DHCPv6 in Fedora Core

[Pekka Savola]

Hi,

I noticed that DHCPv6 was added to Fedora Core a couple of months ago.  
Sorry for not bringing this up then.

I took a quick look at how it has been integrated in the initscripts, 
and the DHCPv6 src.rpm itself.

Two major comments and one justification why this needs to be done
now:

 1) the 'License' tag in the RPM is 'GPL', while the project has been
(partially) ported from KAME's 3-clause BSD license.  I.e., this is
critically incorrect because it's not even compatible with GPL, and
the tag should probably be 'BSD-like', because the advertising clause
is in there.

(Note: I think the source could use some updates, as well, because 
KAME's version has seen significant developer activity, while this has 
been recently quiet -- but this isn't a 'critical path' problem.)

 2) I'd like to ask what those major customers who have been asking
for DHCPv6 (as stated as the reason for including it) have been
requesting: have they asked for DHCPv6 support for 1) address
assignment, or DHCPv6 support for 2) configuring information (e.g.,
DNS or SIP servers) on their [v6-only] systems?

Following up from that, I think an important use for 2) would be to
run DHCPv6 in so-called 'stateless' mode, where no addresses are
obtained at all (for example, this is the only mode supported by KAME)
-- because the addresses can be obtained using stateless address
autoconfiguration as well.

Thus, I'd like to see if that could/should be integrated in the
initscripts.

So, I'd like to integrate both the 'information-only' ('-I' toggle)  
and 'full' DHCPv6 in the initscripts level.  One way to do that would
have 'DHCPV6C=' take two arguments: 'YES' and 'FULL'.  YES would call
dhcpv6c with -I toggle, full without it.

(FWIW, just for simplicity, I'd rename that to DHCPV6 or just DHCP6.)

Thoughts?

3) Q: Why this should be done now, rather late in the FC3 development 
process?  

A: this is the first release where this feature is introduced.  One
doesn't need to worry about backwards compat issues yet.  If this is
postponed to FC4, introducing this logical separation might become
quite challenging.