[Alsamodular-devel] Re: [Fwd: Re: [Alsamodular-user] i'm still alive...]

[Steve H. <S.W...@ec...>]

NB, I'm not on als...@li... so it will ge
held...

On Thu, Feb 13, 2003 at 07:19:32 +0100, Lukas Degener wrote:
> >
> >
> >I dont think it is, I would guess that a grid is most natiral for a mod
> >synth, simulating the rack rows and slots.
> >
> Well, ok. i think this is really an implementation detail.

Sure.
 
> >The metadata format is not part of LADSPA, but it conforms to  a standard
> >for metadata descriptions of stuff in general.
> >
> >My plugins provide it, the guy who writes the BLOP oscillators does, and I
> >will write some for CMT.
> >
> Sounds interesting. What's the name of that standard? Can you point me 
> to some more info?

RDF. http://www.w3.org/RDF/
 
> >As it is external its not part of LADSPA, and doesn't have to be. Hosts
> >can still interpret LADSPA plugins normally, they just miss some
> >functionality.
> >
> So how does a host know if a particular plugin provides such an xml 
> file, and how does host know where to look for it?

The files are in /usr/(local/)share/ladspa/rdf/ the top level library
tells you information about plugins based on what it finds in there.
 
> >Yessss, but this stuff is grim XML, it requires two extra libraries to
> >parse it sensibly.
> >
> Huh?! What 'kind' of xml is this? Cannot be parsed by a SAX2 or DOM 
> parser? QT supports both apis i think.

They can parse it perfectly well, but your application cant make much sense
of the data, multiple RDF/XML syntax structures map to one metadata graph
structure (to make it easier to write basicly). I think this is wrong, but
its a bit too late to argue. There is one library that turns the XML data
structures into a graph and another that allows you to do easy queries
over the graph (eg. "what values does this scale have").
 
> >>- What about polyphony / voice assignement ?  In ams, each modules 
> >>processes data one time for each voice during one cycle.
> >>   
> >>
> >
> >I'd say this is a design error. The hardware equivalent is that you
> >have one set of modules for each voice. Sure, there is a CPU overhead,
> >but its not that big. AMS must handle LADSPA this way allready.
> >
> Not sure about this. actualy what ams does right now (like in the stable 
> releases) is  not so much different from creating a copy of the module 
> in question for each voice. Tthe main difference is, that those "copies" 
> share the same parameter instances in ams. This apart, both aproaches 
> are moreless aequivalent. Maybe it would be better to let the user 
> decide if he wants to "group" modules together. I need to think about 
> this in more detail.

In LADSPA you can point the parallel plugins control parameters to the
same floats and just change them once.
 
- Steve