[quagga-dev 10850] Re: [PATCH 2/4] zebra: quagga-re-onlink-support.patch

Christian Franke nobody at nowhere.ws
Thu Oct 24 21:45:19 BST 2013

On 10/24/2013 12:58 PM, James Li wrote:
> Our main concern is unnumbered interface support which requires onlink
> support. At the time of the development, this diff was the only one
> available and worked fine. I was under the impression that since it's in
> Quagga-RE branch, it's already accepted, just a matter of merging.

As far as I know, the quagga-re-onlink-support.patch didn't make it into
master as there was the feeling that it was not the right way to go.

That might have been differenent at the time it was merged into RE, but
by now we have a working implementation of NEXTHOP_IPV4_IFINDEX which
the quagga-re-onlink-support.patch seems to duplicate in large parts.

In general, my understanding is that we are trying to get features that
have been added to RE into master. However, RE is not treated as
"known-good". This means that every change which goes from RE to master
will be subject to review as to whether it is acceptable for master in
respect of design and quality of implementation.

> If this is not the case, could you comment w.r.t your patch and Denis's
> patch? Which one is chosen? Is there a conclusion? Also do you have
> plans to extend the ONLINK flag to IPC protocol?

As far as I can tell there isn't any unadressed criticism of my patch
around from which I would expect it to be merged eventually.

Hacking up a patch which allows to set the onlink flag via IPC should be
quite straightforward and useful for your unnumbered changes and the
babel changes. I guess I can just write it and send it to the mailing
list the next few days if peopl need it that badly. ;)


ps> One thing I would be interested in: What interface types (P2P, P2MP,
NBMA, Broadcast) are you doing unnumbered on?

More information about the Quagga-dev mailing list