It's suboptimal to not have a clue on what's going on when an imager is taking some more time to finish. Wouldn't be cool to have a 'Processing.. imagers' at the bottom left of the ARV for example ?
Instead a general core thing still not optimal is that it looks like imagers cannot be stopped while running.
I mean .. our enduser-oriented workflow (for dev I could go with verbosity) to know when an imager has done its job 😎
How cool is this.
Hmm, that doesn't sound right, I'll log a defect and see what's best to do. thank you for the feedback and suggestions as always
Great. However I see this kind of things very hard to 'fix' until you won't introduce a way to re-init/stop/reset arnold sessions.
It's like with verbosity. It's completely nut stuff. It doesn't make any sense that when an end-user stops the renderer (after having exhausted max warnings.. that's most likely if the warnings are from the render loop) .. he won't see any verbosity until maya is re-started (that will make him believe whatever was warning out, is now working correctly) .. even if he keep increasing max warnings. That happens for the same reason imagers keep running.. arnold persistent session mode in maya, you really need to introduce proper session lifecycle management APIs etc.
For rombotools imagers I solved porting them to CUDA and by putting stuff in the imager loop to prevent them to run indefinitely.
🖖
