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

comments on address selection API



I have some comments on draft-chakrabarti-ipv6-addrselect-api-01.txt.

1. I'm puzzled by the following IPV6_SRC_PREFERENCES settings:

	IPV6_PREFER_SRC_TUNNEL  /* Prefer tunnel interface  as source  */
	IPV6_PREFER_SRC_NATIVE   /* Prefer native IPv6 source  addr */
	IPV6_PREFER_SRC_LARGESTSCOPE /* Prefer largest scope source */
	IPV6_PREFER_SRC_LOWERSCOPE  /* Prefer less than largest scope */

   What do they mean?  It would be nice if there was a description of
   what source address selection rules they affect and how. One point of
   confusion is that there are no source address selection rules that
   have to do with tunnel vs. native addresses, so I can't really guess
   what was intended here.  The second pair, largest vs. lower, are
   confusing.  There is a "Prefer appropriate scope" source address
   selection rule, but neither of these constants represents the
   "default" behavior for that rule.  What was intended here?  I'm
   guessing that this was an error and that these four settings shouldn't
   exist...

2. Some of the AI_ flags have strange names.  Here's what I mean:

   AI_PREFER_SRC_LARGESTSCOPE
   I'm guessing you meant AI_PREFER_DST_LARGESTSCOPE since this settings
   presumably alters the "Prefer smaller scope" destination address
   selection rule...?

   AI_PREFER_SRC_LOWERSCOPE
   Same.  Also, why is this "LOWER" instead of "LOWEST"?  Its buddy is
   "LARGEST".

   AI_PREFER_SRC_TUNNEL
   I'm also guessing that you mean AI_PREFER_DST_TUNNEL, as this setting
   presumably alters the "Prefer native transport" destination address
   selection rule...?

   AI_PREFER_SRC_NATIVE
   Same.

3. There is no explicit text on what an implementation should do if they
   don't support Mobile IPv6 or CGA addresses.  Should they define these
   socket option constants and ignore them if set, or should they return
   EINVAL if set?  Same question for the AI_* flags for getaddrinfo()...
   My inclination would be to define them, but ignore them when set on
   such implementations.

Thanks,
-Seb

--------------------------------------------------------------------
IETF IPng Working Group Mailing List
IPng Home Page:                      http://playground.sun.com/ipng
FTP archive:                      ftp://playground.sun.com/pub/ipng
Direct all administrative requests to majordomo@sunroof.eng.sun.com
--------------------------------------------------------------------