[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: fixnumXXX and fxXXX names, and other things

This page is part of the web mail archives of SRFI 77 from before July 7th, 2015. The new archives for SRFI 77 contain all messages, not just those from before July 7th, 2015.

Concerning multiple values, whose only connection to this SRFI
is that several procedures return multiple values,
Aubrey Jaffer wrote:
> If the system eschews them, what are the bounds of the system; are
> libraries part of the system?

System designers don't need my advice on where to draw the
boundaries, but Marc was talking about implementors who
might want to use the reference implementation as a basis
for their implementation of R6RS arithmetic.  In my opinion,
arithmetic is an important part of any Scheme system.

> Is it incumbent on platform-neutral
> libraries to have multiple-value and non-multiple-value alternates?

No.  I was stating my intention with respect to the reference
implementation for this SRFI only, to acknowledge Marc's point
that multiple values perform poorly in some systems, and that
the heavy use of multiple-value-returning procedures that would
perform extremely well in some systems would not perform well
at all in less performant systems.

> What mechanism is there for library code to discover whether the
> implementation running it has fast multiple-values?

None.  The reference implementation of R6RS arithmetic will be
a starting point for implementors who wish to conform to R6RS
arithmetic, not a library for users to use as an alternative
to the native implementation of R6RS arithmetic.