Am Mi., 28. Okt. 2020 um 00:49 Uhr schrieb Arthur A. Gleckler
<xxxxxx@speechcode.com>:
>
> On Tue, Oct 27, 2020 at 3:29 PM Wolfgang Corcoran-Mathe <xxxxxx@sigwinch.xyz> wrote:
>
>>
>> Right, and whether the NaNs we examine have a quiet bit is also
>> implementation-dependent.  So the C implementation will just have to
>> be as-portable-as-possible.  The final say goes to the SRFI editor on
>> whether that, plus an FFI layer, is an acceptable sample
>> implementation.
>
>
> When portability is impossible, it certainly isn't required.  A sample implementation that illustrates the API and the challenges faced when implementing it, even in the context of a particular Scheme implementation, is still valuable.
>
> By the way, all these primitives appear to be defined by MIT Scheme here and here, but under different names.  If you find that I'm right about that, you could just reference that as a sample implementation.

The problem is that Schemes that are supposed to be portable would
have a hard time to implement SRFI 208. In general, I agree that it
doesn't matter for some specific SRFI, but it would matter should that
SRFI become included in R7RS-large.