> other option also nice.
I thought about doing an object like that. There are two possible approaches,
both of which I didn't like:
1) access the htmlcanvas by sending Pd messages to the inlet. This seemed
really limiting, especially when you consider how deeply web programmers
leverage javascript and javascript frameworks when doing html5 canvas
drawing/animation. If you do similar such work in Pd to retain/manage state
and render then you're getting in the way of audio computation. If you leverage
a higher-level framework to do it on the GUI side then you're tied to it as a hard
dependency for all time.
2) Have a simple interface for just defining height/width/vis, and give a hook for
the user to specify and use their own js script to do the drawing. But this adds
an entire language to the UX which IMO is too great a burden, esp. for doing
prototyping.
Instead I just extended data structure drawings with the [draw] object to leverage
the SVG spec and do declarative vector drawing. If you look at
"Pd-L2Ork Data Structures" in the help browser you can see some examples.
-Jonathan
Yes, that's certainly possible.