VisibleRenderables problem and other bugs...

Hi, I got a few problems with PLSM.

I'm using:
SVN version: https://ogreaddons.svn.sourceforge.net/ ... glandscape
Compiled successfully with VS2005 SP1 the plsm2_vc8_SDK.sln using OgreSDK 1.4.8
Running it with the same physical data I got the precompiled demo running fine (no apparent bugs).

Problem is my compiled demo isn't paging the terrain, it fully loads and displays it disregard VisibleRenderables configuration.
Release or debug doesn't matter.
Actually the problems get worse in Debug because the ogre camera goes nuts. Position is like 3.24e+008. Watching mCurrentCamera variable with the debugger shows that the variable is initialized correctly, but then after the first frames(or the first one) the camera data gets invalid somehow.
After a few minutes debugging i found the problem.

PagingLandScape2FrameListener.h
mTranslateVector is not being initialized.
mCurrentCamera->moveRelative(mTranslateVector); puts invalid data into the camera
mTranslateVector = Vector3::ZERO; fixs it over the FrameListener constructor

Now terrain shows up in Debug but I mean the entire terrain, without paging.

I'm using the grand canyon to test. Again, the same data and configuration is being shared between the precompiled demo I got from the official page and the one I compiled. Precompiled demo shows only something around 75 renderables creating a very distinct vision radius like expected.

Any thoughts?

Thank you.

I must correct myself. It isn't fully showing the terrain, but it's definitely not paging it correctly.

The paging of precompiled demo somewhat follows your camera like expected, but with my version a big chunk of the terrain, like a page or two is being fully displayed and you can get the camera to the border and beyond to where a new page should be displayed but this never happens.

LOD is working fine when camera moves.

Any help would be appreciated.

Anyone using PLSM could give me a hint or what?

Thx

After some more experimentations I found out that PLSM under release compilation is working better then I have described above.

So I must concluded that there might me a hard-to-find bug (probably another variable initialization problem) built in to the PLSM library that's affecting the debug version only.