improving the complex number division

[Gabriel Dos Reis]

dewar@gnat.com (Robert Dewar) writes:

| > What actually surprised me is that neither J=E9r=F4me nor you  noticed
| > that that implementation isn't being used for anything we have defined
| > semantics for.
| 
| But if it is used at all, it is a mistake. 

It is a mistake only if you define precisely under which circumstances
it is a mistake.  It is not a mistake in absolute.

| After all, Only recently has
| C tried to define semantics for fpt at all, but that is not an excuse for
| doing something completely wrong :-)

True, but irrelevant here.

-- Gaby