std-unix@longway.TIC.COM (Moderator, John S. Quarterman) (03/26/88)
From: uunet!wlbr.eaton.com!etn-rad!jru (John Unekis)
We recently had some SUN reps come to give a presentation about SPARC.
They were strongly suggesting that due to their relationship with AT&T
(that is AT&T will soon sell SPARC) it will soon be the case that if you
are not a SPARC machine you will not *really* be UNIX compatible. They
were talking about a coming binary standard, so that you could buy a
program written for UNIX and know that it would run on your UNIX machine
the same way you know that PC software will always run on your Intel/PC.
This binary standard would assumably be based on the SPARC instruction
set.
Is this stuff true or is it just marketing hype? Is UNIX really going to
become hardware dependent? What about all of us out here with our 680x0
or 80x86 or VAXen or whatever? Are we going to be second-class UNIX users,
unable to run the bulk of UNIX software? Can anybody out there clarify
this?
-----------------------------------------------------------------------
Any opinion expressed above is mine only. {ihnp4 or voder}!wlbr!etn-rad!jru
Volume-Number: Volume 13, Number 34hedrick@athos.rutgers.edu (Charles Hedrick) (03/27/88)
From: hedrick@athos.rutgers.edu (Charles Hedrick) uunet!wlbr.eaton.com!etn-rad!jru (John Unekis) says some Sun sales types claimed that because of the ABI (binary standard), if your machine isn't based on SPARC (the Sun RISC chip), it wouldn't *really* be Unix compatible. He asks whether this is just hype. This issue has been discussed in the trade press a lot. A number of vendors are quite upset at the Sun/ATT cooperative deal for precisely this reason. They want Unix to continue to be open. ATT has been reassuring people that Unix will indeed continue to be open. ATT is helping create binary standards for SPARC and 386. Other vendors with serious interest in Unix are sponsoring similar standards for their chips, e.g. Motorola for the 680x0. The other major issue is whether other vendors will get advance copies of new releases soon enough that they can come out with their own ports, and not have Sun and ATT always be a year ahead of them. ATT is now saying that they will continue to do this, and that their own field test and other overheads are long enough that other vendors should be able to come out with releases at roughly the same time as the ones that ATT produces. It will be interesting to see how things develop in this area. I'm sure both ATT and Sun have toyed from time to time with the idea of closing Unix (whatever that means), but I don't see any way that that could really happen. [ Please pardon the paragraphing, Charles: I couldn't read it as one big block of text. -mod ] Volume-Number: Volume 13, Number 35
jmb@patton.SGI.COM (Jim Barton) (03/28/88)
From: jmb@patton.SGI.COM (Jim Barton) In article <147@longway.TIC.COM>: > From: uunet!wlbr.eaton.com!etn-rad!jru (John Unekis) > > We recently had some SUN reps come to give a presentation about SPARC. > They were strongly suggesting that due to their relationship with AT&T > (that is AT&T will soon sell SPARC) it will soon be the case that if you > are not a SPARC machine you will not *really* be UNIX compatible. They > were talking about a coming binary standard, so that you could buy a > program written for UNIX and know that it would run on your UNIX machine > the same way you know that PC software will always run on your Intel/PC. > This binary standard would assumably be based on the SPARC instruction > set. > Is this stuff true or is it just marketing hype? Is UNIX really going to > become hardware dependent? What about all of us out here with our 680x0 > or 80x86 or VAXen or whatever? Are we going to be second-class UNIX users, > unable to run the bulk of UNIX software? Can anybody out there clarify > this? > > ----------------------------------------------------------------------- > Any opinion expressed above is mine only. {ihnp4 or voder}!wlbr!etn-rad!jru > > Volume-Number: Volume 13, Number 34 Before one bites too hard on this hype, you should remember that Sun will soon be selling 80386 based boxes. How can they sell said systems if they aren't standard UNIX? What about all those Sun-3's out there, do they suddenly become useless to everybody? You should also check out AT&T's side of the story, which is different than Sun's. As far as they are concerned, there will be a binary standard for UNIX for every type of processor. Thus, there will be a SPARC ABI, but it will be just one out of the collection, including 386 ABI (signed with Microsoft just last year), 68K ABI (signed with Motorola very recently) and others on the way. Sun would like to be the IBM of the workstation industry. Backward technology, marketing hype, trashing competitors and the like. AT&T doesn't even claim that SPARC is state-of-the-art, or best performance anymore either. Just that it is standard (Remember, what an "open standard" is by the new rules: I design something I like, publish barely enough information for somebody else to >expensively< duplicate it, and proclaim it a standard. I don't need your concurrence or opinions, nor do I wan't them. After all, you might have a better idea ...). Finally, think about a sales organization that would send their salepeople out with such a story. You are obviously concerned by it, and see the flaws. They trash their own current and future sales of 386 and 68K boxes to scare you into buying SPARC boxes. Would you buy a computer from these people? [ Ok, folks: this is a technical newsgroup, for technical discussions. It's hard to talk about standards without talking about politics, but let's try to avoid casting aspersions on companies or people. -mod ] -- Jim Barton Silicon Graphics Computing Systems "UNIX: Live Free Or Die!" jmb@sgi.sgi.com, sgi!jmb@decwrl.dec.com, ...{decwrl,sun}!sgi!jmb -- Volume-Number: Volume 13, Number 37
std-unix@longway.TIC.COM (Moderator, John S. Quarterman) (03/29/88)
From: uunet!wlbr.eaton.com!etn-rad!jru (John Unekis) In article <150@longway.TIC.COM>: >From: jmb@patton.SGI.COM (Jim Barton) > >In article <147@longway.TIC.COM>: >> From: uunet!wlbr.eaton.com!etn-rad!jru (John Unekis) >> >> We recently had some SUN reps come to give a presentation about SPARC. >> They were strongly suggesting that due to their relationship with AT&T >> (that is AT&T will soon sell SPARC) it will soon be the case that if you >> are not a SPARC machine you will not *really* be UNIX compatible. They > >Finally, think about a sales organization that would send their salepeople >out with such a story. You are obviously concerned by it, and see the ..... I realized after posting this article that it sounded critical of SUN, and might result in some flames in their direction. That was not at all my intention. SUN is an excellent company, and thay have always supported us well. My question was actually more concerned with the future of UNIX as an open standard. Obviously keeping UNIX open is a double edged sword for AT&T, they gain great credit as the source of the most widely accepted non-hardware dependent OS, but does it really benefit the sales of their machines? If AT&T were ever to pick a hardware standard as the basis for a product dependent UNIX, the SPARC would be an excellent choice. The real question, I suppose, is can an open standard like UNIX really survive in today's feircely competitive marketplace? ----------------------------------------------------------------------- Any opinions expressed above are mine only. {ihnp4 or voder}!wlbr!etn-rad!jru Volume-Number: Volume 13, Number 38
jmb@patton.SGI.COM (Jim Barton) (03/30/88)
From: jmb@patton.SGI.COM (Jim Barton) Having been justifiably chastised by the moderator, I'll keep my personal opinions to myself, and stick with the issue at hand, which is "standard" UNIX. In article <151@longway.TIC.COM>: > From: uunet!wlbr.eaton.com!etn-rad!jru (John Unekis) ... > > ........ My question was actually more concerned with the future of UNIX > as an open standard. Obviously keeping UNIX open is a double edged sword > for AT&T, they gain great credit as the source of the most widely accepted > non-hardware dependent OS, but does it really benefit the sales of their > machines? If AT&T were ever to pick a hardware standard as the basis for > a product dependent UNIX, the SPARC would be an excellent choice. The > real question, I suppose, is can an open standard like UNIX really survive > in today's feircely competitive marketplace? > AT&T has some real problems in attempting to manage UNIX, and looks to Sun to solve some of them. There are several points to consider about the current mess: 1) AT&T >owns< UNIX, and don't you forget it. Sun doesn't own it, Berkeley doesn't, and CMU doesn't. They feel very strongly about it. 2) Cassoni (President of AT&T Data Systems) believes that a hardware platform is necessary to the success of UNIX, similarly to the IBM PC. I obviously don't. The PC is basically hardware, and thus a different beast than UNIX. By making UNIX hardware specific, you destroy it. 3) AT&T licenses UNIX to a large number of people in various ways. Were they to take action that would seriously harm a large number of licensees (especially the big ones) they would be facing an anti-trust suit so fast your head would spin. 4) AT&T and Sun are attempting to define a new standard by a simple existence proof, the infamous "open" standard. This group has carried much discussion about P1003 and other efforts; those are a "standard". AT&T and Sun wish to define a standard suitable for their own uses without going through the hassle of getting everybody to agree - thus bypassing most of the technical community. Thus, they want an IBM style "standard". This is the current, openly announced, plan for V.4. 5) The "open" standard and promises to licensees are efforts by AT&T to pacify competitors. It is clearly to AT&T's advantage to have a "standard" which looks open on the surface (i.e., everybody >could< duplicate it without AT&T source code), but is really as closed as possible (you really have to buy AT&T source and hardware to realize it). This is true today. Consider RFS, which is in SVID Issue 3, but is un-implementable from the SVID as specified. Somehow, though, this is a "standard". 6) You've obviously been hiding under a rock for the past six months. A large number of UNIX licensees have complained directly to AT&T about the AT&T/Sun deal and the pronouncements by Sun salespeople. AT&T has come out with many assurances that they will take care of everybody. Rumoured deals are in the making to work on competing versions of UNIX (though obviously >not< with that name). I'd like to make one last point. A "competing" standard arose with Berkeley UNIX because of the legally restricted support that AT&T had to give originally and because of the anti-UCB sentiment back in New Jersey. By attempting to force one standard without the technical and business input of the UNIX licensees, AT&T engenders an environment where another competing standard can rise and flourish. After all, V.4 will have Sun's favorite enhancements, but it won't have mine, or yours, and it may break what I do have. The heavy-handedness of the AT&T/Sun actions may do more to trash any emerging UNIX standard than to promote it. -- Jim Barton Silicon Graphics Computing Systems "UNIX: Live Free Or Die!" jmb@sgi.sgi.com, sgi!jmb@decwrl.dec.com, ...{decwrl,sun}!sgi!jmb -- Volume-Number: Volume 13, Number 39
guy@Sun.COM (Guy Harris) (04/05/88)
From: guy@Sun.COM (Guy Harris) We recently had some SUN reps come to give a presentation about SPARC. They were strongly suggesting that due to their relationship with AT&T (that is AT&T will soon sell SPARC) it will soon be the case that if you are not a SPARC machine you will not *really* be UNIX compatible. If they were truly suggesting this, it merely indicates that Sun (not all caps, PLEASE! - it's not an acronym for Stanford University Network when used to refer to Sun Microsystems, Inc.) salespeople are as 1) uninformed about reality and/or 2) over-eager to sell their product as any other vendor's salespeople. If you are not a SPARC machine, you will be 100% UNIX-compatible as long as you pass whatever validation suites the person asking you "are you UNIX-compatible" wants to use, such as the SVVS. The SVVS won't test whether a machine is a SPARC or not. There may be tests to see whether a SPARC-based machine conforms to the SPARC Applications Binary Interface, but if there is there'll probably be similar test for the 80386-and-up ABI, and the 68020-and-up ABI, and.... They were talking about a coming binary standard, so that you could buy a program written for UNIX and know that it would run on your UNIX machine the same way you know that PC software will always run on your Intel/PC. This binary standard would assumably be based on the SPARC instruction set. There are several binary standards arriving on the market. For each one of them, you could buy a program written for UNIX *and* compiled for the architecture in question and know that it will run on UNIX machines using that architecture. There are, for example, standards coming out for the 80386-and-up family and for the 68020-and-up family. There will be one for SPARC. There may well be others coming out for various non-"in-house-only" chips (MIPS, Motorola 88000, etc.). Is this stuff true or is it just marketing hype? Is UNIX really going to become hardware dependent? What about all of us out here with our 680x0 or 80x86 or VAXen or whatever? Are we going to be second-class UNIX users, unable to run the bulk of UNIX software? Can anybody out there clarify this? They won't be able to run SPARC binaries, but SPARC-based machines won't be able to run 68K binaries, either (unless somebody does emulation product - no comment on whether such a thing exists, or is planned, or...). It may be that some chips will end up having more UNIX software built for them than others, and that the others will end up being "second-class UNIX users", unable to run a lot of UNIX software. Such are the vagaries of the marketplace; there's no guarantee that SPARC will end up in one or the other of those sets. Obviously, we'd like it to end up in the former set, just as Motorola would like the 68020-and-up and 88K to end up there, and Intel would like the 80386-and-up to end up there, and MIPS would like the MIPS chips to end up there, and.... Volume-Number: Volume 13, Number 43