Path: utzoo!utgpu!water!watmath!clyde!bellcore!decvax!ucbvax!pasteur!ames!ll-xn!mit-eddie!apollo!mishkin From: mishkin@apollo.uucp (Nathaniel Mishkin) Newsgroups: comp.sys.apollo Subject: Re: SR10 and me Message-ID: <3a569abe.c366@apollo.uucp> Date: 17 Feb 88 14:43:00 GMT References: <8802090007.AA09431@ELI.CS.YALE.EDU> <1670@ssc-vax.UUCP> Reply-To: mishkin@apollo.UUCP (Nathaniel Mishkin) Organization: Apollo Computer, Chelmsford, MA Lines: 70 In article <1670@ssc-vax.UUCP> benoni@ssc-vax.UUCP (Charles L Ditzel) writes: >I too have not attended any of the non-disclosure meetings, however, >their have been some announcements from Apollo. > >Their Uniforum announcements (according to InfoWorld,etc) state that >they will be releasing System V.3 and Berkeley 4.3 version of their >operating system. Vague discussions that the OS also allows for >parallelism was also mentioned. > >I have no idea if *this* is SR10. >i think not. > >by the way SR10 is rumored to still have Aegis...i believe Aegis >is here to stay (on Apollos). DOMAIN/OS is sr10. Lest anyone got too pumped up by the DOMAIN/OS press release and is feeling deflated by this statement of fact, fear not, the changes made at sr10 are worthy of a naming refinement. Some things worth pointing out: Many people have gotten the impression, and still have the impression that Apollo's Unix support is an side interest for R&D. I won't deny that in the (now fairly distant) past, this was the case. I want to stress that now, this impression is definitely wrong. Unix is not an "optional product". Everyone gets Unix. We do Unix. All of R&D "does Unix". We changed an enormous number of things at sr10 to make DOMAIN work "like Unix" (where it wasn't before). We don't do things differently from Unix if there's no extraordinarily good reason to. We may not do Unix exactly the way other people do (i.e. by starting with a port of the BSD or ATT kernel code) and that may make some people unhappy, but we continue to believe that our Unix (a) is what the market wants/needs, and (b) is the best basis for extending Unix functionality. Now I know that there may not be many "Aegis" diehards out there in this audience, but for those that are, well, don't worry. DOMAIN/OS has some changes that may irk you, but we broke nothing that didn't have to be broken. We have compatibility support so that virtually all "old" programs will run, un-changed, un-recompiled, un-relinked. "Aegis" has not gone away. From my perspective, it's really sort of funny to think that way. To most people in R&D, "Aegis" simply means the OS kernel. GPR, the DM, the compilers, the "/com/sh" -- are not "Aegis"; they're libraries and programs. The tons of system interfaces we've defined -- they're not Aegis -- not in the sense that they can't be used from "Unix programs". (There aren't "Aegis programs" and "Unix programs". There are just programs written in one language or another that use some set of system services or another. The choice of language and set of system services may determine portability to other "Unix systems".) I like to think of things like this: We have some base set of functionality, expressed through procedural interfaces. Some of the procedures are implemented directly by the OS kernel proper, and some are implemented in user space (some in global libraries and some in private libraries). Some of the base functionality pays attention to whether the user/program wants System V behavior or BSD behavior. On top of this base functionality are a number of higher level access points. These are what users see: the various shells (csh, Bourne shell, "/com/sh") and the various standard commands (in "/usr/bin", "/com", "/etc", etc.) If you want to think that Aegis is "/com/sh" and the contents of "/com", and that somehow, that's a separate universe, feel free, but I think you get the picture that it's best to look at things as I've just described. Just trying to make sure people have a clear picture... -- -- Nat Mishkin Apollo Computer Inc. Chelmsford, MA {decvax,mit-eddie,umix}!apollo!mishkin