Hi Kristofer,

I've just reviewed your changes, and while I'm still need to do another
review, my initial reaction is "ooohh that's clever" :-)

You've come up with a solution that I never occurred to me, my first pass
review it seems to be the best solution proposed so far.  As it's rather
novel I do need a bit more time to reflect on it to see if there are any
potential problems.

In terms of CPU costs, I don't think it should be too expensive as I
wouldn't expect that to be a high number of AntiSquish nodes in the scene
graph, if it does turn out to be high then we can look at a
Transform::dirtyCount() but my hunch is that the cost will be negligible
compared to the rest of the CPU overheads.

Cheers,
Robert.


On 10 January 2014 08:34, Kristofer Tingdahl
<[email protected]>wrote:

> Hi Robert,
>
> I don't necessarily agree that a way to check if the transforms along a
> nodepath using integers is not worthwhile, as you replace hundreds of
> double-operations with a few integer ones, and the comparison becomes the
> comparison of two ints. That seems useful, especially as there is hardly
> any overhead (a few bytes per transform).
>
> I'll leave that discussion and submit a new update of the AntiSquish. It
> now inherits Transform rather than MatrixTransform, so there are no changes
> to the object while traversing (apart from one cache). This solves most of
> your objections as we rely on the standardness of the Transform. One
> question is however what to do when there is no node-path available, such
> as the Transform::computeBound(). In those cases I rely on the cache (if
> available). But if I do that, I should really dirty the bound when I
> re-compute my cache. This is implemented now in the version I'm sending,
> but I'm not sure it is necessary. It will create an overload of
> computeBound calls.
>
> I opted to lock the cache as this is the only way to protect from corrupt
> cache if there are multiple traversals going on at the same time. We don't
> do that, but I thought it's safer.
>
> Best regards,
>
>
> Kristofer Tingdahl
>
> _______________________________________________
> osg-submissions mailing list
> [email protected]
>
> http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org
>
>
_______________________________________________
osg-submissions mailing list
[email protected]
http://lists.openscenegraph.org/listinfo.cgi/osg-submissions-openscenegraph.org

Reply via email to