- I didn't realize that sigrenderer->tick counted down from speed to 1, rather
Apply delta to compressed samples regardless of creating tracker version
Merge branch 'master' of github.com:kode54/dumb
- Rewrote arpeggio implementation to use note offset tables
Fixed songs with envelope loops on the last tick
Fixed envelope start and end shortcuts
Merge branch 'master' of github.com:kode54/dumb
Fixed envelope loop timing
Fixed so it only initializes envelopes when instruments are enabled
Fixed envelope carry to only occur if the running note has not already been cut or released
Further fixes to IT envelope behavior
Reimplemented envelope carry to be more like other players
various fixes to address GCC warnings
Moved DUMBFILE structure to its own header file to unify the rare cases where it needs to be visible outside of dumbfile.c
Eliminated all clang compilation warnings
Fixed C++11 compilation error on Mac OS X
Fixed another stupid error
Cleaned up several warnings, including a buffer
Disabled MSVC 2012 default of SSE2
Fixed intermittent noise with SSE resonant filters by special casing zero length render calls
Subsong scanner always tries to play the first order now
Re-enabled SSE support and rewrote SSE filter function
Fixed timekeeping when starting a sigrenderer with a time offset
Disabled SSE as MSVC 2012 breaks it
Fixed position status reset on callback-aborted loop
Added MSVC 2012 v110_xp platform
Fixed duplicating some timekeeping state variables
Small fix for possible issue with time reporting for looped modules if callback ends playback
- Implemented loop-accurate time position reporting into DUMB
M_PI may not be defined
the MEMFILE filesystem didn't have seek and get_size functions
renamed "restrict" parameters to "restrict_"
Removed unnecessary optimization function
Fixed regular surround mixing
Optimizations and bug fixes for the Lanczos resampler
Fixed resonant filter handling in surround mixer
Implemented 3 channel surround mixing
Whoops, that was Blackman-Harris, not Lanczos. Lanczos produces less aliasing, anyway.
Fixed phase calculation when downsampling
- Implemented Lanczos resampler - Fixed overhead of performing effects updates on background voices
Replaced FIR resampler's IIR low pass filter with a fixed set of cutoff stages
- Removed *16/11 scale from S3M global volume
Fixed panning correctly this time
Implemented support for compressed stereo samples
Changed panning rule from 3dB to 0dB to match most trackers
Fixed loop start for STK modules
Fixed serious error with MPT extension reading
Fixed envelope reading to gracefully handle nodes which are out of range
Implemented support for the obscure S9F sample reverse effect
More fixes for seeking support
Fixes for C89 limitations and some const types
Implemented full seeking in DUMBFILE and modified several readers to account for this change, which also reduces their overall memory usage
- Added silence padding for samples which are too short to predict padding for
Eliminate clipping from the low-pass filter stage by increasing intermediate buffer type size to int
- Simplified low-pass filter
Added a low-pass filter to the FIR resampler
Merge branch 'master' of github.com:kode54/dumb
Made SSE filters configurable
Added new files to MSVC projects
Cleaned up MSVC 2010 project file
Removed stale source control file
- Implemented FIR resampler
Fixed pattern counting, and swapped pattern counting flag
Fixed portamento up effect when NNA channels are active
Oops, forgot MSVC doesn't have snprintf
Buffer entire file into memory to allow for weird file offsets, and add minimal Open/MPT extension reading
Fixed memory reader for formats which perform empty skips at the end of the file, like Asylum AMF
Silenced MSVC warning, removed unused variables, and reordered some code for pre-C99 compilers
Added new 'any' reader to MSVC 2010 project
Added format tag to IT reader
Included SSE optimized it_filter function, enabled if _USE_SSE macro is defined at compile time
Implemented simple 'any' format reader which attempts to detect the format of the input file and passes it off to the correct reader.
Only provide our own log2 for MSVC, since other platforms should already provide it
Removed unused variable from function declarations, a leftover from copying from the MOD reader
Added limits.h for LONG_MAX
Fixed older pattern counting mode
Made the new MOD pattern counting system optional
Fixed ADPCM sample support
Fixed MOD pattern count calculation in weird cases where the last sample is missing four bytes
Added interface for inserting extra DUH signals, and fixed searching for IT sigdata when more than one signal is present
- Implemented final solution for MOD pattern count calculation