On 10/06/2015 15:53, Masanori Ogino wrote:
> I came up with three solutions:
> 1. not defining unsupported constants (I guess it's non-conformant to the SRFI),
> 2. defining them as dummy values (0, #f, the `undefined` value, etc.), or
> 3. defining them as the platform's values and masking them in make-*-socket.
Preferred way would be #2 as fallback library of reference
implementation. It would be nice if the value can be used as a usual
flag so that users don't have to absorb platform differences.

NB: Sagittarius sets 0 for AI_V4MAPPED on FreeBSD.

> Which solution is permitted and/or recommended?
> I'm sorry if the problem has already been discussed.
As far as I remember, it wasn't. Good point :)

_/_/
Takashi Kato
E-mail: xxxxxx@ymail.com