OK. A quick re-scan of Will's paper (which I read long ago) convinces me that I have this wrong if we care about input precision. IEEE doesn't take a position on this, but I think we should. Clinger and Steele did some very good work in this area, and I've just asked Will whether Hack's later work would motivate any changes.
To be clear, I do *not* consider this a high-priority item. Swapping out the converter can be done easily; it's a well-isolated change. If somebody has working code they are wiling to provide I'll gladly adopt it, but I don't want to get bogged down in the morass of numerical analysis when I need to be thinking about type systems. I'll happily take this as a bug to be resolved later.
_______________________________________________ bitc-dev mailing list [email protected] http://www.coyotos.org/mailman/listinfo/bitc-dev
