[r6rs-discuss] thinko in spec of fxcopy-bit-field
aghuloum at cs.indiana.edu
Thu Sep 27 22:04:04 EDT 2007
On Sep 27, 2007, at 8:26 PM, Thomas Lord wrote:
> What you say is that a restriction was added in support
> of a particular theory of what strange compiler-related
> features it might be nice to pile on. Damn if that
> doesn't describe a lot of R6.
If you look at the primitives in (rnrs arithmetic fixnums)
and (rnrs arithmetic flonums), you'll see that they mostly
duplicate the functionality found in (rnrs base) and (rnrs
arithmetic bitwise) with the "restrictions" being the only
difference. If you want "the right addition procedure",
use + not fx+ or fl+. Use bitwise-copy-bit-field, not
fxcopy-bit-field, when you want an unrestricted version.
In short, use fixnums and flonums libraries when you know
the domain of your data. Use generic arithmetic when you
want the general unrestricted operations.
I don't see this as a piled-on feature as fixnum and flonum
operations exist in many R5RS Scheme implementations (maybe
some implementations choose not to expose them even though
they exist under the hood).
More information about the r6rs-discuss