To check out this repository please hg clone the following URL, or open the URL using EasyMercurial or your preferred Mercurial client.
hg clone
Working. Loopback example.
Added averaging, not very useful, tbh
Improved thread-safety, prevented segfaults when calling cleanup() and setup()
Added timestamp check. NOTE: it works fine as long as the host is running at a low enough block size. Otherwise the ReceiveAudioThread buffer in the bbb is too small.
The assumption in the last commit was just wrong :) . Issue was I was expecting the thread to run DURING processBlock(). Solved (trivially) by adding multiple buffers, so that more than the full blockSize can be buffered before being sent
The issue was not with DatagramSocket::write() being too slow. Rather, the issue was that the scheduler did not call the SendAudio thread often enough. Reducing the block size of the hots to 128 makes everything work smoothly. Alternatively, you can send from the audio thread without issues (as long as the socket is always available or you accept dropping packets if it is busy).
Added NetworkSend port from BeagleRT. NOTE: using Juce::DatagramSocket to send data, but it is way too slow (maaany dropped packets). TODO: port UdpClient from BeagleRT
ReceiveAudioThread is now the same with/without JUCE. That is the same file can be shared between the UdpIoPlugin and the BeagleRT repos
A better step
Trying to go towards compatibility with the BeagleRT code
the >= would make the thing fail for buffers smaller than payloadLength, which might have been the cause for #1364
ReceiveAudioThread now has its dedicated files
Working ! Large latency, hardcoded values but working!
Removed two Juce modules which were causing weird issues in gdb
Working, at least with samplingrate=44100
Almost working!
close does not exist !
Added Linux target and updated juce rev
Work in progress towards threading2
Added 2 words header (which is ignored) to match the Scope class
Initial commit, loads of TODOS