> In this I would agree, except to say "watch out" if you get into OSPF/RIP, > because (according to our z/OS brethren) the OSPF protocol doesn't > recognize non-subnetted networks and subnets are required (RFC 3021's > 31-bit masks notwithstanding, I guess).
Hmph. Class D routes and non-subnetted networks are perfectly legal in OSPF; they're just not efficient use of routing table space and router resources compared to route summarization. Probably has more to do with implementations taking two or three times as much CPU if you permit large #s of discrete routes rather than summarizing them properly. > This is why the OSPF configuration in z/VM 5.2 no longer allows a mask of > 255.255.255.255. I'm not saying z/OS is necessarily correct, I'm just > pointing it out to avoid further confusion. (Yeah, right. Sure.) Bug, IMHO. Valid route, should be valid syntax. The fact you *can* shoot yourself in the head is not the tool's problem. Your gun, your foot.
