Cool VL Viewer forum

Hello Henri,

no idea if this is a significant issue.

When starting after some the viewer got killed by a TerminateProcess by accidentially clicking in the wrong row in ProcessExplorer, the next start greeted me with a message box and a crash. So there might have been some state on disk corrupted by previous actions.

Log & dump attached.

Hmm... Very strange, indeed.

Looking at the log (no shader compilation failure), the crash dump and at the corresponding code,I see nothing wrong, and the shader should have been bound before the LLVertexBuffer::setBuffer() call which threw this error.

For next release, I added a check in LLFace where LLVertexBuffer::setBuffer() is called, to check that a shader is indeed bound before calling it, and otherwise warn and abort the corresponding object render...

EDIT: in fact, yes, there is something wrong in the code, where the Moon shader is not bound (but used) when the Moon texture cannot load... Seeing how the viewer was failing to load textures and contact the server (inventory AIS errors, etc) in your log, I won't be surprised if this is what indeed happened. Fixed for next release (if you want to fix it in the current sources, simply remove the silly if() test around line 710 of indra/newview/lldrawpoolwlsky.cpp: sMoonShader->bind(); must be called unconditionally).

Fixed in today's release (v1.32.2.3).