Seperated Vertex data (GE implementation)

[a_noob]

Ok well, I know guDrawArrays uses a tightly packed struct for rendering vertices/meshes I was however wondering if/how the GE interprets the data, does the GE take the vertex type and jump around in the data only snagging the co-ords then jump thorugh and do lighting etc? Or is it all interpreted inline? I may be wrong but I heard that it jumps though the data extracting each part at a time, so I was wondering if it was possible to create a feasible implementation of an opengl like drawArrays/client states. I ask, since IMO this allows for easier implementation of dynamic mesh data/types and it also allows for an easier time in cross platform development.

So basically Im asking How is a vertex list interpreted, and how/is it reasonable to split vertex data.

[a_noob]

BUMP, I would like to find out if this is plausable.