Re: CRN is Outta Their Tree--DARL MCBRIDE?
From: Bill Vermillion (bv_at_wjv.comREMOVE)
Date: 11/21/03
- Next message: JamesDad: "Re: CRN is Outta Their Tree--DARL MCBRIDE?"
- Previous message: bernie: "No Root Disk Controller"
- In reply to: JamesDad: "Re: CRN is Outta Their Tree--DARL MCBRIDE?"
- Next in thread: Tony Lawrence: "Re: CRN is Outta Their Tree--DARL MCBRIDE?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Date: Fri, 21 Nov 2003 16:25:01 GMT
In article <204rrvce14vgndi767m3pokv6avdtpsldr@4ax.com>,
JamesDad <mewnewsARROGANCE@mailandnews.com> wrote:
>On Fri, 21 Nov 2003 03:45:04 GMT, bv@wjv.comREMOVE (Bill Vermillion)
>wrote:
>>>On Fri, 21 Nov 2003 00:14:49 +0000 (UTC), Tony Lawrence
>>><apl@shell01.TheWorld.com> wrote:
>>>>No. I'm saying for the Linux folks to stop trying to control their
>>>>software. If you love something, set it free..
>Bill, this is what I'm asking Tony to clarify. I don't know WHAT
>he means, and like certain other posters in this group, I'm
>not going to presume what he means and jump on him about it.
>Personally, I can't see what "control" could be relinquished or
>"setting free" could be other than releasing Linux as public
>domain. But that doesn't really make sense to me, so I'm asking
>Tony to clarify his statement. No need of arguing a point (if
>argument is needed) when it's not even understood.
Posting in a NG makes anything wide open for any and all to comment
upon. If you wish only to get a comment from one person use email.
That's been tradition since day one of the 'net.
>>>It's free to use, it's free to distribute, and it's free to
>>>modify (the only restriction here is that any modifications have
>>>to be shared with the community for the overall good of those
>>>using Linux and to prevent fragmentation into different flavors,
>>You don't count 200 variants different flavors? That's more than
>>Baskin Robins has.
>I'm skipping over the other stuff you posted, Bill, since it
>mostly deals with the issues I'm asking Tony to clarify. But this
>particular item caught my attention for response.
>I don't claim to be a Linux expert (I have yet to set up my first
>Linux box, but I have done a moderate amount of reading) nor do I
>claim to be a C/C++ programmer in ANY sense of the word (that's
>my brother--at least the C part). What I do understand is that
>the many Linux distributions (you say 200, I won't argue the
>issue one way or the other since I don't know) all focus on a
>common kernel lineage, controlled by Linus Torvalds personally.
First thing to do is get some different versions of Linux and try
them. You will see that not all Linux distributions look alike.
>As I understand it, the different distributions have different
>secondary files (libraries, tools, etc.) that one can always get
>if they need 'em. But all of them depend on the common kernel
>lineage. If two distros (let's say Slackware and Mandrake,
>justfortheheckofit) both state they have, say, the Linux 2.4
>kernel, and if they have everything else that's needed plugged in
>(such as particular libraries), software that runs on one should
>run on the other.
The problem is that people don't leave the kernel alone. If you
read the recent RH announcements you will see that they are taking
some things from the 2.6 kernel and putting them in 2.4. BINGO
- now RH's kernel is like no other in that part alone. And dont
count on having libraries that are alike. Back porting is not
something that is as easy to do as going forward. More than a few
Linux users are not happy with the RH approach in this area.
When you go to install a new RPM you often find you have to
install another RPM to fix dependencies the first needs, , and
that one may require yet another. Quickest way around that is to
compile the target program from sources. So many seem intent on
RPMs but the sources are there and >usually< dont have these
problems. Debain and Gentoo are often mentioned as the best in
this area - where things work and install as they are supposed to.
> For example, I just checked the OpenOffice.org
>page, and confirmed that there isn't one version for Slackware,
>one for Mandrake, one for Red Hat, one for TurboLinux and one for
>Caldera OpenLinux. There are two Linux versions there, but it's
>because of the CPUs (x86 vs PPC), not the distros.
That is true of many program, and I can run the Linux Netscape
on my FreeBSD system. But that is not true for others.
>My understanding is that such a level of compatibility doesn't
>exist among the different Unix flavors; a program that runs on
>IBM's AIX may or may not run on Sun Solaris or SGI's Irix.
That's about like complaining that you can't run you car on diesel,
gasoline, LPG, or hydrogen - whatever strikes you. AIX is
primarily the PPC, Solaris is primarily Sparc, and SGI is primarily
the MIPS. Then there was the other chip at HP - whose name I've
forgotten momentarily - that ran HP/UX. And what about A/UX - it
ran on 68000 CPUS. Most people have forgotten A/UX - that was
Apple's version of Unix. And when that went away they stayed pretty
much out of that area until OS/X.
> Programs that run on SCO OpenServer won't necessarily run on SCO
>UnixWare and vice-versa.
Most programs on OpenServer will run on UnixWare. Driver will
differ. SCO has been about the most backward compatible company to
hit the marketplace. Even if the binaries are a.out [from the days
of Xenix] or COFF from the days of SysV2, or ELF from today's
world, almost all the old programs will run. I have one site with
a small Xenix binary that was compiled [based on the date]
originally back in 1982. It's a.out.
When SGI went to the Irix 5.?? [forget the exact number] the
install looked for all COFF files on the system and removed them,
as the new ELF system would not run any COFF files. SCO made their
system backward compatible.
As to UW7 binaries running on OpenServer the reason is the same
that parents do not inherit things from their children.
OSR5 is a SysV3 based OS and UW7 is SysV4. So your argument doesn't
hold much water there.
My observation after using *n*x systems for so long is that one
reason that Linux succeeded - and it started on the iNTEL base - is
that virtually all of the iNTEL based *n*x systems disappeared.
First I suggest looking for the thread "LINUX is obsolete" that
Andy Tannenbaum [author of Minix] started back in early 1992.
The first reply was by Linus Torvalds. It's really long but
worthwhile to see the history. Use 'your friend' google.
You'll note that Tannenbaum wanted a teaching tool and took the LCD
of the 8086 PC. Some tried to port to newer architectures but
the licensing restraints put for by Prentice-Hall kept the model we
see in Linux from appearing. That was one reason Linus brought
forth into this world Linux.
The first ports were iNTEL based before it broadened.
But this was also the era when the *n*x for iNTEL started
disappearing one after the other and left SCO to be the only
SW only vendor left suppling *n*x for iNTEL based systems.
Intel shut down their distribution. [Intel did sell *n*x as a
software solution at one time].
Eastman Kodak stopped selling *n*x [many people don't remember they
also were in the *n*x business - when *n*x looked to be the way of
the future].
Dell gave up on their SysVR4. Many people have forgotten that
Dell also sold it's own branded version of Unix.
Esix moved from V3 to V4 and was bought by James Rivers and that
soon disappeared.
Interactive Unix also went bye bye. I think Sun bought them.
Microport Unix stopped being sold directly and they became strictly
a porting house.
All during this time all the people on non-SCO iNTEL based Unix
systems complained about SCO. This goes back to the inception of
SCO so this is nothing knew.
Since virtually every *i*x for Intel disappeared except SCO you see
there was a need for Linux.
BTW - all the other *n*x distributions did cost money. The
Microport was fairly cheap in comparison. I found my sales invoice
a year or so ago and it was just about $500 - for everything.
Even the almost free *n*x system - from people like Mark Williams
[did I get that right] and others - were in the $99 up.
If you bought a hard drive from Microport you got their SysV
installed on it free.
Many complained about SCOs sales model too - not including the
compilers or the text processing tools.
Then again SCO was restricted in two places - one by their AT&T
license which required additional fees for things outside the base
system, and their highly restrictive license from Microsoft - as
SCO started as both a porting house, with scads of cross-compiles
and assemblers for about 15 different CPUs that were in common use
in the early 1980s, and as a Xenix SW house. The also were the
company that was distributing Xenix for Apple's LISA.
MS license - until the EU threw part of it out a few years ago -
REQUIRED SCO to keep in Xenix backward compatibility and that
complicated a few things also.
In the end if you ran an iNTEL based processor you had two choices,
SCO or Linux.
>That's why SCO had to develop the "OpenServer Kernel Personality"
>for UnixWare. You can't just drop a missing library in and make
>it work.
And I can't put parts of a Toyata engine in my Honda either.
Unless I use a compatibility device - aka - a metal lathe.
>For example; somewhere in my collection of computer stuff is a
>copy of FoxPro 2.6 for Unix. One thing I recall is something
>about "SCO" on the floppies, meaning that if I have any desire
>to run it (once I ever find it), I have to have whatever type of
>SCO Unix it was designed to run on. I *do* have a media kit and
>binary license for a version of Sun Solaris, which is also Unix
>of course, also based on System V, but I seriously doubt that my
>copy of FoxPro will run on it. Am I correct? I think I've got the
>picture, but am willing to be corrected if I haven't.
Talk to Microsoft about that. They bought FoxPro. Bitch to them
about not running on other platforms.
I think you need to read up a bit on the history of *n*x and just
how we got to where we are today.
I'm pretty much OS neutral. If someone calls me to fix a problem
on some machine, I fix the problem if I can, and move on. Often I'm
must applying a band-aid to keep them running until they make their
conversion to something else. I get nightly email from client
backup software showing that the backup was successful from
OSR5, Linux and FreeBSD machines.
As to the ORR5 sites, One is looking into a web-based application
with their data distributed to many servers around the country.
I am anal enough that I don't like to see clients loose control
of their data - as the data is their lifeblood. And when the one
vendor said they'd guarantee 99.9% uptime and that is a guarantee
of not being down more than 90 hours per years. Their current SCO
system has been up 163 days - and it was restarted when power to
the building was out for many hours over a weekend while new power
distribution was being done. Prior to that the machine had been up
for over 450 days.
But I'm seeing many of the SMB market move away - but the larger
customers have such a large SW investment that they will stay on
SCO as their vendor supports the OS. When you pay $75K for SW and
about $10K+/year for support to keep your 8 figure per year
business going, that really isn't that much.
Bill
-- Bill Vermillion - bv @ wjv . com
- Next message: JamesDad: "Re: CRN is Outta Their Tree--DARL MCBRIDE?"
- Previous message: bernie: "No Root Disk Controller"
- In reply to: JamesDad: "Re: CRN is Outta Their Tree--DARL MCBRIDE?"
- Next in thread: Tony Lawrence: "Re: CRN is Outta Their Tree--DARL MCBRIDE?"
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
Relevant Pages
|
