JINMEI Tatuya / 神明達哉 wrote:
As for the relevance to the addrselct API, I was confused about one point. New applications that support this API would probably test the availability of the flag first like this:hints.ai_flags = 0; #ifdef AI_NEWFLAG hints.ai_flags |= AI_NEWFLAG; #endif getaddrinfo(..., &hints, ..); Thus, the problem occurs only when the flag definition and the getaddrinfo() implementation are inconsistent (as Mark pointed out). It would be fair to say such inconsistency is a bug and ignore such a case. I think it's still helpful if the addrselect document assumes the consistency between the supported flag definitions and the getaddrinfo() implementation.
We can add this to the document. Erik -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www1.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
