"David W. Fenton" <[EMAIL PROTECTED]> writes: > Surely for a data bus that is assumed to have multiple devices using > it, *all* data has to be time-stamped in some fashion.
Why? For a multiple-device data bus, all you have to do is ensure that devices are tagged correctly and that events for a given device are in the correct order. Everything else is just detail, defining how responsive the entire system is. The requirement you state implies that every device needs to provide a time-critical component with short latency, which would raise the cost of every single device. Useful for RT, but not for a general purpose bus. > If not, then that means that nobody using a USB QWERTY keyboard with > a USB MIDI interface would be having success, since it would mean > that there's no way for USB to appropriate serialize events generated > by different devices destined for the same software process. That may have little to do with the bus (although with two devices that are requiring bounded-latency response to small packets of data, you could have problems with bus contention). If the software process is reading information from two different locations, the software may not be properly synchronizing. -- Stephen L. Peters [EMAIL PROTECTED] GPG fingerprint: A1BF 5A81 03E7 47CE 71E0 3BD4 8DA6 9268 5BB6 4BBE "Justice is the only tyrant we need obey." -- Urinetown: The Musical _______________________________________________ Finale mailing list Finale@shsu.edu http://lists.shsu.edu/mailman/listinfo/finale