On Wed, 1 May 2002, Martijn Sipkema wrote: > > The following patch: > > > > http://fett.net/m/alsa/alsa-0.9.0rc1-unitor8.patch > > > > implements a driver for the Emagic Unitor family of multiport MIDI > > interfaces. Only serial connection is supported. You need to rerun the > > autoconf toolchain after applying the patch. > > > > The driver is heavily based on serial-u16550.c, which may be the wrong > > concept. A more correct concept seems to have a generic-serial.c and > > generic-usb.c, which only implement a communication channel. The interface > > drivers could then load&use them and only implement their own > > init/port-select/whatever dialect. Lots of MIDI interfaces come with both > > USB and serial connectors nowadays. Any thoughts on this? > > This is nice. What documentation did you use for writing the driver?
http://www.math.tu-berlin.de/~sbartels/unitor/unitor8_doc.txt It's in German. I have a translation of it I made using some online translation thingy a while ago, but I still find the German version a bit more readable (and my German really sucks). > I think a driver for interfaces like the unitor should not be in the > kernel. The standard serial port driver already allows access to the > device. I agree. However, I needed something working fast =) > Also, using a kernel driver does not allow using many different serial > hardware, right? I don't see why not. As long as it doesn't limit itself to a maximum of one instance at a time. > A solution could be alter libasound to support non-kernel drivers. Or > use a loopback device in the kernel. Perhaps only supporting the alsa > sequencer is an option. I like the libasound driver idea. A while ago, there was some talk about moving the sequencer stuff to user level too, with all that low-latency stuff happening in the kernel nowadays. What do you ALSA guys think about this? > I am myself still working on a library for MIDI i/o that uses plugins > for device access and IMHO provides a better interface than alsa > rawmidi. I thought of this as well, but I would like to use the alsa sequencer API for some experiments. -- Manush
