> I actually mentioned this exact problem several months ago.
> I have not looked at the internet headers for 1.3 in detail
> yet, but using "long" for an ipv4 address seemed to be all
> over the 1.2 sources, where it needs to be unsigned 32-bit int.

It's quite fine (bug-wise, at least 8-) to use 'long' for an IP
address in most situations.

However, it's _not_ fine to treat a (long *) like an "ip address
pointer".  That's what the code in question was doing.  There aren't
_too_ many instances of this in the source tree, because I debugged an
awful lot of them.  (For instance, rpc.bootparamd used to work for me.
I wonder if a change reintroduced a problem.)


chris