This subject has been mentioned quite a bit in the past, but I thought a separate thread might be a good idea.
First, I'll start with how I think the program processes input data. I may be completely wrong -- I'm just guessing by observation.
It seems that either keyboard or slider inputs are processed almost immediately. The input is "pushed" through the scene calculation to update the preview or simply to prepare the scene for a render. I imagine a slider input is sampled in a time slice and that value is "pushed" through the scene calculation.
The problem is, especially when using a slider input, you can enter a value that causes a crash (real small inputs can cause this) even though your intent was not to enter that "real small number", but you went a little too far to the left with the slider.
Many times, I want to input several numbers, like X/Y/Z values. I don't need the x value going through a new scene calculation until all three values are entered -- nothing is gained by this. Try entering new data for the "Make River" function.
Of course the preview can be stopped, which helps, but I wonder if a "stop recalc" button would be practical? Maybe make slider entries effective only after the slider has been stationary for a period of time (1 or 2 seconds)?
I suspect there are times when an input will cause a crash but in conjunction with a second input, a crash won't occur.
Anyway, I'm not sure if this makes sense at all, but I do see this happening frequently in other node-based applications (like World Machine.)
Any thoughts on this?
