Thanks for the quick fix, Allin. It works fine now. Seems that not many
people use/like the Rq version for imposing restrictions :-)
Best,
Artur
 Hmm, that's apparently a bug that has been lurking for a long
time. It's
 now fixed in git. Note, however, that the simpler formulation
 
 restrict
   b1 = 0
 end restrict
 
 was not broken like the matrix R,q version. I guess that's why the
 problem hasn't been noticed before now.