i wouldn't be so sure... i think whats going to happen is its going to be pretty close to FVF 338 format, just using some of the extra Data assignments - structured as pure per vertex data rather than a solid model.
i'm releasing a FVF exporter next week for Milkshape so atleast that'll help alot of people. its not the same as a standard exporter because you script the format you use on exit so to export DB's in the script window you'd do
#def EXPORT( XYZ | NORMAL | DIFFUSE | TEX0 )
structdata
{
FLOAT px,py,pz
FLOAT nx,ny,nz
DWORD d
FLOAT tu,tv
} 338;
with that its simple to see what goes to what and then the exporter will take this and apply it as it would a C program
types of EXPORT ->
DIFFUSE - dword 32bit diffuse colour
NORMAL - 3float Normal Vector (Cannot be used with XYZRHW)
PSIZE - float point size, TnL distance Bias
SPECULAR - dword 32bit specular colour
XYZ - 3float Position Point (Cannot be used with XYZRHW)
XYZRHW - 4float Translation Point (Cannot be used with XYZ or NORMAL)
XYZB1 -> XYZB5 - 1 to 3float Position Point w/vertex or matrix weight blending
TEX0 -> TEX8 - (*TEXFORMAT Defualt 2float) Texture Coords, capable of 8 per vertex
TEXFORMAT1 -> TEXFORMAT4 - Format1 is 1float Texture point Coords, Format4 is 4float Texture point Coords. Now if you use this the exporter assumes that each Tex Coord is defualt Format2 unless otherwise stated.
TEXCOORDSIZE
n - you'll probably never need this but you can use this instead of TEXFORMAT
n to specify a number of floats you can't have with them ... but its always good to have something there eh

TEXCOUNT_SHIFT - dword number of bit to shift count of the Tex Coords into the main format memory offset
personally i can't wait until they finally release the details on this format cause i'd love to get stuck into it
Tsu'va Oni Ni Jyuuko Fiori Sei Tau!
One block follows the suit ... the whole suit of blocks is the path ... what have you found?