One way to do float format conversions (among a whole bunch of formats) is to
use the arbitrary-precision and variable-format float machinery in the GCC
compiler (file real.c). It probably doesn't quite give you what you need --
conversion from one format to another -- but I imagine that adding this
wouldn't be all that hard. If you need to do exotic stuff, like convert from
PDP-11 D format to IBM-360 double, that would be a way to do it.
It can readily be extended to add formats it doesn't currently have. I added
PDP-11 float to it in the past, that was pretty easy. I suspect stranger
formats like CDC 6000 might be harder, but probably still doable.
paul
