amanda@intercon.uu.net (Amanda Walker) (07/06/89)
In article <2750@asylum.SF.CA.US>, sharon@asylum.SF.CA.US (Sharon Fisher) writes: > That's why I said "implementation." MacTCP is a > toolkit that allows you to build an implementation. [long, deep breath :-)] I don't particularly want to get into an argument about this, but I think you are defining your terms a little strangely. Here is how I define them and how I see MacTCP. MacTCP is a complete implementation of ARP, RARP, IP, UDP, and TCP, with partial implementations of RIP, ICMP, DNS, and BOOTP. Roughly speaking, it is an implementation of the transport layer and below. This corresponds to the level of service usally presented to applications programs, and as such, fits in very well with Apple stated system software strategy of providing system software but not being in the application software business. > As of a month before the announcement, Apple had planned to release their > own implementation of TCP/IP, which would include an NFS client. Apple may indeed provide implementations of TCP/IP services (such as NFS) that provide additional system-level services, but these are not part of MacTCP, and are not neccessary components of an implementation of TCP/IP. NFS in particular is a proprietary protocol that *uses* TCP/IP--it is not part of the TCP/IP protocol suite itself any more than, say, X is. It's ridiculous to expect every implementation of TCP/IP to include every protocol in the white books (if we did, nothing I've ever heard of would qualify :-)), and I think Apple made a very good decision in sticking to the ones that they did as system services. I must admit that the fact that I write higher-level TCP/IP software for the Mac may bias me in Apple's favor on this issue. -- Amanda Walker <amanda@intercon.uu.net> InterCon Systems Corporation
sharon@asylum.SF.CA.US (Sharon Fisher) (07/06/89)
In article <1151@intercon.UUCP> amanda@intercon.uu.net (Amanda Walker) writes: >In article <2750@asylum.SF.CA.US>, sharon@asylum.SF.CA.US (Sharon Fisher) >writes: >> That's why I said "implementation." MacTCP is a >> toolkit that allows you to build an implementation. > >[long, deep breath :-)] > >I don't particularly want to get into an argument about this, but I think >you Me either. :-) So let me go back to my original question, and I'lll try to phrase it diffeently. Apple had been planning to announce a product on June 12 that was based on stuff that CITI at the University of Michigan did for them. This product was to include an NFS client and TCP/IP. (The sname of the product hadn't been determined.) Does anybody know why this product was not announced.? Sorry about the typos; the terminal I'm using doesn't like Emacs very well. -- "Why should I let a loathsome little toad like you touch my breast when you haven't even read my books!" "Starstruck," by Elaine Lee