Thanks for reporting these bugs.
The first one is indeed a serious one. In PHI, I modify Mathematica's Conjugate to have it expand a bit more than the default, but I seem to have assumed any exponent would be real. Not good.
The second one is also a bug. ComplexConjugate is simplistic and assumes anything not containing an explicit I (Complex) is real. But it used NumberQ to check for this. Also not good.
Both have been fixed in CVS and the fixes will be part of the next release.
This archive was generated by hypermail 2b29 : 04/28/17-10:00:02 PM Z CEST