Rafael Garcia-Suarez wrote: >2009/9/26 Zefram <zefram@fysh.org>: >> On a quick test with blead, just removing those two lines from S_sortcv >> gave precisely the behaviour that I'm proposing. > >Does it break any test ? No. Apparently the current error case isn't being tested for. Attached is a patch that makes the change in behaviour, with corresponding test and documentation. -zefram