Re: [DSSI] Re: [linux-audio-dev] DSSI (Disposable Soft Synth Interface) 0.4

[Chris C. <ca...@al...>]

On Wednesday 08 Dec 2004 10:20, Steve Harris wrote:
> Hmm... well the host ip and port number of the imcoming message are
> unique for the source, so the host could remeber it from the original
> registration message, and treat messages from that source as coming
> from the UI and not reflect them.
>
> But, I dont think there is a way in the liblo API of getting the
> address or port of an incoming message, its a longstanding bug.

We need to do something about that, because it's a pretty fundamental 
problem.

What I was trying to do was very simple -- write a script that starts 
fluidsynth-dssi with GUI visible and a particular soundfont loaded.  It 
can't be done with our current tools, at least not reliably and with a 
consistent UI, because the host won't update the UI when it gets an 
external configure request from dssi_osc_send.

We absolutely need to be able to do that.  Either we find a way to 
identify incoming messages, or we adopt Fons's suggestion (always 
updating the UI even for things it initiated itself).  The latter would 
mean changing the structure of existing GUIs, so we should do that now 
if we were going to.

(btw, the same example works as an example of a case where the existing 
GUI would need significant modification if it were to fit Fons's 
scheme.  Calling configure twice with the same key and value could 
produce different results -- for example reloading a file that has 
changed in the meantime -- so a UI can't just ignore a configure 
request when it happens to have the same value for that key already.  
This means the Fluidsynth UI would have to act to load its soundfont 
only when it got the configure update back from the host, not when the 
user first clicked the file load OK button, or else it would be loading 
it twice.  That would work, but probably be a bit less pleasant for the 
user.)


Chris