I feel writing off appnotes of several reputed companies as absurd is not right and many of us have derived reasonable amount of help from these very appnotes and nobody can deny this fact.
The appnotes are written to show how CHEAPLY you cn make stugg (just see all the RC resets in appnotes) not how SAFELY. If you make this as a potted unit, that is not connected to the mains till after potting, it is perfectly safe.

Senior members do not want to divulge on this topic citing safety issues.
How will one ever know the result unless he tries it
what if the result is death? Then he will never know it

the whole circuit is enclosed in a ABS plastic box. you will still be sued if a customer opens it and kroaks.

NOW, one thing has been missed: if you are, indeed, making a dimmer, you NEED to have the mains connected to the unit. THIS IS NOT AN AMATEUR PRODUCT.
My client has shown a competitor's product which uses capacitive resistance P.S and Direct Mains ZCD and the final product is as heavy as a pack of cigarettes. its been in the market since 4 years and asks me, why can't we do ours on similar lines ?
However, if this is a mass-produced unit, you may get by with a scheme like fig 10 of http://ics.nxp.com/support/d.../an467.pdf NOTE this uses a chip with a comparator which TOTALLY solves you 'directional' problem, I strongly suggest you switch to one such (some can be had for $0.50).

Now, again THIS IS NOT AN AMATEUR PRODUCT but if this is a mass-produced unit, DO NOT use an external switch unit, then the low voltage wires will cary high voltage. CONSIDER potting the assembly to protect prying fingers that open the box. BUY a 1:1 power transformator to make the supply to your testbench 'groundless' (do not rely on a rubber mat or such).

Erik