Hi all,
I would like to have a standalone pd engine for performance, when i don't need the user interface. I checked the source code, but it's not easy to follow every part of the code path between the engine and the tk interface. How difficult/long would you evaluate such a task, and is there a document somewhere mentionning the interaction points between the engine and the interface? Is there anything in the code which may render this task difficult?
-Nicolas
Nicolas,
This seems pretty difficult to do well as quite a lot of the gui is in pd (g_canvas.c, etc.). Perhaps the easiest thing is to let pd think it is connected to a gui without actually starting one. You can either write a separate program that pretends to be a gui but just ignores all the messages it gets or rewrite the communication functions to do nothing (s_inter.c). Two things to think about - pd expects a couple of things from the gui on startup (current working directory and font dimensions) that you can easily fake and you don't want to remove the select from sys_domicrosleep because pd uses this to do microsecond sleeps in addition to polling sockets. Good luck!
Karl
On Sat, 30 Sep 2000, Knos wrote:
Hi all,
I would like to have a standalone pd engine for performance, when i don't need the user interface. I checked the source code, but it's not easy to follow every part of the code path between the engine and the tk interface. How difficult/long would you evaluate such a task, and is there a document somewhere mentionning the interaction points between the engine and the interface? Is there anything in the code which may render this task difficult?
-Nicolas