[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [sc-dev] Towards SC 4 SERVER?

On Tue, Nov 19, 2013 at 1:40 AM, Scott Wilson <i@xxxxxxxxxxxxxx> wrote:
This is actually an interesting idea. Currently Node IDs are arbitrary ints, which makes it easy to have multiple clients node-wise. What matters is that the nodes have a unique ID, and that the total number of nodes is not exceeded.

If busses and buffers worked that way, and if Synths explicitly kept track of buffer and bus usage, it would a) be much easier to deal with them in multi-client situations, and b) we would have the makings of a way to very easily model dependancy. Then node order could be implicit rather than explicit, and parallelism could be inferred.

There's a reason why the intra-synth operation embodies a dataflow model, and why the inter-stream operation does not: the internals of a synth do not change dynamically, while synths themselves come and go (rapidly). The reason why inter-synth connections go through a simple bus model is precisely that dynamic dataflow reconfiguration is a very complex topic.

That being said, and taking into account that JITLib actually does build an inter-synth dataflow model on top of the bus architecture, it's definitely not impossible. But I just wanted to express my belief in the importance of the distinction between intra-synth and inter-synth worlds and be aware of their respective limits.