●●●○ — It prevents me from completing a specific task with Vuo.
Steps causing the bug to occur:
Running the attached composition "Kinect Server v5". Works. I did see output from kinect.
Export it as a Mac App
Run the mac app: it crash.
Have you found a workaround?:
No. What's worse - no idea it's related to this bug - later on neither the composition started not to work - the Kinect patch now outputs nothing. Switching USB port the kinect uses, it works again. Next time you try, it doesn't. Booting from a different OS makes it runs again. Soon, neither on the other OS I can't use the comp any more.
No other software I own that used to run with kinect seem to get any input from it anymore. Some of them report "they're still running" - like if they find an instance of something already using kinect, perhaps. Some other can't find the device at all anymore.
I still can't exclude a hardware problem - as soon as I can test the kinect with other hardware I'll report it.
Note also that a previous compiled version of the app - Kinect Server v2, it had a few nodes less, just syphoning the kinect output - doesn't give any error on startup, but simply still doesn't give any output apparently from the kinect - while it worked a couple of days ago.
BTW dealing with this issues made me discover that "expression" input port is of type "List of math symbols". Seems that it works fine when connected to a "share value" node, but there's no way to connect to other kind o f ports, or set other generic type ports to that specific type. I'd like to know if there's a way to convert from text type to that type so that math expression could be somehow calculated on runtime. Don't know if this is a FR or there's something I'm missing.
Rethinking to this, I propouse a different behaviour. a thing like this:
if a new list reaches the node
if count item of the new list < actual internal node position
actual internal node position = count item of the new list
This way, the last backward event I described in point 3 would output item 2 of the new list.
This behaviour seems to me more reasonable.
