Date: Tue, 23 Jul 2002 07:54:20 +0900 From: [EMAIL PROTECTED] Message-ID: <[EMAIL PROTECTED]>
| here are a little (incomplete) list of RFC2460-compliant | implementations that does not speak/understand HAO: This is a totally irrelevant argument. It simply doesn't matter which current implementations do or do not understand the option when it comes to deciding if it should be specified as a MUST or not. That question is relevant when designing the option, and its processing in the first place though. That is, someone has to decide what happens when a new option is seen at an old node that doesn't understand it. There the choices are "too bad, old node can't participate", "old node will just ignore this and do things this other way - not optimal, but works", or "this option will be useless with all those old nodes, there's no point having it at all" (perhaps others). Here, from what I have seen, that question isn't an issue - it seems (I believe) that the "old nodes will work sub-optimally" is the approach taken - that is, they don't get communications denied (which they would if this was a new header) nor is the option just considered worthless to have at all. On the other hand, when deciding whether new implementations MUST or SHOULD implement some option, the question is entirely related to the benefit vs the cost of the option processing, for nodes that can & would implement it (or might decide not to). What old code, shipped before the option was invented, might do is clearly irrelevant - that's not going to implement it, it cannot. Continuing to talk about the existing non HAO-supporting code base when deciding MUST vs SHOULD isn't helping anyone. If your point in there somewhere is that mobile-ip won't work with old nodes, and that is important enough to worry about, then the solution would have to be to redesign the protocol, not change one compliance word. But I don't think that's it - it seems to be more related to whether all those old implementations would be conforming to a standard that didn't exist when they were shipped. And the answer is that of course they won't be, and who cares about that anyway? kre ps: this message is not stating an opinion on whether MUST or SHOULD is the right label for this option, just about this one argument point. -------------------------------------------------------------------- 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 [EMAIL PROTECTED] --------------------------------------------------------------------