I once did a MIDI extension for SpeechBasic to program a real time MIDI sound sampler on BASIC for the C64, for example
$1810 LDA $DEO6 $1813 LSR $1814 BCC $1810 $1816 LDA $DE07; read MIDI event byte, usually followed by RTS This are 6 bytes of code, to get a MIDI data byte from the UART. I did an experiment by the signature that makes Adrian going wild. Some lines for a signature are spam and I do agree. How many lines of code are used by ALSA MIDI to do the same? Lets say not per USB, but PCI. I wonder if MIDI on Linux could become better and if audio could need less resources when for Linux audio and MIDI, the concept might vary to the usual Linux concept. Some very elaborate Mac and Win audio apps do not need much resources or much bytes. I hope at least one person would understand that I'm not nagging. I try to understand Linux code, but it's very long and hard to understand, not comparable with the 4 lines above or even not comparable to longer code for professional Atari sequencers. Is there really a reason, I might not understand, to make Linux audio that complicated? Please, I guess some will write that I'm the only one having issues with ALSA MIDI and JACK, but c'mon ;) ... Cheers! Ralf PS: Latest ALSA MIDI latency tests fortunately did show that USB most times is useless, see also the 64 Studio lists archives, but PCI MIDI most times seems to be very good. Perhaps, if you should agree, there could be an advice on Linux audio web sides to prefer PCI MIDI. _______________________________________________ Linux-audio-dev mailing list [email protected] http://lists.linuxaudio.org/listinfo/linux-audio-dev
