CVS Checkin

Just an FYI, I'll be checking in a few things a little later that include (but not limited to):

-LWJGL 1.0b update (with a custom lwjgl.jar that we had them build as we found an AWT issue)

-Improvments from Three Rings

-Badmi's exporter switch over stuff.

-The Font3D items

-Some memory optimizations from Gustav

Any explicit information on the improvements from Three Rings?


Any explicit information on the improvements from Three Rings?


any implicit? :slight_smile:

Ok, committed (here's the log).

ISSUE 207 -Addition of Font2d, font3d classes for text rendering.
ISSUE MINOR - Spatial transformer bug fix
ISSUE 197 - Model Converters switched to using BinaryExporter/BinaryImporter
ISSUE 198 - Nulling of renderqueue list items.
ISSUE MINOR - Memory optimizations of allocating lists to only 1 element (as that is the most common number)
ISSUE MINOR - Removal of batchCount for size call.
ISSUE 204 - Addition of GeometryTool, allowing for the optimization of batches (removal of duplicate verts, etc).
ISSUE 203 - Fix for strip/fan support when using VBO.
ISSUE MINOR - Fixes to DummyDisplaySystem to again allow use of jME without OpenGL
ISSUE MINOR - Fix for OggInputStream for small Ogg Files.
ISSUE MINOR - Updated to LWJGL 1.0b

Code donated from NCsoft, Three Rings, GamaLocus, and others.

Awesome. :slight_smile:

Dunno if this is a new thing, but any test I tried that uses a non-jme model format (maggie.obj, bike.3ds) seems to be broken here. Binary Header doesn't match.

The converters convert to the new jme format not the old one. If use the converters you need to use the new loader.

Ah, so all the old code is just overwritten now? Without changing the tests?

Some of the tests where updated some where not.

Just to let ppl know, it seems the default behavior of SpatialTransformer has changed from CLAMP to REPEAT.

That is a mistake, we'll correct it asap.

Checked in a few fixes and tweaks:

-Moved MaxToJme back to using the new exporter.

-SpatialTransformer back to clamp by default (this was just a testing artifact)

-Fixed a few of the tests.

Note: JmeBinaryReader would automatically apply boundings to models loaded in. However, the new importers and exporters simply load/save the scene as it exists, so you might not get boundings (converters do not apply them for example). Therefore, the setModelBound(Bounding Volume) and updateModelBound methods have been promoted to abstract methods in Spatial, and Node's implementation passes the call down to it's children. So setting boundings on complete models should be easy enough.