On Tue, Oct 17, 2006 at 03:09:03PM +0100, Andrew STUBBS wrote:
The expression parser/evaluator must use host types at some level. Does
this not have similar horrors hidden away? If this can convert target
types to host types (of at least the same precision) can the printf not
be implemented with the same mapping somehow?
Well, I suppose all we need to do is appropriate widening. We can pass
everything to the host code as LONGESTs and DOUBLESTs.
Are you interested in working on this? If so, I can quickly pull
together a patch to add the printf backend that I was thinking of;
all it'll need is glue.