A tool that uses some solid part of the bike as an anchor point is the best solution. It avoids putting any lateral load on the gearbox shaft like the factory tool does, loading up the drive train as with the locked rear wheel, or loading up the primary chain and sprocket teeth as with the chock between the primary drive sprockets.
BTW, the chock between the sprockets is by far the worst solution. Because of the relative angles between components, it puts extroardinarily high loads on the shafts, primary chain and sprocket teeth - we're talking numbers in the order of 1000kg. You could quite easily bend shafts, break sprocket teeth or break the chain. At the very least, it will stretch the taut side of the primary chain. I wouldn't use a primary chain that had been abused in that way. My recomendation would be don't even think about using that method - it's just asking for trouble. Even if nothing breaks while you're doing it, you could end up running on a weakened primary chain.
The primary case bolt holes are convenient anchor points because they are nearby, but the further away the anchor point the better. It's just like having a long handle on a spanner - the longer the handle, the less pressure it needs to turn a nut. So if you're going to use the M6 threaded holes for the primary case attachment, use the holes that are furthest away like Andy Wagner's tool does.
I assume the two tubular spacers in the photo of AW's tool are to space the tool the correct distance out from the primary case mating surface (no idea why they have nuts on them though). But those spacers will have a bending load on them because the tool applies a force to the outer end of the bolts. That bending load could potentially bend the bolts, or even tend to pull the bolts out of the aluminium casing using the edge of the spacer as a fulcrum (similar to the way a claw hammer pulls a nail). If that happens, the spacers could put an indentation in the primary case mating surface. A better solution would be to make the spacing setup more rigid by welding a plate between the two spacers or welding the spacers to the back side of the tool, or both. That way, there will only be a shear load on the bolts, and no need to do the bolts up particularly tight.
The distance from the centre of the crankshaft sprocket to the two rear primary case bolt holes is about 262mm (260.99 and 262.87 for the upper and lower holes respectively - for those of the obsessive compulsive pedantic persuasion
) so the load at the end of the torque arm when 100Nm is applied to the nut will be 381N (about 39 kg). That's spread over two bolts in shear. By doing a few more sums, the shear stress in each M6 bolt works out to be about 14MPa. Mild steel will handle around 600Mpa in shear before it will deform, so the bolts are nowhere near over-loaded. Even if you have to apply 200Nm or more to the nut to undo it, you still have a big safety margin.
I think I'll modify my tool to look something like Mr Wagner's.
Cheers,
Cam
now you've got me wondering what the right torque looks like. 
