if we are loosing a burst ACK because of fading signals or bad conditions or just because - combined with a changing speed level - theres now a good chance we are not loosing the entire data frame because of a corrupted CRC, anymore. We are now searching in already received data, if our new (smaller) data chunk already exists. If so, we are stripping the entire data to this position and adding our new data chunk. If not, it seems we received new data the regulary way. However, this could cause problems if we have a too small data entropie. We might also check if we have an increased NACK counter or if a data chunk exists, twice or more - in a further step
it seems, pyaudio isnt terminated correctly. running it within a subprocess which will be closed after execution, pyaudio seems to be terminated and not causing a problem anymore on raspberry pi. Yay
...and a lot of different changes. Also deactivated single mode transmission. This needs to be optimised another day...Time is the missing ressource...
added precompiled libcodec2 version for 64bit / 32bit / armv7 (raspberry pi). Also updated library search which is now much more flexible and (hopefully) less confusing.
changed protocol so IRS is now the speed-level master / send ptt state via network / introduced no rig mode / disable scatter and waterfall in gui and tnc/ increased network chunk size / ...
..and some other changes to the gui so its compatible again with the latest socket commands. The rx buffer has now a unique id, and new structure. Also all messages and files will be saved to the same buffer. All commands which will be sent to the tnc or dameon are now written in lowercase
an attempt with a locking state for the mod_out queue so we can process audio only, if we finished filling our mod_out queue. Possibly this solves the problems #99#127
