[R6RS] fixnum operators and overflow detection

William D Clinger will at ccs.neu.edu
Fri Mar 31 15:54:51 EST 2006


> Can't we think of some more mnemonic convention, or at least one that
> forces people to figure out which is which?

I can think of some, but I can't think of any that would fly.
I'm afraid to ask your opinion of Z_{2^N}+ for the wrapping
version of +.

> Like wfx+ and rfx+ ("wrapping fx+", "(overflow-)raising fx+"?

Ugh.  Maybe Z_{2^N}+ isn't so bad after all.

Seriously, I'm happy with fx+ for the signalling version and
fixnum+ for the wrapping version.  The shorter name should be
enough to encourage the signalling version.

I don't want to revise the arithmetic SRFI until we have some
agreement on this, because I don't want have to go through it
twice.

> (I'm quite afraid of fx+ becoming a common-place operation
> in everyday programs, in the name of speeding up things.)

That seems like a happy outcome to me, assuming fx+ is the
signalling version.  It would be better than what people are
doing now, which is to write non-portable code or to define
macros that expand into non-portable code.

Will



More information about the R6RS mailing list