> -----Original Message----- > From: Markus Stenberg [mailto:markus.stenb...@iki.fi] > Sent: 05 December 2014 13:44 > To: Jonathan Hansford > Cc: Markus Stenberg; homenet@ietf.org > Subject: Re: [homenet] Comments on draft-ietf-homenet-hncp-02 > > > On 3.12.2014, at 18.07, Jonathan Hansford <jonat...@hansfords.net> > wrote: > > I’ve just come to the homenet WG and have some questions and > comments for draft-ietf-homenet-hncp-02: > > Heya, and thanks for the review ;) > > > 1. I notice from the homenet charter the work is based on IPv6. Is > > there > anything in the HNCP protocol that precludes its use for IPv4? I am looking > for > a similar capability for use within multiple IPv4 broadcast domains with no > IPv6 in the mix at all. > > Do you need PA, SD, or just share some arbitrary TLVs? If last, you could use > some existing routing link-state protocol, or possibly <unnamed thing we are > generalizing out of HNCP>.
I'm currently exploring using Zeroconf capabilities to establish a single broadcast domain from scratch within an "IP island" (i.e. initially isolated from any larger internet). However, I know there will be occasions when there will be multiple broadcast domains within the island and will therefore need to be able to move from link-local to private addresses so they can be routed between the domains. Also need to perform service discovery. The one anomaly is that I need to do this prior to receiving any information from "the outside world" (e.g. allocated prefixes), hence the use of private addresses. The network needs to "discover" itself whilst it is an island, providing full service capability within that island prior to being connected to the internet. > > > 2. What is a node? I presume it is a router. > > Router/node is awfully mixed in hncp-02.. Router is a node, but not all nodes > are routers.. > > Thanks also for corrections - I will incorporate them in the next version if > they > are still applicable (it seems the next iteration will be drastic, given what > we > have now in github (public development and all) but more on that later > perhaps). > > (Terms ‘link’, ‘router’, ’node’ were horribly abused in the hncp-02. Sorry.) > > > 3. Section 1, 3rd paragraph: s/information how/information on how > > > > 4. Section 3, set of TLV data: s/The set of TLVs have/The set of TLVs > > has > > > > 5. Section 4.2 > > > > a. 1st paragraph: s/as purely as/purely as > > > > b. “TLVs MUST or MAY be present” – is that a decision that can be made > per node, per link or across the implementation? > > Good point, have to specify that (if this thing even exists). MUST be ones > obviously are always required, and MAY ones are local choice. (I think we got > rid of the MAY ones in the next iteration though, but I have to double check > to make sure.) > > Cheers, > > -Markus _______________________________________________ homenet mailing list homenet@ietf.org https://www.ietf.org/mailman/listinfo/homenet
