BREAKING CHANGE: the order of these operations was inconsistent
across the different VMs. Go VM was the only one to first modulate
and then apply note tracking multiplication. But that made most
sense. So now all different VM versions work in this same way.
The sample-based oscillators converted the samplepos to an integer
and did samplepos < loop_end comparison to check if we are past
looping. Unfortunately, the < comparison was done in signed math.
Normally, this should never happen, but if the x87 FPU stack
overflowed exactly at right position, we then got 0x80000000 in
samplepos, which is equal to -2147483648. Thus, we considered that
sample is not looping and read the sample table at position
-2147483648, well out of bound. TL;DR changing jl to jb makes sure
we always wrap within to sample table, no matter what.
Fixes#149.
When voice was silent, the exponential decays in the filter unit
were causing the high pass component to eventually denormalize,
causing high CPU loads. The solution is the same as in the delay
unit: add and subtract a small number from the value, causing
essentially a flush to zero.
https://en.wikipedia.org/wiki/Subnormal_numberFixes#68.
The modulated delay time was converted to int with i32.trunc_f32_u.
This throws runtime error if the modulations caused the delaytime
to become negative, because _u implied that it should be unsigned
integer and negative numbers were out of range. Using
i32.trunc_f32_s fixed this.
The Model was getting unmaintanable mess. This is an attempt to refactor/rewrite the Model so that data of certain type is exposed in standardized way, offering certain standard manipulations for that data type, and on the GUI side, certain standard widgets to tied to that data.
This rewrite closes#72, #106 and #120.
ConstructPatterns was never used except during compilation, so it
makes sense to have it closer where it is used. We could consider
making it even private function, as the pattern table construction
is quite specific to how compiler compiles and probably not that
reusable elsewhere.
The Render name misleading as it did not do the same thing as normal
Synth.Render, because it disregarded time limits. Conceptually, as
the function modifies the state of the synth, it would be better to
be synth.Fill(audioBuffer), but we cannot define methods on
interfaces; therefore, it is audioBuffer.Fill(synth) now.
The VMs now release all envelopes by default, so this mechanism was
useless / did not actually start them as unreleased even when you
thought they did.
The commands and values were not very good names to what the
byte sequences actually are: opcodes and their operands. In
many other places, we were already calling the byte in the Command
stream as Opcode, so a logical name for a sequence of these is
Opcodes. Values is such a generic name that it's not immediately
clear that this sequence is related to the opcodes. Operands is not
perfect but clearly suggests that this sequence is related to
the Opcodes.
The -er suffix is more idiomatic for single method interfaces, and
the interface is not doing much more than converting the patch to a
synth. Names were updated throughout the project to reflect this
change. In particular, the "Service" in SynthService was not telling
anything helpful.
Throughout sointu, we assume stereo audiobuffers, but were passing
around []float32. This had several issues, including len(buf)/2 and
numSamples*2 type of length conversion in many places. Also, it
caused one bug in a test case, causing it to succeed when it should
have not (the test had +-1 when it should have had +-2). This
refactoring makes it impossible to have odd length buffer issues.
BREAKING CHANGE: The problem with crush was that it had very few usable values. This changes the crush to map the value nonlinearly, so the crush resolution is bits. Still the upper portion of the values is not very usable (bits 12-24 i.e. hardly any crushing), but at least the lower portion is usable. But now crush resolution has slightly different meaning.
Specifically:
* Added win32, elf32 and elf64 asm player and wav writers using winmm.
* Added dsound player in C.
* Separated the ALL target and the examples; introduced a new examples target.
go v1.21 is more strict about giving methods to C.structs and was complaining about "cannot define new methods on non-local type *C.Synth". The solution was a local type alias: type BridgeSynth C.Synth
The RPC and sync library mechanisms were removed for now; they never really worked and contained several obvious bugs. Need to consider if syncs are useful at all during the compose time, or just used during intro.
There is a new "sync" opcode that saves the top-most signal every 256 samples to the new "syncBuffer" output. Additionally, you can enable saving the current fractional row as sync[0], avoiding calculating the beat in the shader, but also calculating the beat correctly when the beat is modulated.
The old "native" compiler bridged version is now started with cmd/sointu-nativetrack,
while the new pure-Go bytecode implemented bytecode interpreter is started with
cmd/sointu-track
Thus, you do not need any of the CMake / cgo stuff to run cmd/sointu-track
