On Thu, Jul 13, 2000 at 01:13:07AM +0200, Christian Limpach wrote:

> May I suggest a different solution:  change the syntax of the file arp
> reads, i.e.
> hostname * [temp] [pub]
>    will use /etc/ethers to lookup the ether address for `hostname'

Hm. That still adds to the amount of administrative work to be done, as
you're still maintaining two files. The main impetus for my desired
change is simply maintaining one file. I guess if no one else wants
the -F flag, allowing wildcarding in a file destined for arp's use
would be a reasonable thing to consider, although it doesn't minimize
the administration.

> This way you don't have to add to the /etc/ethers syntax, you have fine
> grained control which entries from /etc/ethers get used and you don't add a
> new meaning to /etc/ethers.  Adding a new meaning would indeed be a really
> bad thing to do (cf /etc/shells).

How is it a bad thing? The format of /etc/ethers is not changing in
meaning or form, and if you're going to use "arp -F" you're hopefully
aware of what you're doing and why... I'm not absolutely desperate for
the change, but I'm curious as to why it's seen as such an evil thing.
Maybe I'm missing out on some common but funky use of /etc/ethers, but
it seems logical and natural.

In any event, I'm still reading to try to answer something jhawk posed
about how arp entries can make their way into the translation table.
It's possible that the concern that makes me want to nail all my MACs
down isn't valid, but I haven't seen that to my satisfaction yet. So
far my reading of RFC 826 is telling me the opposite, but I haven't
finished reading our if_arp.c yet either.

-- 
Mason Loring Bliss             (( "In the drowsy dark cave of the mind dreams
mason@acheron.middleboro.ma.us  )) build  their nest  with fragments  dropped
http://acheron.ne.mediaone.net ((  from day's caravan." - Rabindranath Tagore