>> The major problem I see with arp(8) is that there is no way to
>> specify what interface an arp entry is to be created on.  [...]
>> arp -s 10.1.2.3 0:1:2:3:4:5 if de2
> I like the idea, but I don't like the syntax.

I'm not wedded to the syntax.  Someone else wanted -ifp, for similarity
to route(8)....

I don't much care what the interface is like.  I added it the way I did
because that looked minimally disturbing to the existing code, but if
someone feels like doing the work to fit it in some other way, I'm not
going to squawk.

> (It might, in fact, be most reasonable to *always* require the
> interface be specified in which case a keyword or flag are both
> unneeded.)

Perhaps.  But that would break backwards compatability.  It also seems
a little silly for the (probably large) class of machines that have
only one arp-capable interface.

>> When using this to create proxy entries, you also have to be able to
>> tell arp that it's supposed to be a proxy entry, because specifying
>> an interface bypasses the normal logic for determining whether the
>> entry is to be a proxy entry.
> Maybe the logic should be fixed instead of the program?

Perhaps, but as otherwise identical proxy and non-proxy entries are
both useful, you'd have to be able to override it in any case.  (Or at
least I think they are; I'm not 100% clear on how a proxy entry differs
functionally from a non-proxy entry.)

/~\ The ASCII				der Mouse
\ / Ribbon Campaign
 X  Against HTML	       mouse@rodents.montreal.qc.ca
/ \ Email!	     7D C8 61 52 5D E7 2D 39  4E F1 31 3E E8 B3 27 4B