Re: Stream event structure

[Ralph Little]

Erik Sandberg wrote:
> > According to the thesis, the stream event contains a "context" field
> > which contains the id of the outlet context for the event. I can't seem
> > to find this in the event structure at all and I can't think of a good
> > reason for it to be there anyway for the implementation as it stands.
> >
> > I'm playing with exporting streams and being able to identify a context
> > id for the Voice would be really useful to me. Does anyone know how
> > reproduce the listing that Erik has in his thesis including the outlet
> > context id?
> >
> > Any help would be really cool.
> >     
>
> There is no official external representation format for music streams; in the 
> internal representation the context information is represented by appearance 
> (if the event appears in voice V, then it belongs to voice V). A music stream 
> dumper will typically consist of one listener per context, which exports the 
> events to some textual format. This listener will typically associate each 
> context object with some unique human-readable ID, and tag each heard event 
> with this ID (when exporting you could e.g. use a context -> id hash table, 
> and when importing a stream, use a hash table for id -> context).
>
> Note that you will need to listen in each context for the creation of new 
> child contexts, so that you recursively can add listeners to them as well.
>
> FWIW, I've attached the code I used for import + export in my thesis (beware 
> that they are written for an old lily version, and that they are not cleaned 
> up, and that context ids were hardcoded in C++ code back then).
>
> In that particular version of lily (2.6), I could override the translation 
> procedure by setting the global function translator-proc; this is the entry 
> point of my dumper/undumper. I don't know if this still is the way to go.
>
> IIRC, I have written some similar code for the official implementation of 
> streams, but it was lost when my laptop was stolen. If you're lucky, I might 
> have published some code on lilypond-devel (but I don't remember).
>
>   
> <etc...>
Hi Erik,
Many thanks for this - your comments below confirm what I suspected 
about ids - the context that gets the event is whatever happens to be 
the bottom listening context at the time.

I will look through what you have sent and see if I can get something 
working.

I'm really trying to get this Braille export thing off the ground and it 
seemed like a good place to start. I know others are looking at Midi 
also so perhaps I will contact them and collaborate as the framework is 
likely to be very similar.

Cheers,
Ralph