Yes, I am aware of the problem, but did not quite know how to deal with them. For instance, should there be something added to configure to make these functions conditional?
I am happy to have a go at writing a fall-back version of isnan, finite and isinf for doubles and floats that could be used by OpenDX and that work on IEEE platforms. Is it at all any point in considering other platforms? I have seen a OpenDX IEEE related macro definition, but I am not sure what it is used for (and I couldn't find it right now).
