Post a Comment
Isn't UnixWare a SCO product? Why would anyone touch it then? SCO is downright evil!
Yes, it is a SCO product, but on the same time it is an operating system. And OSNews is about Operating systems (and not just OSes), but not politics and laywers.
Please keep the discussion here about the UnixWare product and not about the ethos of the SCO management or their lawsuits. These are off topic here. Keep your anger for SCO articles related to the lawsuit.
Destruction is always easier than construction.
I too believe that SCO is dead wrong. But I believe that a product review should also be kind of objective.
Like Eugenia said: OSNews is about OS'es and Unixware is one of them...
SCO has never been a major unix player, they have been known as a unix for economy hardware for quite awhile. i dont know, i wouldnt want to install an os whose main claim to faim is its ability to run on mcdonalds cash registers on my server....
If we open this up to a discussion of why SCO sucks, then the discussion of the OS and the review will be drowned out. Let's just assume for the sake of this forum that SCO is bad, very very bad, so don't bother mentioning it.
Now that you have the cds maybe you can do a test of FreeBSD SCO compatibility and let us know the results. That would be an interesting follow up imho.
I've been running SCO Openserver as my server OS since 1996. The company I currently work for has been using it for years longer. I've got ADSL and a proxy server and a firewall to protect my internal LAN from the bad guys on the internet. I've never had an exploit. I use the SCO box for file serving to my internal LAN pc's. I've never had a problem with it. It's been rock solid and reliable. Likewise the company experience. Santa Cruz Organization (the original SCO) developed UnixWare as the upgrade path from Openserver. So I don't doubt that it's as enterprise ready as any other *nix out there. Even if Santa Cruz Organization was not a big player in the enterprise market, it still produced excellent, enterprise software. I will not, however, be upgrading to any products offered by the current company that calls itself SCO. When the time comes to replace my current server, I will be going either with Linux or OSX.
What makes no since to me is there pricing matrix. If they wish to grow there user base they should at least put the product in a competitive price bracket with similar products in the market. Also, with the need to license other software components, the product becomes even more unattractive from a business perspective.
My guess is that they price so high because they have a very small user base and they need to increase the per-unit pricing to make money off the product.
Back in 2000 I was tooling around getting Linux PPC installed on my old PowerComputing box. I ended up with a horrible mess requiring an exact mix of hard drives (I had three, but one was supposed to be optional) in order to boot.
Then I took a course in (SCO) Unix. SCO becase this was in SC (Santa Cruz, California) and they could get some of the nice people from SCO to teach classes for cheap.
My impressions mirror the reviewer's on most points. It was pretty bad, and the drop rate of students and working installations was about equal (and high). Not something I wanted to use on a daily basis.
One thing did stick out to me: playing with a real Unix hammered home how horrible Linux's disk management was. Linux was trying to give SCSI drives drive letters? Linux's ties to cheap hardware still seems to be a downfall for it, just as Unix's ties to expensive hardware are for it.
Sorry, not too many other thoughts (sorry), other than they were pretty impressed with their config prog.
I was quite intrigued by how this news begins... Very good! Very refreshing! Quite provoking. A good move. Not that I care about SCO, but this news item was not like all the rest. Good.
But I've long been curious about performance benchmarks, its long been my understanding that Linux (and most assuredly *BSD) has been spanking it for a while now. What kind of threading support does it have? Is it as poor as LinuxThreads or is it decently scalable? The LPK does seem rather impressive, and perhaps the only way you could run apps on it in the future. Which naturally begs the question, why run Linux apps on a tremendously expensive, backward, and potentially slow(er) o.s. when you can get the real deal? Oh, and I hadn't yet figured out the different between UnixWare and OpenServer. As an aside my old job had an old SCO Unix box (do not recall if it was OpenServer or UnixWare) that ran the database (it was a small vocational school) and along with our Red Hat Linux box that handled email, firewall, proxy, and www serving it was the most stable box there. The Windows servers always demanded our attention for something, but never those two. :-)
Very nice, intelligent, balanced review.
One point - I think you will find that most companies who are running UnixWare are doing so for some very industry specific vertical market applications and not for generic server roles like Samba or Apache or Oracle or things like that. I don't know this for sure, but I think that it is very likely and would explain why Tony had so many problems just doing everyday common things with UnixWare that he could take for-granted with other Unices.
"Oh, and I hadn't yet figured out the different between UnixWare and OpenServer."
SCO OpenServer still uses System V kernel version 3.2 while
Unixware is on SVR5.
One thing did stick out to me: playing with a real Unix hammered home how horrible Linux's disk management was. Linux was trying to give SCSI drives drive letters? Linux's ties to cheap hardware still seems to be a downfall for it, just as Unix's ties to expensive hardware are for it.
Maybe I haven't had enough exprience with other flavors of *nix, however I'm not quite sure what u mean by this statement. I've used Linux, FreeBSD, and Solaris, and all have a similar /dev/*** reference for harddrives. In linux's case you would mount /dev/hda as / or /boot or whatever you decided. I don't get where you say "drive letters". Did you maybe use Suse and see that it mounted your windows drives as /windows/c or /windows/d? I just don't get where your coming from.
Again, I could be completey wrong about something, but as far as I know in linux scsi drivers are listed as /dev/scd0, /dev/scd1, etc, and in other unixes you just replace the scd* w/ some other abreivation.
There is no reason to consider SCO UNIX evil. Before it changed management and became a company of lawyers, there were a lot of very good engineers at SCO. SCO UNIX is from that SCO, not the current one.
Not that I use any SCO product, or that I would consider it, but this really was a nice to read review. By the looks of it, SCO has a nice IP in its hands with its "Personality" stuff. If they could leverage this knowledge and experience for other *nix like OS'es they might actually have a viable product to compete in a market which is still wide open (virtual computing, compatibility layers). As it is now though, SCO seems to be more interrested in making money from litigation than from competing in an honest market.
Anyway, thumbs up for this review, keep up the good work.
PS. What worries me is that "Enterprise Linux" though cheaper than SCO's licenses still is rather steep priced. Where did this come from all of a sudden? Does the little proprietary addon (read: installer) and more or less professional support really make up for the rather large price difference between free and $1500? Personally, I think not. It's exploiting the work of the community which basically built these products. But alas, with open source also comes the freedom to choose. We're fortunate for that. All we need now is a company like Oracle to also certify for a free linux distribution. Debian, of course, being the most viable candidate. Does anyone have any idea how to convince a company of that size to consider such an action?
I was able to get a COBOL app and runtime compiler running under FreeBSD 4.x, it was originally running on dozens of OpenServer 5.0.4 boxen. The app was for POS, so I had to change some of the settings for printer mapping, term settings were also modded. This got them outta of having to hassle with SCO licensing. Their corp IT "mothership" decided to revamp the whole thing to Windows, thus making it a learning experience more than anything else. I have ran UnixWare (test drive), and OpenServer (production). I no longer offer support for SCO products, and have not found any need to recommend UnixWare nor OpenServer. There are better solutions out there, imho.
-iGZo
re skaeight
The /dev/sda is what the original poster is refering to. Most "enterpise" (term used in contex) use /dev/controller id/ drive id construct. I think there is a way to do this in linux but I don't remember right now. For the record the reason is large drive systems.
1 primary controller + 1 backup controller per drive box
3 drive boxes each with 800 GBi of storage
Veritas Raid 10
Lots of drives etc, the bus id and drive id is a lot easier to manage nd to trace disk io bottlenecks to specfic drive.
Donaldson
The LKP doesn't work like VMware at all; rather, it's much closer to the Linux binary compatibility in *BSD. Rather than providing a virtual machine, it simply supplies Linux binaries with a different system call table to the native one, so that the syscalls in the LKP are used. It's entirely unsuprising that speed tests show no difference between LKP and native Unixware; nothing is virtualised or emulated. If there's no syscalls being made (which for RSA generation, there won't be many - it's all calculation), then there's no overhead; if there are syscalls being made, you'll probably find the overhead is neglible.
A better speed comparison would be Linux 2.4.x vs Unixware's LKP on the same hardware. Probably be interesting to compare it with the speed of FreeBSD doing the same.
Another speed test which could be worth doing is a relatively carefully crafted 'grep' on a dictionary file, or something similar. Lots of file I/O, so lots of syscalls.
Turn devfs on:
/dev/ide/hostN/busN/targetN/lunN/{disc,partN}
/dev/scsi/hostN/busN/targetN/lunN/{disc,partN}
Yep, very gutsy, Tony. Especially considering UnixWare doesn't seem all that great and is apparently overpriced as well. It must be embarrassing for them to own Unix and be outdone by SGI, IBM and Sun.
A lot of what "Enterprise Linux" gives you isn't so much code, but work done in the trenches.
When OpenSSH came out with vulnerabilities, if you were an Enterprise customer, you had access to a patched OpenSSH that you can download. Same with sendmail and the Kernel vulnerability.
Many people are fine with compiling their own versions, and that's what I do. But for some companies it's worth it to them to pay that sub
The great thing about Linux is that you have that choice. You can either DIY, or you can pay someone to do it for you, if that solution makes sense. And for some companies, having someone else do it makes sense.
"PS. What worries me is that "Enterprise Linux" though cheaper than SCO's licenses still is rather steep priced. Where did this come from all of a sudden? Does the little proprietary addon (read: installer) and more or less professional support really make up for the rather large price difference between free and $1500? Personally, I think not. It's exploiting the work of the community which basically built these products."
If you take RHL Enterprise, I think all that is "licensed" is the artwork/name and that is tied to a support contract. I do wish they made this more clear in the EULA. So this means that if you wanted you could strip the artwork and do what you wanted as long as you didn't call it Redhat.
Is the support worh it? I have no clue. It seems people have been screaming about "lack of support" for a while. This is support shoved down your throat, which I guess is what they wanted.
you also have to remember, unixware is a unix that has surived, there is a litany of those that havent (much to my disappointment, BSDI!). it WILL be the next commercial unix to die, but it has survived this long, if it was really bad, it would have died out long ago.
using unixware is like using an unpolished old version of solaris. if sun didnt put the work into solaris they have, youd get unixware.
it served its time well and should be retired.
I have found OpenServer to be a much better product then UnixWare was. I did a technical comparison for Groklaw and I found much of the lower level functions to be much like a UnitedLinux distribution. If you ordered the OS from SCO how come the author did not get a license key. I got one when I ordered the packages. Also why did the author go to the skunkware site, The Optional Services CD has alot of the extras that the author has brought up and they are more recent. I have personally had no problem with UnixWare and I still have it to play with, I have built MySQL, Postgre, KDE and other consumer apps for it. I wouldnt use it as my main OS nor deploy it as a primary server solution because Windows 2003 and SuSE Linux have a much broader application support, dont ask me about RHEL because I wouldnt touch Red Hat with a hundred foot pole, but overall I found UnixWare to be pretty much fully functional. OpenServer now on the other hand, the other SCO product, I would use that as a primary because OpenServer is much better than UnixWare and in terms of uptime and recovery has beaten Linux to the floor. What I want to know is this, Did the author order the evaluation or did he prompt for evaluation? Thats my question.
I have found OpenServer to be a much better product then UnixWare was.
Now that would have to be the epitomy of damning with faint praise...
(I'm grumpy because I've just started a new job with a bunch of OSR5 boxes and have thus far found it one of the most user-hostile unixes I've ever used. Even an unmodified Solaris install is nicer.)
From all the BS that SCO throws at Linux you would think that they would have a far better poduct. But in reality they aren't even close to Linux, Solaris or W2k3.
I still find it humorous that I can buy a UnixWare license for only marginally less than I can buy a server to run it on for. I mean, Dell was running a special last month to get a (admittedly minimal) server for $299. I just can't see spending 2.5-3.0x as much for the OS license than I did for the hardware.
I have used SCO products since the XENIX days.
I am currently running Linux.
SCO has always been:
...rock solid, simple, expensive
...rock solid, lacking good tools, expensive
...rock solid, boring, expensive
You get the idea.
B
"Also, with the need to license other software components, the product becomes even more unattractive from a business perspective".
Well, that's what Microsoft and other companies have been doing all these years. It's not like you can do a lot of stuff with a just installed Windows box.
But anyways, I see your point. A GNU/Linux CD comes with a thousand free software packages to choose from.
The only benchmarks run were comparing OpenSSL computation in native UnixWare mode versus Linux Kernel Personality (LKP) mode. This is an extremely poor test and shows that the reviewer doesn't know what he's talking about.
LKP is basicly system call emulation like that which is available in FreeBSD. This has NOTHING to do with pure user-space number crunching required of crypto computations! This kind of test would only show the most eggregrarious scheduling or interrupt handler errors in providing the LKP functionality. This wouldn't (shouldn't?) even show up any compiler differences between UnixWare's cc and GCC since OpenSSL is heavily assembly optimzed on x86.
These numbers arn't even compared to running under a real Linux kernel, which would be the most logical course of action given the reviewer's incomplete understanding.
But regardless, with comments like the following, it becomes painfully obvious the reviewer knows little about this:
The Linux kernel version number piqued my interest, because of the recent kernel vulnerability responsible for the compromise of some Debian project servers. I'm not sure if the same kernel exploit would work in the LKP, but it'd be an interesting test.
If anything, benchmarking system calls should have been done. Something along the lines of these tests: http://bulk.fefe.de/scalability/
The reviewer makes his bias very plain with passages such as:
I want to be as objective as possible, but I'd be a fool to think such a review could possibly avoid the controversy and raw emotions surrounding the company offering the product I've chosen to evaluate.
This combined with the lack of objective and useful benchmarks makes this article little more than a piece of cheerleading propoganda.
-molo
and have thus far found it one of the most user-hostile unixes I've ever used
I have a copy of UnixWare 7.1.1 with an educational license. This was before Caldera acquired SCO. I don't install it for about 1 year but CDE is a rock and the admin/config tools are quite good for an Unix. However, the shell is the most unfriendly and naked interface I ever saw. (but I've no big experience).
Prices: This kind of OS is to be used with hundreds/thousands of Windows clients not just to dploy a web server.
(I know that here in Europe, at least, an airline and Rover Car Spare Parts base their WANs on SCO, they don't complain).
just thought i should point that out as it seems that that change happened before your article was posted. So to sumarise linux, bsd, solaris are cheap or free more supported and arent followed by a pack of vultures ready to grab at their carcas when they inevitably die. Damn id be even less likely to recommend UnixWare or whatever they are calling ti now then windows
I found it and the comments very interesting. Thanks to all of you.
This is one of the very few reviews I have ever found on Unixware. With the current mess going on it is nice for a OS idiot like me to be able to see some of the details of this OS.
Here are some problems I have with UnixWare and SCO.
1) Sent the sales department an emale requesting information on the purchasing of a workstation license for UnixWare, I never received a reply. The whine they have no customers yet they do nothing to improve the situation. How is this related? why would I use a product with poor pre and after sales support? why should I part with thousands mearly to he ignored?
2) UnixWare is stagnant. They have done nothing to improve the product. As pointed out in the article, it still uses X11R5 and still sticks to CDE. People may give Solaris and SUN a hard time but atleast they're doing their best to improve these products rather than sitting around, whining about a lack of customers and trying to sue everything and everyone under the sun.
3) UnixWare has the potential to be a great product, but it is unfortunate that the management are too concerned about suing rather than actually building a better product. Maybe when the ship goes down and the rats jump overboard, someone will purchase the remains and do something constructive with it.
If it were me, I would firstly purchase a "jumbo" license off QT, adopt KDE, majorly tweak the interface, migrate all the management tools over to it, oh and update the X server to R6.7.
Here are some problems I have with UnixWare and SCO.
1) Sent the sales department an emale requesting information on the purchasing of a workstation license for UnixWare, I never received a reply. The whine they have no customers yet they do nothing to improve the situation. How is this related? why would I use a product with poor pre and after sales support? why should I part with thousands mearly to he ignored?
2) UnixWare is stagnant. They have done nothing to improve the product. As pointed out in the article, it still uses X11R5 and still sticks to CDE. People may give Solaris and SUN a hard time but atleast they're doing their best to improve these products rather than sitting around, whining about a lack of customers and trying to sue everything and everyone under the sun.
3) UnixWare has the potential to be a great product, but it is unfortunate that the management are too concerned about suing rather than actually building a better product. Maybe when the ship goes down and the rats jump overboard, someone will purchase the remains and do something constructive with it.
If it were me, I would firstly purchase a "jumbo" license off QT, adopt KDE, majorly tweak the interface, migrate all the management tools over to it, oh and update the X server to R6.7.
Although there appears to be some effort at an impartial review of SCO, politics is important. Reality is and will always include politics because were all human. Uh..I think. Anyway...The decision to apply an operating system to a problem cannot ever be devoid of economics either. They're all intertwined. What does this all mean? It means that I would be reluctant to bet my company's investment in an IT resource that I thought would be derived from a company on the edge of extinction. The animosity SCO has aroused has seriously impacted it's long term viability as an ongoing OS business. Sure a corporate skeleton may linger as a way to collect license fees for certain intellectual properties. I'm not sure the current Linux challenges will be one of fees this carcass will collect from. They will go on in that mode only long term. Development of new intelectual properties will wane. No more OS development will occur. The OS will become an orphan or it will be acquired by another organization. I also would refer to Rambus as a hardware analogy to consider.
Thank you for this excellent review. Especially given the political load of the company in subject on other grounds.
"UnixWare = VaporWare
By Alex (IP: ---.client.comcast.net) - Posted on 2003-12-16 01:40:42
From all the BS that SCO throws at Linux you would think that they would have a far better poduct. But in reality they aren't even close to Linux, Solaris or W2k3."
Quoteth Webopedia.com
Vaporware definition: "A sarcastic term used to designate software and hardware products that have been announced and advertised but are not yet available." Source: http://www.webopedia.com/TERM/v/vaporware.html
My 3 examples: GNU/HURD, AmigaOS4, Microsoft Longhorn.
UnixWare is not for SCO UnixWare is available. I think that last thing is quite clear.
Wired.com writes:
"[..] And remember, if it shipped -- even if it stank -- it's not vaporware."
Source: http://www.wired.com/news/technology/0,1282,61586,00.html?tw=wn_top... (about the vaporware awards 2003). You would not be able to succesfully submit UnixWare.
"unix survivor
By df (IP: ---.in-addr.btopenworld.com) - Posted on 2003-12-15 23:31:47
you also have to remember, unixware is a unix that has surived, there is a litany of those that havent (much to my disappointment, BSDI!). it WILL be the next commercial unix to die, but it has survived this long, if it was really bad, it would have died out long ago.
using unixware is like using an unpolished old version of solaris. if sun didnt put the work into solaris they have, youd get unixware.
it served its time well and should be retired."
I don't think the life cyclus of an OS, hardware, or software is necessarily correlated with the quality. Example: DEC Alpha, Amiga...
As for your other comment regarding Solaris: it thrives mainly on another architecture. Or are you comparing with Solaris x86?
Not that I use any SCO product, or that I would consider it, but this really was a nice to read review. By the looks of it, SCO has a nice IP in its hands with its "Personality" stuff. If they could leverage this knowledge and experience for other *nix like OS'es they might actually have a viable product to compete in a market which is still wide open (virtual computing, compatibility layers). As it is now though, SCO seems to be more interrested in making money from litigation than from competing in an honest market.
Forgive me if I'm misinterpreting the purpose and/or implementation of this "Personality" business, but isn't this basically just the same thing that FreeBSD is implementing with its support for Linux binaries? And isn't the FreeBSD implementation available at no cost, through the BSD license, to anyone who wants to use it? I couldn't imagine SCO, even had their reputation not imploded, making any money off of this.
What a fantastic review! I love it! It would be better if you can provide some screenshot on CDE GUI.
For costing hundreds of dollars more, UnixWare does not have a single feature which is stronger than Linux, not one! You have to admire SCO for having the nerves to put such piece of crap on the market, much less putting such a ridiculous price tag on it. You have to be out of your friggin' mind if you actually think about using UnixWare.
I wouldn't be surprised if SCO would hold a press conference declaring success as soon as some retard actually buys the first copy, just like they did with their Linux license.
To me there are not much hope for the future of UnixWare as long as the current management of SCO still in controlled. By looking at the pricing scheme anyone can tell how greedy they are. Unless UnixWare are a lot more superior to Solaris or other OS, the pricing just not justifiable especially when considering the availability of open source OS such as *BSD and Linux at the price that a lot more cheaper.
MS make their fortune not by greed alone by the strategy they use which SCO should learn. Yeaaaahhh, bringing Linux to court maybe one of their strategy but in my opinion it will just fail. My only hope is that the good engineer/programmer in SCO are tnot as greed as their management.
no i was comparing unixware to the most polished of the sysvr5 bases. unixware IS like an unpolished solarish. at one time it did have potential, but I dont think SCO are investing the $$$ on development on it.
unixware sites exist, but i really think it will be the next proprietory unix to become extinct, and running it will be akin to running coherant or something.
Somebody wrote:
"One thing did stick out to me: playing with a real Unix hammered home how horrible Linux's disk management was. Linux was trying to give SCSI drives drive letters? Linux's ties to cheap hardware still seems to be a downfall for it, just as Unix's ties to expensive hardware are for it."
Linux DOES not assign SCSI drives "Drive Letters".
It assigns /dev/sd?x where ? is a letter and x is a partition
number. And of course via the /etc/fstab and /etc/mtab you
mount the /dev/sd?x to a particular place in the filesystem.
About Linux and hardware. Linux supports both cheap and expensive enterprise level hardware of many architectures...
I don't know where you are getting your information, but you are CLEARLY uninformed.
well, the benchmark in the article means less than 0. It would mean exactly 0 if it wasn't misleading, but as it stands, it's counterproductive at best.
Linux SCSI drive and partition enumeration: above poster is right, but that doesn't make it any easier to work with lots of SCSI drives. /dev/sdXY, where X is a letter of the english alphabet, and enumerates the SCSI drives, makes it more difficult to work with large SCSI or fiberchannel setups.
I understand that reviewer actually didn't work with Unixware before, but still, before making review it would be pretty useful to check what are the real features of OS comparing to other.
I didn't touch Unixware for 2 years but then in some areas it was clearly ahead of Linux.
1. AIO - pathetic on Linux. Dirty hack from Oracle would so the trick, but only for Oracle. It's almost impossible to compile any serious program using AIO on Linux. Seems like 2.6 catching on this.
2. Volume management and filesystems. If you have only one HD - you don't need this, but in other case it's just could be a lifesafer. Until recent port of JSF and XFS, xmm.. Linux wasn't close to enterprise ready.
3. Multiprocessor and Threads. 2.4 was baby on 4 processors server comparing to Unixware 2 years ago (my personal tests). Seem like 2.6 is picking on this as well.
4. System monitoring. Taking care of one server is nice, but what about dosen of them. Remote monitoring was hugely important. Don't really know current state on Linux on this, but I assume all these enterprise distribution have this.
5. Windows compatible networking. Samba of course is very nice and all, but then 2 years ago it couldn't be integrated with existing Windows domains so easily as SCO solution (having MS code was very handy of course).
So, the bottom line is: It could be really interesting to see review which measures performance and scalability of enterprise OS in enterprise configuration and comparing ENTERPRISE features. Even 4 years ago Linux on uniprocessor for workstation was much more usefull than SCO, but on server it was completely different. 2.6 vs Uniware vs Solaris x86 on 8 way box would be extremely interesting reading.
I personally don't like current course of the SCO, but we shouldn't compare the OS based on politics.
P.S. Sorry for my bad English.
At my last job we used a 486 machine still on our SCO v3.2 box. This box was a literal workhorse, it ran a company wide management package which included: Full POS support, Accounting package, departmental inventories, specialized management, etc.. And yes it was very specific to SCO as it ran using the old BuisnessBasic 90 language that many older coders will know. I had alot of respect for the operating systems stability, and yes I agree with many of the previous posters in saying the SCO now is not the SCO that developed these workhorses.
"Linux offers a far greater value and far better enterprise application support. In addition to Linux, FreeBSD and the other BSDs, and Linux offer a far better open source environment, better hardware support, and more features."
Which explains why UnixWare doesn't compare favorably to Linux. Looks like SCO should spend time and money trying to improve their product instead of waging a jihad against Linux. When its all said and done, Linux will bury them.
AFAIK, UnixWare was develloped by Novell and acquired in 1995 by SCO, that had its own system - Open Server. SCO tried to merge these systems, but failed. We have to agree that Santa Cruz Operations, Novell and pre-McBride Caldera have made a lot of good to *nix technology. But now these two SCO's products are totally deprecated and highly overpriced. SCO now isn't interested in developping products - they only pretend to be technological company, they even dont't care to market these products properly.
SCO's business model is litigation only.
If I had to choose between UnixWare and Windows, I'd go with 2003 Server, it's cheaper and application support is superior. But... I already chose Linux instead.
--
Quote: 2) UnixWare is stagnant. They have done nothing to improve the product. As pointed out in the article, it still uses X11R5 and still sticks to CDE. People may give Solaris and SUN a hard time but atleast they're doing their best to improve these products rather than sitting around, whining about a lack of customers and trying to sue everything and everyone under the sun.
--
You missed one very important note with UW's age - not only is the included software for it stagnant, the hardware support for it is even worse. It has the absolute worst hardware support you will ever see. Dont even bother trying to get it running well on anything cutting edge, and even semi-modern hardware gives it trouble.
Dont even get me started ranting about security or stability issues with it.
I used Unixware 2.X and 7.0 before at a previous employer. I still have nightmares about it.
"Dont even bother trying to get it running well on anything cutting edge, and even semi-modern hardware gives it trouble. "
Thats just plain not true. You were using Unixware 2.x &7.0 which are OLD. UnixWare 7.1.3 has much better hardware support.
Further with regards to the CDE and X11R5 comments... Agreed that CDE is fugly, but most customer sites I've been to that run UnixWare aren't using it for it's pretty desktop.
regards
-jim
Just wanted to quickly say that was a very nice review.
Still no benchmarks on programs against other similar OS's but very good in terms of installation, compatability and software etc. If more people write like this I will be very happy!
Even Microsoft is a better company than the SCO. That's a low blow to the sofware industry because Microsoft is a criminal monopoly. Why do these types exist in the land of liberty and freedom. Surely they are connected to Alciada in some way, shape, or form.
His speed test is very flawed. Generating rsa/dsa keys is almost purly cpu work. Little to none system calls. So unless the code actually runs in an interpreter, one could not expect these to differ much.
Again, I could be completey wrong about something, but as far as I know in linux scsi drivers are listed as /dev/scd0, /dev/scd1, etc, and in other unixes you just replace the scd* w/ some other abreivation."
Linux numbering of SCSI devices suffers the same problem as w32 drive letters; if you add a device at a SCSI id between two pre-existing devices of the same class all devices with higher bus,target,lun values get their /dev entries incremented. This sucks.
SysV UNIX uses absoluste /dev entries for SCSI, much as Linux does with IDE. This does not suck.



