Peter Memishian wrote: > > It is because that sometimes the libdladm API users needs to use a more > fine > > classification, and I feel it is really inconvenient to differentiate > these > > two. For example: > > > > Some wifi operation (show-wifi) needs to call dladm_walk_datalink_id to > walk > > all wifi datalinks, but some operations (say show-phys) needs to walk all > > physical data-links. > > > > Or some linkprop are wifi only, or some are physical links only. > > > > Therefore, I think it is more convenient to the user to export the WIFI, > > ETHER as (sub)classes. > > Can't that be accomplished by introducing a "media" field? If we mix > class and media now, it's going to only get worse in the future.
We could. But again, I think it is just inconvenient. For example, if I'd like to walk all aggregations, it needs to indicate both "class AGGR" and "media ETHER"? The latter does not even matter to the user. - Cathy
