[PD] Sending data from PD to another program
cyborgk at gmail.com
Mon May 7 16:19:51 CEST 2007
I thought of something like your solution too, Frank, I'll have to
give it an inspection after I get off work.
As for my own solution, I decided that it could be very useful for
Markov Chain / random walk type of operations, especially when they
are more cyclical and NOT tree structures. For a smaller number of
states, say under 20, it would be really easy to prototype using the
node method where you simply chain things together. In fact, an AI
generative music machine would greatly benefit from being able to do
such a walk.
However, is there any method for making dynamic select and number of
outlets in a node, without coding an external? Or would it be simpler
to hard-code the node abstraction with more outlets and let the user
choose to use less?
Also, I decided it would be nice to have node names in some kind of
namespace for the node, where it has the form 'name.message'. The idea
is that while the name might be unique, messages might not. This would
make it easy to write linear chord structures as a chain of dataflow
objects. So I'm trying to remember, can .39 vanilla PD split a symbol
based on some delimiter, such as '.' (decimal) ? It seems like I rely
on zexy for things like this sometimes! Or do I need to give the
abstraction both a name, and a message parameter?
On 5/7/07, Frank Barknecht <fbar at footils.org> wrote:
> David Powers hat gesagt: // David Powers wrote:
> > Here you go, just use the tree abstraction and create named nodes for
> > every event in your story!
> > Note that the nodes are hard coded to send to "master", you'd want
> > another "master" receiver for your GEM abstraction to select the image
> > which goes with the current node.
> > Oh, and of course replace the PD selector patch, which sends random 0
> > or 1, with your actual voting mechanism.
> > ~David
> > On 5/6/07, David Powers <cyborgk at gmail.com> wrote:
> > >I think I will try to do a quick prototype and email it to you in
> > >15-20 minutes.
> > >~David
> > >
> > >On 5/6/07, Jared <linux at fridaythang.com> wrote:
> > >> David,
> > >>
> > >> Thanks for the reply. That's basically what I want to do. My concern is
> > >> that there are around 80 nodes which have to be displayed in specific
> > >> if-then sequences (if it's at node 12 and the vote is A, go to node 33,
> > >> if the vote is B go to node 37). My understanding of PD means this isn't
> > >> *impossible*, it'll just be a bit clunky. If you have any suggestions to
> > >> the contrary, I'd love to hear 'em.
> I attached another solution to your problem, which may be easier to
> extend to 80 nodes, because not much patching is involved, instead you
> write a textfile with a definition of your state transitions (because
> basically what you seem to long for is just that: a state machine).
> The idea is to write a textfile where every line represents a certain
> state of your system indicated by the line number. The content of a
> line then specifies the possible transitions.
> In the example patch every state has two possible follow-up states
> written as numbers. Depending on the choice of a user (0 or 1 in the
> example) a new state is selected by selecting either the first or the
> second follow-up state and making that the new active state.
> Frank Barknecht _ ______footils.org_ __goto10.org__
> PD-list at iem.at mailing list
> UNSUBSCRIBE and account-management -> http://lists.puredata.info/listinfo/pd-list
More information about the Pd-list