On Sat, Apr 07, 2012 at 11:00:32AM -1000, Joel Roth wrote: [...] > Thinking some more, we would certainly see errors if ECI commands got > garbled somehow. > > I'm not sure what you mean by 'piecemeal', already only > one command goes at a time, and I'm sure Ecasound > buffers them. Perhaps we could experiment with > some small delays after every command. I guess I was mainly thinking of serialising the ECI commands with possibly some small delay between commands, since ecasound is possibly allowing them to come faster than it can digest. The possibilities offered by the new logging mechanism you discuss would certainly help us in that respect. > > We do have one other--perhaps minor--way to help > troubleshoot. While I believe both Audio::Ecasound and > Net-ECI both are fundamentally the same under the hood, > it can be worth switching between them to > see if the same error patterns appear (-l and -n flags). Fortunately, the thought already did occur to me as I was trying to track down the bug earlier this week, so I used both libecasound and netECI throughout. Unfortunately, the behaviour remained completely consistent, and so i was none the wiser for it. Cheers, S.M.