SQL handles [about] 90% ... of what the standard end-user wants to do VERY easily

Not if you have poorly-normalized tables that have corporate cruft that's built up for decades. Further, as described, in this case I have to denormalize copies for speed purposes. Anything more than a 2-way join on slim tables tends to be too slow for their needs.

SQL handles 90% (random percentage pulled out of my ass, of course) of what the standard end-user wants to do VERY easily. And easily does lots of complex joins that truly SUCK in any X-base derivative.

Joins only semi-suck in X-Base. Further, most modern dialects let one use *both* SQL and X-Base as needed. Thus, you can get the best of both worlds.