Message ID: 259043
Posted By: leclite
Posted On: 2005-04-24 05:42:00
Subject: Unixware == worthless.

> Unixware code before IBM began contributing to Linux?

Biff, there IS no usable code from Unixware, period. None, zero, zippo. Its a load of crap that is totally worthless and unusable in Linux.

If you actually knew anything about OS'es you'd realize this.

For the sake of argument, lets pretend that newSCO isn't a bunch of scam artists. Lets pretend newSCO is still oldSCO and actually owns Unixware IP.

Suprise me. Tell me what wonderful technologies existed in Unixware circa 1990 that Linux needed.

Nevermind, we all know you don't know jack shit about anything vaguely technical.

Message ID: 259057
Posted By: br3nsc
Posted On: 2005-04-24 09:14:00
Subject: Re: Unixware == worthless.

you know for us nontechies this is an interesting post.
when i think of the linux kernel,there is a lot listed for it,then the apps that go on top of it.
for unix i have no basic understanding.
what would be a good way for comparing the two since unix is proprietary?

Message ID: 259059
Posted By: manyhats23
Posted On: 2005-04-24 09:31:00
Subject: Re: Unixware == worthless.

when i think of the linux kernel,there is a lot listed for it,then the apps that go on top of it.
for unix i have no basic understanding.

Your understanding is correct. "Unix" is structured the same way. There is a Unix kernel, and then there are applications that run on top of the kernel. Many of the applications and utilities are common between Linux and the "closed" Unices, in that the same utilities, for example, exist in both.

The "closed" part of the proprietary Unices is mostly in the Unix kernels (for example, Solaris' kernel) rather than the applications. I can usually, with a little effort, take a open source application and compile it to run on either Linux or a closed Unix.


Message ID: 259061
Posted By: br3nsc
Posted On: 2005-04-24 09:48:00
Subject: Re: Unixware == worthless.

now a question that may sound silly?
how comparable would the code be.similar structure due to unix specifications i know.but would same terminology be similar thruout because of that?
word for word ,i have been told,is not possible?
so this is all very confusing.
i have looked at bsd and linux similarities but it doesnt *appear* the same to me,but i am not a coder?
it might use terms in both that are similar but they seem structured differently?
and they are to the specification i think called *posix*?

Message ID: 259062
Posted By: peragirn
Posted On: 2005-04-24 10:05:00
Subject: Re: Unixware == worthless.

>> Many of the applications and utilities are common between Linux and the "closed" Unices, in that the same utilities, for example, exist in both.

This is interesting because those utilities were designed on the orginial closed AT&T unix. Now that wonderful contract that SCO claims control over made sure that any third party program became public domain(aka vi, and the like) Many of those programs over the years were then created for/from the BSD's to replace the functions of an ever tightening grip by AT&T. Now the BSD's and most other Unices were POSIX compliant. Even though they are different code, they are designed to implent the same funcitonality. Hence why Linux when it came around in 1991 Linus could easily port bash and the other toys.

Think of the original unix as a Lego blocks. Great but tightly controled by Lego. Also you can't have just one block you need a large variety of blocks to build functional toys/buildings/ anything.

Over the yaers people saw things they wanted to build but Lego wouldn't help them out(old coroparte poilcy actually) So other people designed lego-compatible blocks, that could work with exsiting Lego's blocks. Now since the new products weren't called Lego, but lego-like. The purists will only use pure Lego, but many don't care so much about purity as functioinality. With Lego-like blocks filling in gaps, and even duplicating core functions.

You know I don't think that helps any. Oh well :-)

Message ID: 259065
Posted By: truth_in_government
Posted On: 2005-04-24 10:15:00
Subject: Re: Unixware == worthless.


"Today, the definition of UNIX takes the form of the worldwide Single UNIX Specification integrating X/Open Company's XPG4, IEEE's POSIX Standards and ISO C. Through continual evolution, the Single UNIX Specification is the defacto and dejure standard definition for the UNIX system application programming interfaces. As the owner of the UNIX trademark, The Open Group has separated the UNIX trademark from any actual code stream itself, thus allowing multiple implementations. Since the introduction of the Single UNIX Specification, there has been a single, open, consensus specification that defines the requirements for a conformant UNIX system."

UNIX is therefore defined through sets of published standards and must be certified for conformance to these standards to be capable of carrying the UNIX trademark as a branding. The definition is broad and specifies kernel, libraries, interfaces and commands and utilities. Caldera has maintained a "competing" specification that is not endorsed by the open group and can be considered depreciated to which its higher end UnixWare offering alledgedly conforms:


note, however, that these documents no longer define a compliant UNIX system and in fact are to be considered a partially compliant subset of an early version of the SUS with nonconformant extensions. UnixWare itself is not compliant or certified beyond the UNIX95 branding level, which effectively places it at SUS v.1 compliance. OpenServer, which was apparently at the time of the Santa Cruz Operation's acquisition of portions of Novell's USL assets intended to be integrated into a "merged product" is compliant only to earlier "Base" and UNIX93 standards. One of the stated goals of the Monterey project was to produce a UNIX98 branded oldSCO operating by incorporation of IBM licensed code, a goal that seems to have been as sucessful as previous "merged product" development into which Novell apparently invested ~$20MM.

Caldera's best probability with achieving UNIX98 certification for one of its products actually appears to have come out of its collaboration in development of Lasermoon Ltd's Linux-FT, a product which was intended to achieve Open Group certification at XPG4 (equivalent to OpenServer's level of compliance) in a Linux distribution.

Caldera did obtain (third hand) certain licensing rights within a range of source code which can be helpful in the development of a UNIX brand system, exactly what the extent of those rights are and any restrictions that might apply are deliberately obfuscated by them. The history of the "exercize" of those rights by Caldera subsequent to their acquisition is a more interesting and fruitful study to make than any attempt to analyze the thicket of partially revealed contractual agreements by which those rights (whatever they are) were obtained.

Message ID: 259071
Posted By: div_2n
Posted On: 2005-04-24 11:07:00
Subject: Re: Unixware == worthless.

Maybe I can shed some light on this since I have actually tinkered with Minix code and have peeked at Linux code.

First a definition (in case you don't know):

algorithm - a series of unambiguous step-by-step procedures for solving a problem in a finite number of steps

The reason I post that definition is that it is the "finite number of steps" part that really gets us to the heart of the reality of programming and, as an interesting side note, why software patents are just plain silly.

As algorithms increase in complexity, the various ways in which the solution can be accomplished increase. Maintaining efficiency, there is only so many ways you can concatenate or combine two strings together. That is a relatively simple task. Compare that to process scheduling, for example. There are a number of different ways you can manage which process gets the attention of the CPU. Some are more efficient than others, but there is more than one way to skin that cat.

How does this help answer your question? If you assign the same simple programming problem to beginner computer science students (combining two strings) then you will likely get some with almost identical code. Some may be exactly identical as they may independently choose the same variable names (StringA and StringB).

Assign the more complex task to graduate level students (rewrite the process scheduler for Minix) and you will most likely see some divergence. But unless a student develops a new algorithm, then they will all be selecting one existing algorithm or another. So inevitably, you will see some codea as similar EVEN THOUGH THEY WERE DEVELOPED INDEPENDENT (emphasis for SCO junkies).

Now that I have thoroughly bored you, I will finish answering your question. There are standards (POSIX) and common interfaces (think common error codes) that UNIX and Linux share due to them being based on the same technological underpinnings. Therefore, there will most likely be some coding overlap due to the law of averages and due to some things being independent implementations of the same simple idea.

It is my opinion that SCO finding these small overlaps is what got them started in the first place. It is them discovering too late in the game that these are, in fact, coincidental and not intentional (nor legally tortious) that has them in the nasty position they are now in. It is this reality that BIFF and others absolutely refuse to acknowledge. It is this fact that will be SCO's undoing.

Message ID: 259091
Posted By: ColonelZen
Posted On: 2005-04-24 12:58:00
Subject: Re: Unixware == worthless.

It's both harder and easier than a source code comparison. An operating system kernel or any program itself, in unix is a collection of named elements called functions. In c these functions are the atomic elements that the linker puts together into the program (or kernel). The compiler builds these from source code separating out global elements which are shared by all functions (and references to them) stack frame descriptions, and actual executable code.

The existance of other languages add complexity but because unix is built on C other compilers reduce what comes out of the language to this same model.

Now there are tools (nm, and ar to name two I remember using of the top of my head) which "tear apart" a program to tell you what they're made up of.

Now you can map the program into these functions map the globals and stack frames and disssemble the executables. At that point you have essentially source code in assembler ... and if you've used the same tools to tear apart the different programs, the conventions will be the same.

Now you can compare the dissasembled code the same way you would do a source code comparison, except now all the practices and bad habits of those dorky programmers are gone, there are no comments to skip, all the complexity and value hiding in headers and macro's has disappeared. It's just raw data to compare (well, I've simplified somewhat). Now you look for sequences of execuatble code which match referencing corresponding values in the global and stack frame areas. It's still complex but the "human element" has been almost completely filtered out. There are ways to hide from this kind of comparison as well, but the time and effort to do so is generally more trouble than rewriting the reference code.

As said this is complex but trust me, IBM knows all about how to do this!

Once done you can track correspondences in this comparison back to the source code and check where it came from.

My bet is that IBM long ago did this between SysV and Linux and *that's* when they decided to tell Darl to stuff it.

I'd also bet BIG bucks that a third party has been hired to do exactly this comparison between Linux and Legend and we'll hear a bit about it in the next two months!

-- TWZ

Message ID: 259094
Posted By: br3nsc
Posted On: 2005-04-24 13:08:00
Subject: Re: Unixware == worthless.

< It is my opinion that SCO finding these small overlaps is what got them started in the first place. It is them discovering too late in the game that these are, in fact, coincidental and not intentional (nor legally tortious) that has them in the nasty position they are now in. It is this reality that BIFF and others absolutely refuse to acknowledge. It is this fact that will be SCO's undoing.>

you are much more generous toward scox than i am.
i feel it was deliberate and not an accidental fault.they planned to reap big bucks from the FUD fear of legal suits for the licenses.
that sorta collapsed right i agree they are scrambling to find something,but i sometimes wonder if they are holding one little item that they feel may be enough to raise doubts and get this before a jury.
scox will lose against IBM patent claims and a bunch more claims but they only need one win of anything to keep sec and other regulators off their backs.

Message ID: 259095
Posted By: bill_beebe
Posted On: 2005-04-24 13:13:00
Subject: UnixWare was Quite Excellent - in 1995

>> Its a load of crap...

And you, sir, are full of shit yourself. I have my red Novell UnixWare box that I purchased in April 1995 for the princely sum of $249. I got in that red box two CDs loaded with UnixWare 2 (2.01); the Application Server and the SDK. It included all media (including 3.5" and 5.25" boot floppies) and a full suite of manuals. In looking at the SDK manual, it lists the following features:

* Software Packaging Tools
* C Compilation System (CCS) for Pentium
* Enhanced debugger (graphical UI)
* Library enhancements: support for MT/MP (multi-threading/multi-processor)
* Linker enhancements (shared libraries)
* C++ compiler
* Profiling tools (fprof)
* X GUI (X11R5)
* Desktop Management Development
* Motif Development (CDE and OSF/Motif 1.2.3)
* Kernel debugger (kdb)
* Windowing Korn Shell
* IHV (Independent Hardware Vendor) Development Kit
* Netware development - full client support
* Network development - full
* Metric Access Support

UnixWare was an evolution of the very successful Univel (the joint venture between AT&T and Novell) Unix that I had used before that (Consensys was a seller of the Univel base package with their own extensions, again for $249). I never had a complaint about UnixWare, and if I needed to I could fire it up today and work with it.

What did I have for Linux? I have very little dead-tree documentation from that period, but starting in 1996 I did have O'Reilly's "Running Linux" bundled with Redhat Linux 2.0, and I have official Redhat 4 with a publication date of August 1996 in the front cover.

Now. The reason I went with UnixWare at that time was because UnixWare was backed by Novell, a much larger company than any Linux distributor at the time, and because UnixWare was far more polished and mature than Linux was. I had tried Linux before RH4, via InfoMagic (remember those guys?). Here's what I had across a three year period before I got UnixWare:

June 1994:
Slackware 2.0, SLS 1.0.5, Debian 0.91 Beta, TAMU 1.0-A, Kernel sources up to 1.1.18, and a preliminary version of WINE.

December 1994:
Slackware 2.1, SES 1.06, Debian 0.91 Beta and kernel sources up to 1.1.72

March 1995:
Slackware 2.2, Bogus 1.01, SLS 1.06, Debian 0.91/3 and kernel sources up to 1.2.1

April 1996:
Red Hat 3.0.3 "Picasso" (ELF release), RH2.1 for DEC Alpha, Slackware 3.0, Debian 0.93R6 and Kenrel sources up to 1.2.13 and 1.3.88

In fact, in looking at other CD collections on my bookshelf, I see Red Hat Power Tools (RH4.1) from 1997, and more InfoMagic collections from 1998, 1999, and 2000. (InfoMagic finally went away in 2001 after publishing a very bad final collection.)

For personal use, Linux was great. For commercial use, Linux generally sucked unless it was in very specific niche areas on the edge of the network. I didn't start to take Linux seriously until I purchased my second "official boxed copy" of Redhat, version 5 in 1997 (Redhat had introduced kernel 2 with RH 4, but the overall distribution was still not-quite-there-yet for me).

You can flame SCOX/SCOG all you want. But UnixWare was and still is damn good. My positive feelings towards UnixWare under Novell is why I have such positive feelings towards SuSE under Novell these days. The quality of Novell SuSE of today compares quite favorably with the quality of UnixWare circa 1996.

>> If you actually knew anything about OS'es...

You should practice what you preach.

Message ID: 259109
Posted By: leclite
Posted On: 2005-04-24 14:32:00
Subject: UnixWare was not Quite Excellent

>> Its a load of crap...

Sorry. I stand by my statement. It was a heap of crap. I wasn't using Linux for serious work at the time, but i was using HP-UX and SunOS extensively.

UnixWare was a clunky toy in comparison.

There is a reason nobody liked Unix on x86 until Linux matured. Its the same reason there aws (and is) ABSOLUTELY nothing in Unixware that is worth porting to Linux. Period.

>> If you actually knew anything about OS'es...

> You should practice what you preach.

You seem to enjoy complaining when somebody flames when that somebody isn't you.

Allow me to retort:

Cram that attitude right up your ass.

Message ID: 259112
Posted By: raoulduke_esq
Posted On: 2005-04-24 15:19:00
Subject: Re: UnixWare was not Quite Excellent

<< There is a reason nobody liked Unix on x86 until Linux matured. Its the same reason there aws (and is).>>

I'll second bill_beebe's comments. I've been running FreeBSD since 1994, version 1.5.x. I'm sorry that the start-up that I put on the net at the time couldn't afford a fancy Sun, HP, or SGI box at the time. I had to scavange an old 486 as a mail, FTP & file server. Back then, we had a class C address space because it was the smalled block Netcom offered for busnesses. After I left, the idiots ran the company into the ground and let the IP space registration lapse.

FreeBSD was the natural step for someone that had been running & working on BSD and "cmunix" since 1982. And later, SunOs, Solaris, Ultrix & AIX. Once I could afford reasonable hardware (AMD 486), I loaded FBSD and I've been running it ever since then. Works fine for me.


Message ID: 259113
Posted By: div_2n
Posted On: 2005-04-24 15:23:00
Subject: Re: UnixWare was Quite Excellent - in 1

>> UnixWare was and still is damn good. <<

Umm, no. It was and still is good at very strictly defined tasks on strictly defined hardware and software.

There is a reason why the ISV support never really took off for it. They found their niche in healthcare, POS and maybe a few other areas and never really made a concerted effort to branch out until recently. They did that by using OSS software.

Message ID: 259127
Posted By: ColonelZen
Posted On: 2005-04-24 17:45:00
Subject: Re: UnixWare was Quite Excellent - in 1

Back in '97 or '98 I was the primary unix guy in a small to mid size biz. Our primary business system was supported on two, and only tww flavors of unix - AIX and Unixware.

We were upgrading and could (theoretically) have saved a few thou/yr in licensing costs on unixware.

I remember telling my boss, ~"no we don't want to do that, (even then I knew) SCO is pretty much a joke in the industry. AIX is well supported and in a couple years Linux will be supported. Stay with AIX for now."

All my predictions turned out correct.

I can't count the times I've toasted myself for my prescience.

-- TWZ

The texts of these Yahoo Message Board posts have been licensed for copying and distribution by the Yahoo Message Board users "leclite", "br3nsc", "manyhats23", "peragirn", "truth_in_government", "div_2n", "ColonelZen", "bill_beebe", "raoulduke_esq" under the following license: License: CCL Attribution-NonCommercial-ShareAlike v2.0.

Copyright 2005 Yahoo! SCOX. Messages are owned by the individual posters.