Return to BSD News archive
Xref: sserve comp.os.386bsd.misc:2970 comp.os.linux.misc:20748
Path: sserve!newshost.anu.edu.au!harbinger.cc.monash.edu.au!msuinfo!agate!howland.reston.ans.net!europa.eng.gtefsd.com!MathWorks.Com!yeshua.marcam.com!charnel.ecst.csuchico.edu!xmission!u.cc.utah.edu!cs.weber.edu!terry
From: terry@cs.weber.edu (Terry Lambert)
Newsgroups: comp.os.386bsd.misc,comp.os.linux.misc
Subject: Re: Usefulness of BSD/Linux Source Knowledge
Date: 1 Aug 1994 01:35:19 GMT
Organization: Weber State University, Ogden, UT
Lines: 242
Message-ID: <31hjgn$spj@u.cc.utah.edu>
References: <1994Jul31.044558.12981@cs.brown.edu> <31grdf$o8g@u.cc.utah.edu> <1994Jul31.224245.24749@cs.brown.edu>
NNTP-Posting-Host: cs.weber.edu
In article <1994Jul31.224245.24749@cs.brown.edu> mhw@cs.brown.edu (Mark Weaver) writes:
] >] This is the vendors' way of competing with the free OSes.
Me>This is false. First of all, venders aren't "competing" in the same
Me>space as the free UNIX(R) clones; the clones leave a lot to be desired
Me>in terms of support, commercial apps, etc.
Me>
Me>At best, the clones are research platforms for computer scientists; at
Me>worst, they are simply "hackerware".
] I strongly disagree with you here. Linux in particular is starting
] to break into a large market of less technical computer users.
] Other than universities and large corporations, there seem to be
] an order of magnitude more people running Linux than any commercial
] unix, and that number is growing fast.
First of all, most UNIX venders I know (and I "know" more than a few), see
no need to "compete" with free OSes. At any point in time they have been
asked to comment and haven't simply declined comment, the competition model
they have cited involves support, code quality, pre-engineering, and
testing. All these are failings of the free OS phenomenon, which tends
to be a collection of small projects bundled together.
It was BSDI that triggered the UCB/BSDI legal problems by attempting to
truly commercialize the code (at which point it really *was* a competitive
threat). I feel strongly that BSDI is *still* the only real competitive
threat out there in the "UNIX(R) clone" category. Linux was nearly
legitimized in this regard by Novell, but fell short. The commercial
support services for Linux (and the lesser support services for BSD)
generally fall into the category of "after market"... that is, they do
little or nothing to "sell" the "product".
I can go into great gory detail on commercial failings in BSD, and into
lesser (if not less gory) detail on the failings of Linux if it were to
be considered a commercial product. Many commercial products have a
similar list of failings, but they are known and being addressed on the
basis of economic importance. Tried installing BSD or Linux on a disk
with > 1024 cylinders as the second or third OS on the drive lately?
Things are little improved over the initial 386BSD days... and you are
in immenent danger of trashing what ou do have on your disk already if
there are geometry translation and/or bad-sector forwarding issues.
] As far as commercial apps go, binary compatibility with commercial
] unixes like SCO will ease this problem quite a bit. Also, commercial
] apps are becoming available for Linux, because software companies
] realize that there's a growing market there. For products that
] already run on a commercial unix, compiling it on Linux represents
] a small investment with a potentially large reward. Furthermore,
] a decent free WYSIWYG word-processor alone could cause a huge influx
] of "ordinary people" to the free unix clones.
As former Senior Software Engineer at an applications software house
that supported their product on over 140 different UNIX platforms at
one time, I can say that the investment isn't that small for an "uncooked"
OS (not stable for over 6 months at a time, like a commercial OS); as
a matter of fact, as the company matured, it dropped support for the
OS's which did not make up 80% of the market -- and it was left with about
5 UNIX platforms and an up-front porting fee and quantity guarantee for
anything other than those platforms.
The only thing I think binary compatability will do for an OS is enlarge
the potential application base for all the OS's, commercial or otherwise,
on the basis of economies of scale. It may make that OS more attractive
on the basis of cost, but that's not as big a factor as some would like
to think.
As a case study, Oracle will not port to a new platform unless they are
given an up front fee. The last time I was glancingly involved in a
negotiation of this type, the fee was $1M.
It has been long held as a maxim in the industry that "software sells
hardware" and "software sells operating systems". That the driving force
and the majority of profit is in the applications, not in the platforms;
this is why I laugh when I see knife-fights between the cable and phone
companies over who will charge for wiring "the information superhighway";
it's not where the money is.
Consider a company that wants to run an Oracle database; they look around,
and they see "what platforms will run Oracle?". Say they see UnixWare,
Solaris, and Linux, all on Intel hardware -- what do they get? Well,
they don't get Linux, and the reason they don't get it is that Oracle
won't guarantee their product on Linux, and there is not a single legal
entity that will guarantee the ABI compatability that allows Linux to
run the Oracle software. So if it doesn't run, what do they do? Nothing;
they are screwed. Without an in-house team of hackers to fix all the
problems (and the time before the system must be online figured as part
of the cost of Linux), Linux is an unacceptable risk, unless they can
convince Oracle to run certification testing of their product on the
Linux platform: $1M, please.
This is not to say that these free systems will not be used as platforms
for vertical market and embedded systems applications; they probably will
be, but since the majority of the market is in the Fortune 1000 business
that Linux, BSD, et al. won't get, this is comparatively nothing.
] Remember also that "The Internet" is getting a lot of press, and
] non-unix OSes are very behind in this area. This also represents
] a reason for a normal computer user to switch to a free unix.
This depends on what portions of "The Internet" your hypothetical
pedestrian is interested in. The majority of these services are
available through cheap dialup services, most notable of late being
"America On Line", which basically functions as a painted crosswalk with
"yield for pedestrians" signs and mistimed traffic lights on "the
information superhighway" (that is, one big obstruction). Other means
of access, if a free soloution is needed, include Trumpet WinSock and
the various related client pieces for Mosaic, Telnet, FTP, News, etc..
All this while not hurting the ability of the end user to run "TurboTax".
] Calling the free unixes hackerware is meaningless to the masses.
If you ask "the masses", UNIX is a guy with a squeaky voice who guards
harems; until very recently, "the masses" thought CDROM was a Rap singer.
This posting was not addressed to "the masses". Before you analogize
me on the rate of change, note the distance in going from a personal,
portable storage technology to the software that may or may not be
ultimately responsible for drawing the pretty boxes on the screen is
a much larger leap. Common knowledge of UNIX is still relatively far
off, and even that assumes an effort on the part of venders to evangelize
it.
[ ... ]
] Although I agree with you that the vendors aren't CURRENTLY competing
] in the same space as the clones, many vendors may be looking at a
] bigger picture than you are Terry.
What is this bigger picture? The "unification of UNIX"? That died
with the purchase of a license-to-fragment by SunSoft. The "support"
and "someone to sue for non-performance" issues are the biggest ones
on the table right now. And the clone people aren't addressing either
of them (and can't, without becoming commercial entities -- basically
shaving their heads an painting targets on them).
[ ... ]
] >] Many say that the enemy is Microsoft, and this is true. But I also
] >] say the enemy are the vendors that try to sell for big bucks what
] >] we all enjoy for free.
[ ... note by me to the effect that the above statement is FUD ... ]
] My post wasn't intended to soothe the fears of managers, it was to
] present my perspective on Sun's forced and unpopular switch to
] Solaris, among other things.
I don't see how the above previous statement relates, nor how the above
statement refutes my claim that the UNIX venders aren't an enemy. I
agree that there needed to be a rallying cry toward unification, if only
to stop the sniping between the free OS groups, but pointing out a
common "enemy" is not the only method of accomplishing this.
[ ... ]
] The vendors are extremely interested in the research market, and
] that's why I'm upset. Even within my university, it's hard to
] stick to just free software because everyone is pushing System V
] and Motif all over the place. For my needs, there is very little
] (if any) value added in this. And getting Sun to do bug fixes is
] IMPOSSIBLE.
First of all, I don't think the venders are seriously interested in "the
research market"; it isn't big bucks, and it only tends to turn out
people like the members of CSRG, and that results in problems. An
indicator of "serious interest" would be the loaning of people to
academic research projects, the ability to communicate your results
(source licenses from all the major players, even educational licenses,
prevent this), and freeing of employees to do research not in the scope
of their job functions, and to make the results publically available.
The pushing of System V and Motif is an application level pushing;
that is, it is predominantly irrelevant to OS research. If you want
to do applications research, then it is relevent, depending on what
it is you are researching. The System V restriction, I think, is
more of a POSIX restriction, and as such, aids in the applicability
of the results to more platforms. The Motif restriction I don't
believe in, simply because the Motif 1.2 specifications are available
and fairly easy to implement to. I don't believe it is a requirement,
in any case.
[ ... suggestion that a University environment would be happier with
NetBSD ... ]
] If more universities did this, you can bet your ass it would take
] a large market share away from the commercial unixes.
I don't think it would; other than possibly MIT, very few Universities
are in the Fortune 1000. If what you say is true, then the UNIX market
is truly small, and they need to address the issue, not by acting
against their replacements at University sites, but by pursung their
real market more vigorously.
Every student that graduates with experience in a clone OS is one more
person likely to want to use a commercial UNIX over NT (where the only
choices will be commcial UNIX vs. NT, and a clone OS is not an option
approved by management).
] Furthermore, as the unix market grows to include computers at home,
] students are more likely to buy from whatever vendor they got used
] to at school.
This speaks more to the "Marketing should be giving it away to students"
issue than to the clone OS issue.
] Speaking of paranoid managers, you're a prime example of what these
] non-compete agreements do. The damage has already been done, before
] my post was ever conceived. I still have a copy of a post you made
] last September, when you made all the *BSD folks drool.
I still have it too, as well as the code it represents, plus even more
code than that that was created since then. I said at the time that I
didn't know "when or if permission would be forthcoming"; it's a long
involved process, and I won't give up. And I agree on the damage; the
Sun-style shared libraries slipped nearly a year on the basis of a 5 day
cleanup I was involved in when it suddenly became a potential conflict of
interest for me to release. I have to say that Paul's code is probably
better (I didn't deal with g++), so it may be an irrelevent point.
I'm still involved in the process, and I'll let you know what happens;
but the point is that only *some* of the paranoia is endemic; much of
it is induced by infamatory comments to which technically ignorant people
ascribe the worst possible meanings.
] Even today, if you released this code it would shake the entire
] free intel unix community. Some day, when I have a few months to
] spare, I plan on repeating some of the work you did.
That's the real shame: that's all it would take for most of the things
that haven't already be redone. Again, the issue is whether or not
venders are being frightened away from allowing their employees to
participate in research... employees capable of doing a better job than
me on a lot of things, and probably in a shorter time. Don't shackle
them by giving their managers unreasonable fears to hold on to.
Regards,
Terry Lambert
terry@cs.weber.edu
---
Any opinions in this posting are my own and not those of my present
or previous employers.