[Bf-committers] No destruction function for nodesystem?

[Ruan Beihong]

Thank you.
My node need some calculated image to be cached for later use, which is not
very likely in other node, but highly useful in my node design. The image
ususally will be reuse for three times.
And my other question followed. How can I tell one image from other? How to
know which one is computed and should be gotten from cache and which one is
not. It there any of knowing the CompBuf containing the same image? Maybe a
digest is needed. Then I find md5 is used in imbuf's internal code which is
not available for external use. What shall I do?

2010/1/29 Matt Ebb <matt at mke3.net>

> Hi,
>
> I don't know what your node is, but we do indeed need an image caching
> system for the compositor. However, where it's needed is at the end of
> the pipeline, to cache the viewed results, so you can scrub back and
> forward in time and still get live updates (like in the sequence
> editor, and in most other compositors). I'm not sure why you want to
> do it for one node only, but on face value it sounds like it may not
> be the right approach...
>
> Anyway, nodes can define their own free callback for their custom data
> in the node definition itself - check for: /* freestoragefunc    */
> down the bottom of a node's C file.
>
> Matt
>
>
> On Fri, Jan 29, 2010 at 1:31 AM, Ruan Beihong <ruanbeihong at gmail.com>
> wrote:
> > Hi all,
> >  I've recently trying to add a new composite node into blender. The work
> is
> > still on early stage. And I need to use MEM_Cache_LimiterC to manage
> CompBuf
> > of my node. I find I can initiate it in init_func for node struct, but I
> > can't find any place that I can free it. In free_nodesystem() in file
> > blenkernel/intern/blender.c it just free the list directly. Shouldn't
> there
> > be a way to do some destruction for node?
> >
> > Thanks.
> >
> > --
> > James Ruan
> > _______________________________________________
> > Bf-committers mailing list
> > Bf-committers at blender.org
> > http://lists.blender.org/mailman/listinfo/bf-committers
> >
> _______________________________________________
> Bf-committers mailing list
> Bf-committers at blender.org
> http://lists.blender.org/mailman/listinfo/bf-committers
>



-- 
James Ruan