[Wien] LDA+U changed between the versions?

[mazin]

This may have been already discussed here, but here is the question:
I have a calculation (CeTe3, if that matters) that was 2 years ago 
converged with LDA+U (FLL, U=0.32 Ry, J-0) and spin-orbit.

Now if I try to run WIEN2k_09 on these files, it finds the calculation 
completely unconverged and eventually converges to a different solution, 
with the lower Hubbard band 1 eV higher and affecting the Fermi surface 
(the UHB is also higher by 1 eV). If I run my old WIEN2k_07 executable 
instead, I get the same result as before. In fact, my old calculation 
then appears fully converged when I run the '07 executable.

Has something drastically changed in the way how LDA+U is implemented or 
in the way how it interacts with spin-orbit?

Thanks!