Frederik,
I am enjoying 3.2.0 but have run into a small problem. With all the changes you made to caching and so forth, the Reload command (cmd-R) only reloads if there has been an actual change somewhere.
I think you presented this as a feature, and perhaps it is, but it creates a small problem for me.
I am constantly trying to improve the efficiency of the subnetworks I make. When a subnet is slow I go on the hunt for some bottleneck or bug that is causing the delay. To do this I step through the network node by node reloading as I go by hitting cmd-R. As I do this I keep a sharp eye on the little processing dot in the upper right corner.
As I work my way down the tree, rendering and checking, the dot barely flickers until suddenly I render a node that makes the dot stay on for full second or two or ten. And that's how I spot possible bottlenecks. This immediate feedback is one of the things I love about Nodebox.
But now this technique no longer works. Without actual upstream changes of any kind, cmd-R does nothing.
I can render a node deep in a sub-sub-network, go upstream somewhere, make an inconsequential change, time the dot, drill back down, and repeat, but this is cumbersome.
Can you think of another way to easily get feedback on how fast each node takes to fire?
Aside: I have long fantasized about adding visualization features to Nodebox at design time. At the top of my list would be a display option to color code each node by how fast it runs. Most of the nodes would show in some neutral color for quick, with slower nodes turning yellow and then red. In this way you could spot possible trouble nodes at a glance. You would probably need some way of adjusting the sensitivity.
I don't really dare to hope for such an elegant addition. But can you think of some simpler workaround, or some easy-to-add addition, that would restore the functionality I used to enjoy with cmd-R?