I’ve noticed that some node classes seem to be immune to hash-testing as a 
reliable means of checking their up-to-date-ness. So far Rotos/Rotopaints are 
the main ones I’ve noticed, and I’m starting to suspect Primattes as well.

Is there a list of node classes that are known to force a tree to be 
re-evaluated no matter what (by 'append'-ing pseudo-random numbers or something 
similar)? I know Reads work something like this to force reloads, but they just 
use an incrementing knob that can then be reset to '0' before re-testing the 
tree hash.

Do I need to develop my own method of node hashing to reliably keep track of 
statuses? Or is this a totally unrealistic/impossible situation? Right now I’m 
using something similar to the code the Write op uses to compute its embedded 
metadata hash, but this seems to be breaking down with certain node classes.

I’d love to know how the DiskCache node is working internally (as far as 
comparing cached files to tree states), as what I’m creating is basically a 
DiskCache node with a user-specified cache path.

Thanks for any hints on any of these,

-Nathan

_______________________________________________
Nuke-dev mailing list
[email protected], http://forums.thefoundry.co.uk/
http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-dev

Reply via email to