danfrz / plebtracker Goto Github PK

I don't think this will fix the issue all of the time, but it might improve it in a lot of cases.
Regardless it's something to check out.

This may help when adding waveforms together, may sound nicer. Investigate

I am trying to add a Dockerfile to this project, to ease installation (you can try it out by just installing docker)

I am having trouble getting it to run with this Dockerfile:

FROM alpine:3.7

COPY . /usr/local/plebtrk
WORKDIR /usr/local/plebtrk

RUN apk add --no-cache alpine-sdk ncurses-dev fftw-dev alsa-utils inotify-tools
RUN cd ./Tracker/src && make && ./INSTALL.sh && cd ../../Interpreter/src && make && ./INSTALL.sh

ENTRYPOINT ["plebtrk"]

docker build .

Sending build context to Docker daemon  58.77MB
Step 1/6 : FROM alpine:3.7
 ---> 3fd9065eaf02
Step 2/6 : COPY . /usr/local/plebtrk
 ---> Using cache
 ---> e456797f511b
Step 3/6 : WORKDIR /usr/local/plebtrk
 ---> Using cache
 ---> bf0c1788b56e
Step 4/6 : RUN apk add --no-cache alpine-sdk ncurses-dev fftw-dev alsa-utils inotify-tools
 ---> Using cache
 ---> e9ba090f6443
Step 5/6 : RUN cd ./Tracker/src && make && ./INSTALL.sh && cd ../../Interpreter/src && make && ./INSTALL.sh
 ---> Running in d5036bb0b085
g++ -c -o ../obj/cmdbar.o cmdbar.cpp -std=c++11 -I../include
cmdbar.cpp: In function 'bool editor::getCommand(const char*)':
cmdbar.cpp:150:16: error: 'get_wch' was not declared in this scope
     get_wch(&in);
                ^
make: *** [Makefile:23: ../obj/cmdbar.o] Error 1
The command '/bin/sh -c cd ./Tracker/src && make && ./INSTALL.sh && cd ../../Interpreter/src && make && ./INSTALL.sh' returned a non-zero code: 2

do I need to install ncurses differently in alpine, maybe?

renderPattern should use different buffers for each renderTick call. Then filters can be performed within those renderTick call. Finally renderPattern would mix all of the filtered buffers together.

There is a short click sound on the beginning of song playback, after the pulse table was implemented.
The click isn't present for the duration of a whole row segment so it's an artifact, doesn't seem to change when pulse parameters change so may not be directly related.

When I wrote some of them I wasn't really focussed on making them work perfectly, I was mostly focussing on perfecting Pulse1. Pulse2 and Hybrid were afterthoughts so their output should be verified.

Confirmed not working:

• 08 - Pulse Saw (only has down-trough). Fixed
• 0C / 0D Not really implemented, substitute with something else later
• 0E 06 - Square Pulse 2 (REALLY Broken). Fixed
• 0F - Square Pulse Hybrid (Doesn't work because 0E doesn't). Fixed
• 14 - Triangle Pulse 2 (doesn't work as intended; is identical to 07). Removed
• 15 - Triangle Pulse Hybrid (Doesn't work because 14 doesn't). Removed
• 16 - Sine Pulse 2 (doesn't work as intended; is identical to 09). Fixed
• 17 - Sine Pulse Hybrid (Doesn't work because 16 doesn't). Fixed
• 18 - Half-Sine (broken, keeps resetting phase mid-waveform). Fixed
• 1A - N Sine W (broken, keeps resetting phase mid-waveform Only when wave param > 1). Fixed
• 1C N Sine Pulse 2 (doesn't work as intended; is identical to 1B). Fixed
• 1D N Sine Pulse Hybrid (Doesn't work because 1C doesn't). Fixed
• 1E N Half-Sine (broken, keeps resetting phase mid-waveform)

When I started the project I underestimated the value of stereo in music. When I get to live interpretation in the tracker (using OpenAL?) I should also add panning. These both will involve a number of infrastructural changes.

Goattrk uses a table for panning, and I think you can do some pretty cool stuff with that.

However there are so many tables visible all the time that you can barely see your composition within 80 character lines, and that's a use case I'd like to support.

Proposal: have filter and pan tables always hidden and provide a modifier key that when pressed displays pan and filter tables in place of the wave and pulse tables.

Bytes Per Row is exactly what the number is, "bytes" is vague.

Hold on. It's only bytes per row for 8 bit implementation. Shouldn't it be samples per row?

Regardless. Refactor this.

Use a window function before filtering, that should make filtering actually sound reasonable I hope.

Also, this turns out to be the proper solution to the problem, not adding more song data to the buffer of the filter operating. As a side note, keep the buffer larger than input data and power of 2n, but instead of copying song data in there, have zeroes. Apparently that can make filtering sound better.

This makes the fourier_buffer_size byte within songs unnecessary.
Also, may end up making window_type option for tracks, set by a function within the filter table.

There's a bug with plebitp where pulse effects will not work properly sometimes. I've noticed it only happens in certain instruments and when it does happen, it happens 50% of the times you play the same segment of the song, which strongly suggests some sort of uninitialized malarkey. This was happening even before the pulse table was introduced.

N action in pulse table currently directly copies the functionality of the wave table's which is not correct.

Write documentation for pulse table,
Remove documentation for functions moved from wave table

Is there any plan to support any of popular tracker formats?

Thanks.

Currently I do not see any movement or position indication when playing a track using plebtrk's :play function.

Would be nice to have.

Hello sir.
You have a great app, unfortunately this app does not have a logo yet, may I donate a logo for your app?

To reproduce:

Start plebtrk
Copy the order
Clone the pattern
Go to Order 0
Remove pattern 0
Go back to Order 1
Crash happens