server:
323        - fixed application exit on broken pipe error (fixes bug #20)
324        - fixed the notification messages delay due to lack of
325          network activity (fixes bug #26)
326        - fixed parser bug which occured on space(s) within device parameters
327          (e.g. "SET AUDIO_OUTPUT_CHANNEL_PARAMETER 0 0 SOMEPARAM='foo bar'")
328    
329      * audio driver:
330        - added aRts audio output driver (by no means RT safe)
331    
332      * MIDI driver:
333        - fixed legacy sysex code which caused dispatching of MIDI SysEx
334          messages several times instead of once
335        - API extension for MIDI drivers which already supply exact time stamps
336          for events (i.e. for offline rendering based MIDI drivers)
337        - added checks for bad MIDI data
338    
339      * linuxsampler application:
340        - show available sampler engine types on startup
341    
342      * general changes:
343        - support for muting sampler channels and solo mode of the same, two new
344          LSCP commands ("SET CHANNEL MUTE" and "SET CHANNEL SOLO") and two new
345          fields ("MUTE" and "SOLO") for command "GET CHANNEL INFO" were
346          introduced for this, the behavior is the same like on a mixer console
347          (patch by Grigor Iliev, a bit adjusted). Also added configure option
348          --enable-process-muted-channels which can be used to enable the
349          processing of muted channels.
350        - support for sostenuto pedal
351        - support for monitoring the total number of active voices
352          (new LSCP commands: GET TOTAL_VOICE_COUNT, GET TOTAL_VOICE_COUNT_MAX,
353          SUBSCRIBE TOTAL_VOICE_COUNT, UNSUBSCRIBE TOTAL_VOICE_COUNT)
354        - fixed some memory management errors
355        - fixed some concurrency problems that could lead to crashes when
356          LSCP commands were executed
357        - fixed crash when instrument loading failed previously on the same
358          sampler channel (fixes bug #36)
359    
360    Version 0.3.3 (15 July 2005)
361    
362      * packaging changes:
363        - fixed compilation with gcc 4.0
364    
365      * Gigasampler format engine:
366        - LFO preprocessing was performed even though the respective LFO was
367          disabled by the instrument patch (minor efficiency fix)
368        - if period time of chosen audio device is too small (< MIN_RELEASE_TIME)
369          for volume ramp downs in the same fragment (needed for current voice
370          stealing implementation) then simply inform the user with a warning
371          about possible click sounds and reduce the volume ramp down
372          appropriately instead of cancelling the audio device connection
373    
374    Version 0.3.2 (24 June 2005)
375    
376      * packaging changes:
377      - updated autotools build files to compile on OS X      - updated autotools build files to compile on OS X
378        (conditional compilation of CoreMIDI and MidiShare drivers)        (conditional compilation of CoreMIDI and MidiShare drivers)
379      - hand-crafted assembly optimization code can be disabled with      - hand-crafted assembly optimization code can be disabled with
# Line 15  Version CVS HEAD (?) Line 388  Version CVS HEAD (?)
388      - volume of release triggered samples now depends on note-on velocity,      - volume of release triggered samples now depends on note-on velocity,
389        note length and gig parameter "release trigger decay" instead of        note length and gig parameter "release trigger decay" instead of
390        note-off velocity.        note-off velocity.
391      - revised voice stealing (fixes crash caused by voice stealing)      - revised voice stealing
392          (fixes crash and endless loop caused by voice stealing)
393      - don't reset scale tuning on instrument or audio output device change      - don't reset scale tuning on instrument or audio output device change
394        - handle key group conflicts right at the beginning of each fragment
395          instead of when the respective voice is actually launched
396          (fixes undefined behavior if stolen voices belonged to a key group -
397          this case was followed by a "killed voice survived" error message)
398        - fixed minor issue with null/silence samples
399          (those stole voices even though they don't need a voice at all which
400          resulted in "voice stealing didn't work out" messages)
401        - don't reset volume, pan, pitch and MIDI controller values on
402          instrument or audio output device change
403    
404    * LSCP server:    * LSCP server:
405      - fixed some crashes (patch by Grigor Iliev, fixes #19)      - fixed some crashes (patch by Grigor Iliev, fixes #19)
406      - fixed LSCP event "CHANNEL_INFO" notification      - fixed LSCP event "CHANNEL_INFO" notification
407        (e.g. did not notify on volume changes)        (e.g. did not notify on volume changes or MIDI program change events)
408    
409      * linuxsampler application:
410        - added command line parameters --lscp-addr and --lscp-port to override
411          default IP address and port of LSCP server
412    
413  Version 0.3.1 (24 May 2005)  Version 0.3.1 (24 May 2005)
414    

Legend:
Removed from v.660  
changed lines
  Added in v.1651

  ViewVC Help
Powered by ViewVC