> The only downside to using the netif API that
I can see is that the
> target will be restricted to always providing the "Requested
IP". Might
> there be a situation where you need one netif using the "Requested
IP"
> feature, and one not using it, or different times when a netif might
> want to use it or not? I'm just worried that once you've turned
this
> option on at compile time you might have no way of not using it, whereas
> if it was a specific (new) function call you would have more control.
Well it is a deliberate decision of the user to create
a netif, give it the desired IP and send it to dhcp_start() OR create a
netif and send it to dhcp_start() directly. That was my idea of deciding. (Hoping that a netif
is created with all nulled fields, haven't checked that)
So if it is documented properly, this feature would
be completely uninvasive and still functional.
I'll come up with a patch and then we can decide if
it is good enough.