On 16/05/2012 15:19, mike.artixels wrote:
Jon,
Thanks a lot for these clarification especially about aggressive
caching. I have a recent experience that after I deleted a DAG node,
there is still one instance not deleted, I discovered this by accident
because my node prints debug message when Nuke prompts it via the
Memory class freeCallback(). So is this also an expected case ?
Yes this is expected, it's basically part of undo. The op may be fully
deleted later at some time but you shouldn't rely on it.
As my Op requires internal memory buffer for processing, so I guess
it's not safe to share common memory buffers among them, right? as
they might be doing rendering at different contexts .....
Well it depends, if you know that the buffers have the same data then
you can share it. You can use the op hash as a key for this data. It
will be unique for the different contexts. You should make sure you
register these buffers with DDImage and free them when requested by the
freeCallback though, if you're not using them. This is important.
Don't just delete them on _close() or in the op destructor as that may
not happen.
Thank you, Jon.
Best,
Mike
_______________________________________________
Nuke-dev mailing list
[email protected], http://forums.thefoundry.co.uk/
http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-dev
_______________________________________________
Nuke-dev mailing list
[email protected], http://forums.thefoundry.co.uk/
http://support.thefoundry.co.uk/cgi-bin/mailman/listinfo/nuke-dev
