Re: Testing netapplet
On Tue, 07 Sep 2004 11:20:18 +0200, Matthew Garrett wrote:
> Yes. Netapplet is not designed to deal with the rare case of requiring
> multiple interfaces to be active. It is designed to deal with the common
> case of laptop users having multiple physical interfaces present, but
> only wanting to use one of them at any time. With hindsight, the
> description could be somewhat clearer in this respect.
Yes, netapplet's extended description should describe the situations in
which the package is useful if it turns out that it can't be general
purpose.
>> Changing the configuration of a wireless card with ifupdown is a matter of
>> ifdowning the interface and ifupping it as another logical interface with
>> different wireless options. My description of what netapplet should do
>> (in my previous message) fits this case.
>
> No. In a situation where you have multiple wireless networks available,
> you do not want to have to put an entry in /etc/network/interfaces for
> each of them. If I'm in the airport, I should not have to hack
> /etc/network/interfaces in order to ensure that I pick up the public
> access network rather than the closed internal network.
Well, _if_ you want to use ifupdown to adjust wireless parameters
then the way to do this is define different logical interfaces with
different wireless-* options and to bring up the wireless physical
interface as one of them.
My own preference is not to use ifupdown for this purpose. I actually use
waproamd which automatically detects available wireless networks and
configures the wireless adapter appropriately. I still define different
logical interfaces appropriate for the different networks I connect to
but I don't include wireless-* options in them.
I take it that you want netapplet to fill the same role as waproamd
except that the Wi-Fi configuration profile switching will be manual.
OK.
>> Instead of placing these restrictions on the administrator I suggest
>> using a special option in the logical interface definition which specifies
>> the name of the physical interface that gets created by that logical
>> interface's pre-up. netapplet can use this field in order to compile
>> a list of emergent physical interfaces and the logical interfaces
>> with which they are associated.
>
> Ok, that sounds like it makes a lot of sense. Have you any preferences
> for how this information should be presented?
Now that we are talking about this, I have noticed that the network-admin
component of gnome-system-tools writes a mysterious "name" option into
every logical interface definition. I don't know what it is there for; in
my case every such option has the value "Unknown interface type". I
suspect that its purpose may have something to do with what we are
discussing here. (I reported this phenomenon in #268334.) Even if it is,
I don't find the word 'name' very descriptive. How about:
iface foo ...
creates bar
...
If this option is introduced then ifup can be modified to enforce it.
That is, ifup would allow only bar to be brought up as foo. You
could still have multiple logical interfaces with the same physical
interface named on their "creates" lines.
--
Thomas Hood
Reply to: