On 28 Mar 2002 09:14:51 -0700
Josh Green <[EMAIL PROTECTED]> wrote:

> On Thu, 2002-03-28 at 04:33, Stefan Kost wrote:
> > hi,
> > 
> > as a small suggestion, what about using xml via libxml2 (www.libxml.org) for the 
>fileformat. This way, you don't
> > need to write a parser and furthermore you know that whats you read is indeed a 
>devicespec. Example
> > <device name="r0x0r tranc3m4chine">
> >    <contection>
> >      <interface>name of mididev"</interface>
> >      <port>1</port>
> >    </connection>
> >    <banks>
> >      <bank name="rezocool phasers" msb="31" lsb="22">
> >        <patches>
> >          <patch nr="1" name="super acid pad"/>
> >          <patch nr="2" name="cool juicy 303 thing ">
> >            <controller>...</controller>
> >          </patch>
> >        </patches>
> >      </bank>
> >      <bank>
> >        ...
> >      </bank>
> >      ...
> >    </banks>
> > </device>
> > 
> > 
> > Stefan
> > 

I basically dont like xml because it's a lot more annoying to write.
These files are meant for people to write, not a program.

> 
> I'm curious about how this config file is going to be used. The interest
> I have in a MIDI instrument querying API is the querying protocol
> itself. I thought that the synth on the other end of the MIDI port (say
> an ALSA sequencer port) would just interface to this API. No need to
> know how this synth stores the actual patch info. It could be in a sound
> font or a fixed set of sounds for an analog synth. How useful would a
> static config file be in this API? Perhaps I'm missing something.. Ohh,
> a thought just occurred to me. You're probably thinking of this for
> external MIDI devices that don't have a method of communicating what
> patches they have? Hmm. Sorry for my ramblings :) Cheers!

hehe.. exactly. This is meant only for devices that dont support querying,
so, even if you have to write a definition file, it keeps transparency in the api.


Juan Linietsky

Reply via email to