From reading the spec, I understand r and t to be field elements and not arbitrary-length integers. The python reference, however, will directly hash as an integer, even if it is greater than MODULUS.
I believe that r and t should be taken as field elements, since it consumes less memory to discard the full hash as soon as a field element has been generated from it, instead of having to keep two copies around. Also, I believe this approach would simplify the semantics.
From reading the spec, I understand
randtto be field elements and not arbitrary-length integers. The python reference, however, will directly hash as an integer, even if it is greater thanMODULUS.I believe that
randtshould be taken as field elements, since it consumes less memory to discard the full hash as soon as a field element has been generated from it, instead of having to keep two copies around. Also, I believe this approach would simplify the semantics.