gstopp at fibermux.com gstopp at fibermux.com
Wed Jan 29 19:15:09 CET 1997

     I have a lot to say on this topic but I can't say it all now... this 
     topic has surfaced many times over the last decade and the time will 
     come when there is an obvious answer as promising protocols 
     proliferate. Could it be that it's getting to be time to re-evaluate 
     it again, now that high-speed packet silicon has dropped to consumer 
     comodity prices?
     TCP/IP probably isn't the right direction - it's at the wrong layer 
     for this discussion. Internetworking is not the problem, it's 
     real-time transport of control data. Besides, protocol stacks such as 
     this are usually bought and are copyrighted and all that yukky stuff. 
     Current LAN phy-layer protocols aren't suitable. Ethernet is a 
     "gutter" protocol - it's designed from the start to actually discard 
     data! Token Ring is a much more reliable tranport method, but the 
     physical layer is so complicated that it makes you suspect that IBM 
     budgeted millions of dollars in development money just to screw the 
     users. But the chips to support these are getting cheap, so maybe we 
     can keep the packet format but change everything else? You could stick 
     a pretty darn big bunch of note-on's into a 1514-byte packet - talk 
     about a big chord...
     Actually MIDI works pretty good as it is. It's just that we're pushing 
     it and we need more "headroom". It sure would be nice to have a tiny 
     $20 box hanging off your MIDI synth with a 100BaseT hole on it, going 
     to a fast Ethernet switch next to your PC, though, wouldn't it?
     Come to think of it, during playback there's only one dude doing the 
     talking anyway. Collisions may not be a problem, if you have different 
     modes of operation like one for setup and another for the actual 
     playing of music (or whatever). This was actually done before...
     There *was* an attempt to apply LAN principles to MIDI devices and 
     maintain backwards compatibility, back in 1989... it was the Lone Wolf 
     Miditap using the MediaLink protocol. Unfortunately the hardware was 
     quite expensive as it was implemented which of course shut out almost 
     the entire consumer world. (If you look at the first page of the 
     Miditap schematic, down in the lower right corner, you'll see my name 
     there...) Lone Wolf does not exist any more, the name of the 
     development group has changed to "MediaLink", and they currently 
     struggle under the finances of Paul Allen up in Seattle, most likely 
     deviating from the original product direction - I don't really know 
     all the details now, I split away years ago in pursuit of a real 
     9-to-5 job to pay the bills (besides, I didn't want to move to 
     OK time to go get more coffee.
     - Gene
     gstopp at fibermux.com

______________________________ Reply Separator _________________________________
Subject: Re: OVERRIDING M.I.D.I.
Author:  "Arnim X. Sauerbier" <arnims at usa.globelle.com> at ccrelayout
Date:    1/29/97 12:42 AM
>A friend of mine suggested TCP/IP as a transport layer... MIDI to 
>TCP/IP converters would fill a similar function that CV to MIDI 
>converters do now. 
To be picayune, TCP/IP and MIDI are apples and oranges.  TCP/IP is a 
networking protocol (or internetworking protocol) which basically lets 
you wrap data chunks into a standard-format packet.  It does not specify 
the physical or electrical characteristics of the transmission medium.  
MIDI on the other hand is a standard for electrical, physical, network 
and application layers of the OSI model.  It specifies everything.   But 
it's a pointless topic since the only standard that can take-over will 
have to be MIDI compatible (This is the lesson of MS-DOS, my son).
>But I reckon there's not enough demand at the
>moment to make it worthwhile. IMHO 99.9% of everyone is either happy 
>enough with the MIDI setup they already have, or can put up with its 
>limitations considering its flexibility.
Well, all the synth heads I hang-around hate MIDI's slowness and would 
gladly pay a good chunk extra for more speed.  But they're the 'pros'.   
The 'amateurs' are perfectly happy to hook-up their casio to their 
soundblaster MIDI port and find that it actually works.  
Perhaps there are technical reasons why a backwards compatible MIDI 
extension can't be designed -- otherwise, why haven't they done it?
Hmm - maybe someone reading this will become inspired to develop an 
improved, backwards-compatible MIDI standard?   Stranger things have 

More information about the Synth-diy mailing list