Finally I have at least submitted something like an ebuild for g95 to
https://bugs.gentoo.org/show_bug.cgi?id=102199
  Testing of my build of Quantum ESPRESSO did not go exceedingly well,
too many segfaults while calculating the examples, but at least
example01, 04, 08, 09, 11, 18, 23, 27, and 29 seem to have gone well.
I'll try to find out, why so many calculations in the other examples
crashed, wheather it was due to some reproducible fault of the ESPRESSO
build, or some misconfiguration of my system and libraries, but my
abilities for that are even more limited then the time I can pay to that.
  In addition, I have compiled the CP2K
(http://cp2k.berlios.de/index.html) CVS code (``The latest CVS entry
found is /xray_diffraction.F/1.10/Tue Feb 7 10:40:09
2006//D2006.06.06.13.40.00,'' says the automatically generated manual)
with g95. The compilation went well with gfortran too (I attach my
Linux-i686-gfortran.sopt addapted for linking ATLAS), but the gfortran
build crashed when tested on tests/QS/H2O.inp. When I addapted
Linux-i686-g95.sopt to interface with gcc-4.1.1/gfortran compiled ATLAS
(I'm attaching that too) and used g95, the code not only compiled, but
also actually worked and computed at least tests/QS/H2O.inp successfuly.
  With best regards
						Honza Macháček