[babel] Summary of my comments about Sandy's talk

[Juliusz Chroboczek <[e-mail]>]

I'd be thrilled to work on BIER in Babel.  Babel claims that it is
a flexible protocol, and getting BIER to work in Babel would be an
opportunity to prove this claim.

I do however have some concerns about draft-zhang-bier-babel-extensions-00.
First, the encoding does not fit Babel's extension mechanism for two reasons:
it speaks about a transitive, optional sub-TLV, and Babel does not have
these bits.  Depending on the intended semantics, this extension should be
encoded either as a sub-TLV of the Update TLV, as a sub-TLV of the
Router-Id TLV, as a new TLV, or as an Update TLV with a new AE.  I believe
I have an opinion, but I'm not sure.

Second, this version of the draft only speaks about the packet format --
it doesn't describe the semantics of the data, it doesn't say what an
implementation actually does with the extra data.  I'm going to do some
reading about BIER (Pierre, fancy a beer at some point?), but I'd
appreciate some references.

(Side note: what requirements does BIER make on the topology?  Are
non-transitive (meshy) links fine?)

Finally, I think this work only makes sense if people are serious about
building a prototype.  I'll be glad to help.

-- Juliusz