Sorry for rambling post...

A few people on cnczone seem to swear by it, I've only ever used the stuff before for patching up rust holes in cars not for anything structural but watching this vid impressed me enough to seriously consider it:

DP-420 3M EPOXY overlap shear test by Doctorbass PART 1 - YouTube

(In part two he adds a load more weight and the join holds perfectly)

The thing with welding properly is it does take a lot of practice, if you don't prepare the joint carefully, don't make sure you have good penetration and don't make a consisitent pool in the weld than the strength of the join drops off dramatically, that is why people doing mission critical welding like nuclear power plants or submarines insist on x-ray inspection of every single weld. A weld can look and will look absolutely perfect but still be structurally weak, especially with arc welding where bits of slag can slip into the weld and be completely invisble to you when you have a welding goggle on.

If the people recommending repeated tack welds ever took an x-ray of the joints or sawed through them they would be shocked at how weak the joint they ended up with is in real life, fortunately the amount of steel that gets used for CNC machines is not there for strength, it is there to provided rigidity. (I know nothing about CNC but I can weld...)

Metal epoxy seems to provide more than enough strength for the job of holding a machine together but introduces another problem because the young's modulus of the various epoxy mixes seems to vary between 2-20 Gpa (compared to about 70 for ally or 200 for steel, higher means more rigid). I'm satisfied that any epoxy joint will structurally hold together on a CNC machine but I still haven't worked out what effect joining two pieces of steel together with what is by comparison a bit of rubber will have on the overall performance of the structure.

If I can't determine that my current plan is to build a frame from mild steel box section that is held together using brackets made from steel angle section welded to plate that is bolted to the box section. Enlarging the holes for the bolts mean I can perform accurate positioning before tightening the bolts and secure the joins using metal epoxy. That way I get the benefits of steel on steel rigidity and the epoxy makes sure that it will never move out position but if I welded a 'perfect aligned bolted structure' the thermal shock of the welding would distort the alignment as metal expands and then contracts. Bolting alone would mean loosening of the bolts over time.

My current thoughts anyway!
*Subject to radical change when I get a clue about what I'm doing. :)