Re: mboned: RE: suggestion for simplifying IGMPV3/MLDV2 to discard EXCLUDEmode

John Zwiebel wrote:
> Let's just leave the spec alone.
> Let implementers ignore the exclude for sources until someone
> figures out how to use it, then it won't have to be revisited.
> 
> FWIW: one can interpret the EXCLUDE to delete specific (S,G) entries.
> ie the opposite of INCLUDE, and since SSM doesn't deal with (*,G)
> entries...
> 
> And consider that it appears that each kernel is making different decisions
> on the type of message to send to join and leave the group, we might not
> want to mess around with this too much.
> 
> For example, Linux is sending out an "allow-new-sources" followed
> immediately by "change-to-include".

What Linux sends seems rather strange to me (I've looked into this
before). I feel it's not done the way it should be, but it seems it was
easier to implement it this way.

Stig

> 
> Windows XP, doesn't seem to support "block-old-sources" (although it
> could be
> the application I'm using).
> 
> NOW, it -could- be the test apps I'm using...
> 
> It is true that 3376 suggests that "MODE-IS-INCLUDE" is only sent in
> response to a IGMP query.  But is it also true that triggering a
> MODE-IS-INCLUDE
> will create the right state in the router.  (current-state-record)
> 
> It is also true that 3376 appendix A explains the need to differentiate
> 'current-state-records' from 'change-state-records'.
> 
> But, as I said, I'm not necessarily seeing implementations follow those
> rules.
> 
> 2cents