kklouzal wrote:
...aside from that I had a brilliant idea to remove all capital letters from the voxel cache location path and that seems to have had a positive effect.
Ok, well that's strange but at least you might be on to something.
kklouzal wrote:
...this is still with the edit to largevolume.inl, removing that puts us back to square one :/
So just to be clear, are you referring to this edit:
David Williams wrote:
Something else you can try is to comment out the relevant assert and set 'm_pLastAccessedBlock' to '0' instead. That is:
Code:
// POLYVOX_ASSERT(pUncompressedBlock != m_pLastAccessedBlock, "Attempted to delete last accessed block.");
m_pLastAccessedBlock = 0;
If that fix is helping then I think I'm happy to commit it to PolyVox. I don't understand exactly why it is needed, but the fix should be harmless.
kklouzal wrote:
Anyways heres what client.exe is telling me now...
I added this "Attempting to insert vertex between two voxels with the same value" assert a few months ago when I was seeing corrupted meshes in Cubiquity. As I recall it turned out to be a threading problem, in that the volume data was being modified by another thread while the surface extraction was taking place.
Are you accessing the LargeVolume from multiple threads at all? If so you will probably have problems - see here for more information:
http://www.volumesoffun.com/polyvox/doc ... ading.html
