On Tue, August 14, 2007 11:32 pm, Henning Thielemann wrote: > I used this implementation so far, but I hoped that sending bundles makes > this synchronous automatically.
this doesn't make sense? the async commands are precisely those that cannot be run from the synthesis thread, they are staged and run from a different thread. (see the scsynth architecture document in the sc3 documentation). > How can I know how long I must wait? Maybe you need to wait until the reply arrives. > I must wait too long and I am not able to send the next message in time. you need to ensure that you don't use resources that do not exists or are not in the state you require. scsynth is usually pretty good at gracefully handling this kind of 'non-sense' (ie. using buffers before they are allocated, nodes that don't exist, etc.) but it will not do as you hope! > I like to use the same data structure for realtime and non-realtime > performance. But as you said, messages with the same time stamp are in that case you will need to somewhat more sophisticated data structure! (it would not need to be complicated though) > executed in unpredictable order. That is > > send fd (Bundle 0 (d_recv g)) > wait fd "/done" > send fd (Bundle 0 (s_new "Anonymous" (-1) AddToTail 1 [])) > > won't work (if I omit 'wait' in non-realtime mode). but that doesn't make sense, it says: "start d_recv, which takes an unbounded/indeterminate amount of time, at 00:00:0 on january first 1970, wait until it notifies this process it has completed, and then start the node at 00:00:00 on january first 1970"? regards, rd _______________________________________________ haskell-art mailing list haskell-art@lists.lurk.org http://lists.lurk.org/mailman/listinfo/haskell-art
