On 25/03/09 11:55 AM, Sowmini.Varadhan at Sun.COM wrote:
> On (03/25/09 09:05), Darren Reed wrote:
>
>> Which is good... and if I refer back to the CLI you presented earlier:
>>
>> # ipadm {create,modify}-interface [-t] [-f {inet, inet6}] \
>> [-if6_intf_id=<IPv6 Interface ID>] \
>> [-O <interface sub-options>] <interface>
>>
>> Then the concept you're pushing of an interface being a "thing"
>> that can have multiple address families requires that the "-f"
>> option be dropped completely. Another alternative might be to
>> make the family a mandatory option before "-t", without the "-f".
>>
>
> If you drop the -f completely, how is the ipadm user going to mandate
> that (s)he will have "no ipv6 interfaces", or "no ipv4 interfaces"?
>
> In other words, yes, we can have all of the following:
> - make ipadm present the (illusory, today) BSD-like impression that there
> is just one interface, that can be used flexibly for ipv4 and ipv6.
> - under the covers, we can continue having the ipv4/ipv6 ill model, where
> both are created by "ipadm create-interface <foo>".
> - ifconfig will still report ipv4 and ipv6 as separate ills
> But what if the adminstrator really does not want to create an
> ipv4 capable interface, and only wants the ipv6 capability? If we want
> to provide this feature (as an option) thorugh ipadm, we'd need the
> -f flag.
>
Hmm, I might have been a bit obscure in how I phrased one of those
suggestions...
Can the command line work like this:
ipadm {create,modify}-interface <inet|inet6> [-t] ...
Is there ever a situation where neither inet nor inet6 will be there?
Darren
-------------- next part --------------
An HTML attachment was scrubbed...
URL:
<http://mail.opensolaris.org/pipermail/brussels-dev/attachments/20090325/c93c2121/attachment.html>
