On Mon, 2009-01-26 at 21:26 -0800, Erik Nordmark wrote: > Sebastien Roy wrote: > > > The potential for the implementation to get complex is a fair > > observation (although I don't think that preventing persistent change to > > a temporary object is hard to do), but I don't think that implementation > > complexity should necessarily result in a change in project scope or > > requirements. > > FWIW I'm not primarely concerned about implementation complexity, but > about the complexity visible to the admin. > With temporary objects all over the complex dependencies we have > (physical datalinks, vlans and vnics, aggregations, ipmp groups, > tunnels, ip addresses) the administator might find that in wanting to > make a particular object or property perstent s/he will get an error > message that some other object, that s/he has never heard about, is > temporary hence the operation fails. > > That doesn't sound like good usability.
No, and it's a valid concern. I'm not sure how prevalent that situation will be, however, if the default mode of operation for ipadm is persistent. > > I think we need to have a big picture in mind depicting the ultimate > > administrative model for networking. We can draw requirements out of > > that. Without this picture, we're making design decisions on a whim, > > which is uncomfortable. > > I take objection to you calling me that ;-) No name-calling intended. :-) -Seb
