1: \begin{abstract}

2: \noindent

3: The Douglas--Rachford algorithm is a simple yet effective

4: method for solving convex feasibility problems. However, if

5: the underlying constraints are inconsistent, then the 

6: convergence theory is incomplete. 

7: We provide convergence results when one constraint is an

8: affine subspace. As a consequence, we extend a result by Spingarn

9: from halfspaces to general closed convex sets admitting least-squares solutions. 

10: \end{abstract}

11: