> OK. What does a scalar with IV and NV both OK mean? I *think* it *should* mean that the cached IV and NV are both valid *and* that there is no loss of precision: (IV)nv == iv && (NV)iv == nv. Obviously, this doesn't seem to be the case. > And what are the private flags actually used form? No idea, sorry. > There seems to be an implicit assumption that NV preserves IV always, Yes, an unfortunate assumption. I've tried shaking it once in a while but too many tests have started failing (numconvert, it nothing else) so I have always backed up. > Am I right in thinking this is going to mean changing the 2nv and 2iv and 2uv > code in sv.c? Sounds familiar. -- $jhi++; # http://www.iki.fi/jhi/ # There is this special biologist word we use for 'stable'. # It is 'dead'. -- Jack CohenThread Previous | Thread Next