Re: [musl] mDNS in musl

[Rich Felker <dalias@libc.org>]

On Fri, Mar 08, 2024 at 04:31:02PM +0100, Markus Wichmann wrote:
> Am Wed, Mar 06, 2024 at 04:17:44PM -0800 schrieb David Schinazi:
> > I'm definitely supportive of the slippery slope argument, but I think
> > there's still a real line between mDNS and NetBIOS. mDNS uses a different
> > transport but lives inside the DNS namespace, whereas NetBIOS is really its
> > own thing - NetBIOS names aren't valid DNS hostnames.
> 
> One thing that came to me after thinking about it: This really puts a
> wrenchs in the works of the "just proxy it" argument.  Not for mDNS, but
> more generally for further possible backends for the list of hosts. If
> you want to support a backend that supports names that DNS can't, then
> proxying isn't a solution.

Who said we want support for names that DNS can't support?

> We may have to look into integrating nscd support into the search for
> host names. We already have it for passwd queries, after all. And then
> of course there are all the attendant questions:
> - Do we query nscd always or only with some config option?
> - Do we query nscd after DNS or vice-versa? Is this not a policy
>   question?
> - Is nscd failure authoritative?
> &c., &c.

The nscd wire format for hostnames is just bad. For example it can't
convey scope_id in IPv6 results, while musl's /etc/hosts
implementation can and does, making it actually useful for link-local
addresses. Maybe it could be extended, but I don't see any point in
worrying about this until there's a concrete need, which has not been
shown.

Rich