On Fri, Jul 25, 2003 at 11:53:45AM +0200, Waldemar Kornewald wrote: > The netstack would store a structure with its settings for each module > in the netsettings file: > > ethernet { > eth0 { > ipv4 { > ip 192.168.0.1 > netmask 255.255.255.0 > } > } > eth1 { > dhcp enabled > } > } > > dns { > ipv4 { > nameserver 192.168.0.10 > } > } I have a small suggestion here and also a question. Calling the setting "nameserver" is a bit ambiguous, maybe it should be called "dnsresolver", "dnscache" or something similar. Not even MS calls it nameserver anymore. Now for the question.... Why separate sections for ipv4 and ipv6 in the dns configuration? A given query, say for an MX record, will return the ipv4 address and also the ipv6 address in the same result set, regardless of wether the dns cache used is on a ipv4 or ipv6 address. The only exception to this rule are queries for AAAA records wich of course only return ipv6 addresses. Having separate sections for ipv4 and ipv6 doesnt make any sense to me. DNS Extensions to support IP version 6: http://www.faqs.org/rfcs/rfc1886.html > The DHCP client module would look for all interfaces that have "dhcp" > "enabled" (eth1) and begin getting the ip and a nameserver. Those > values would be saved as dynamic attributes. IMHO, having different dns caches on different interfaces only adds unnecessary complexity. How would the resolver library know wich cache to use for a given query? Since you dont know ahead of time what the returned result, ie ip address, will be i cant see any way that you'd know wich one to use. Cheers Lars Hansson