APLawrence.com -  Resources for Unix and Linux Systems, Bloggers and the self-employed

Ancient Pick on Ancient SCO

Yesterday I visited a company running a Pick MRP app under a SCO 5.0.2 system, For those of you unfamiliar with SCO, that makes it circa 1995. I almost turned them down flat just for that reason: this stuff is ancient, and it can be difficult and quite scary to work on. You can't reinstall SCO on modern hardware, so that's out. Upgrade? Maybe: most apps that ran on SCO 5.0.x will run on SCO 5.0.7, which can still be purchased and can still install on a lot of current hardware. But upgrading to 5.0.7 is like upgrading your 1974 Gremlin to a 1990 Yugo - whoopee, that's a big improvement!

With Pick, there are better options. Pick can run on modern operating systems and that's the other reason I didn't want to get involved in this. Pick isn't chained to SCO - for a relatively small amount of money, they could take that app anywhere.. assuming availability of source code, of course. And even without that, Google says there are Pick decompilers. That might cost a little more, but how bad can it be? Compared to buying a new MRP system, it's dirt cheap. So why waste time and money on SCO?

Well, because times are really bad. This company deals primarily with the auto industry and we know how well they are doing. So, they found the idea of possibly transitioning to SCO 5.0.7 appealing as that would be less expensive than moving to another OS.

Well, maybe again. I warned him that Pick might insist on an upgrade and relicensing. If they do, then you might as well get away from SCO unless they charge a tremendous amount extra. And sheesh: I can't imagine that they aren't willing to deal a bit in this economy either..

But after we talked about all that, I agreed to come up and help him investigate moving the app to 5.0.7. When and if that idea comes up today, I always suggest virtualizing it to avoid the aging OS problem: SCO 5.0.7 should be able to survive a lot longer as a VM than it will be able to on real hardware. So that's what we intended to look at: SCO 5.0.7 under VMware. VMware considers SCO "experimental" at this point, but we know it works, so..

They had SCO installed and running by the time I got there. They also had their Pick consultant there to help with that. So..

Pick installs with Custom. For some reason it insists on being single user mode - probably just being over cautious, but fine, we can do that. But Custom failed to read the floppies.

My first thought was that the floppies were bad, but no, I could read them from the command line. Well, at first, no: the VMware wasn't properly identifying /dev/fd0 so I had to be specific and read from fd0135ds18. I extracted the files from two floppies and ran "/tmp/init.pick" and the install worked. Great.. but now Pick itself needed to read some more floppies and it wasn't doing well at it. It's configuration file suggested that it expected to read from /dev/rdsk/f0q18dt but that always died without success. I changed the config file to read from /dev/rfd0135ds18 and that worked.

I guess it shouldn't surprise me to have trouble in floppy-land. This is VMware running an "experimental" OS under Vista. I can't imagine that Microsoft or VMware have put a lot of effort into their floppy disk drivers. Why would they?

So at that point, Pick worked. Single user, unlicensed, and none of the customer's data. Encouraging, but far from helpful. The Pick consultant had contacted TigerLogic about D3 (that's current Pick) upgrade pricing during all this, but we had no firm answer yet.

Pick uses it's own raw file. We did try just transferring that over. It appeared to work, though again single user and although programs seemed to run, the main screen said that it was corrupted. My bet is that someone who knows more about Pick than I do could fix that up but there still is the licensing to deal with.

So there it is - it looks like it is possible to run an old Pick app under SCO 5.0.7. Again, I'm not convinced that there is any valid reason to do this unless TigerLogic comes back with ridiculous upgrade pricing.

How bad could it be? The Pick consultant told me that IBM Universe (their version of Pick) runs about $400 a user brand new. This is a 50 user system, so that's $20,000. Sure, not pocket change, but if we're considering limping along with SCO on 15 year old hardware.. it's not that much money, is it?

Staying with SCO is silly. Time to move on.



Got something to add? Send me email.





(OLDER)    <- More Stuff -> (NEWER)    (NEWEST)   

Printer Friendly Version

-> -> Ancient Pick on Ancient SCO Unix


10 comments



Increase ad revenue 50-250% with Ezoic


More Articles by

Find me on Google+

© Anthony Lawrence







Fri Mar 13 15:26:55 2009: 5680   Patric

gravatar
I had a OpenServer 5.0.2 box that I moved to new hardware (motherboard etc) three times, it was a bit of a nightmare, but I am pretty well convinced that I can take a copy of 5.0.2 tell it has a buslogic controller running it, dd the entire instance to an image and bring it up under vmware, there are several gotchas, (one that caused crazy instability only under load was a modern motherboard that could not be set to "Plug and Play OS:NO").





Fri Mar 13 15:35:28 2009: 5681   BigDumbDinosaur

gravatar
5.0.7 with MP4 installed can be made to run on modern hardware, with some limitations. I have one client running 5.0.7 on AMD Opteron hardware just fine (there's a middleware issue that keeps them tied to OSR5 for the duration). In some cases, device drivers for embedded hardware, such as on-board NICs, may not be available for 5.0.7, which wasn't the case with this machine. Otherwise, the kernel seems to be okay with the chipset and dual channel RAM. This particular unit uses an Adaptec U320 PCI-X host adapter, for which a device driver was available. It runs real fast and has been rock solid.

That said, I agree that your client should see about moving their Pick MRP installation to Linux and scrapping OpenServer. As the old joke goes, "SCO: the 1980s are calling. They want their UNIX back." SCO doesn't deserve to be in business any more.





Fri Mar 13 15:42:57 2009: 5682   TonyLawrence

gravatar
Yes, 5.0.7 can run NOW, but will it run on current hardware of 2015? Probably not - that's why I say if you have to do it, virtualize it so it can last as long as you need it to.





Sat Mar 14 14:23:53 2009: 5702   BigDumbDinosaur

gravatar
Yes, 5.0.7 can run NOW, but will it run on current hardware of 2015?

The same question could be asked of all current operating systems, right? Can you get MS-DOS 6.22 to run on a dual core Opteron system with 16 gigs of RAM? A question immediately followed by why would you want to run DOS? <Grin> Will Windows whatever of today run on the PC of tomorrow?

It is likely that as long as the fundamental architecture of PC hardware stays as it is now, 5.0.7 could be loaded and run. As with DOS, however, why would you want to?



Sat Mar 14 14:46:07 2009: 5703   TonyLawrence

gravatar
If you put it on Linux, Linux WILL keep up with hardware so you'll always be able to move to new hardware. SCO will never move beyond 6.0.0.







Sat Mar 14 16:25:29 2009: 5705   BigDumbDinosaur

gravatar
Apparently so.

Speaking of hardware supporting Linux, someone has successfully compiled a Linux kernel that will run on the W65C816S microprocessor (MPU). I'm trying to found out more about that, as the '816 was used in the Apple IIgs and the Super Nintendo game system, and is also the MPU of choice in numerous homebrew computers (see (link) if you're interested). I'd be interested in seeing how the problems of memory protection and virtualization were solved, as the MPU itself lacks the memory protection features found in x86 processors.

For those who aren't familiar with the W65C816S, it is a 16 bit static CMOS incarnation of the venerable MOS Technology 6502 8 bit MPU, the latter which powered the original Apple 8 bitters, as well as almost all of the Commodore 8 bit machines (variants powered the C-64 and C-128). It is estimated that some 5 billion 65xx MPUs have been used, some of which have found their way into space and into life-support apparatus. I discovered my microwave oven has a W65C134S microcontroller in it, the core of which is a W65C02S MPU.

The '816 and its predecessor, the W65C02S, was developed by William D. Mensch, CEO of the Western Design Center, and one of the original 6502 architects (he was also part of the Motorola 6800 design team). Mensch essentially took the 6502 architecture, added 16 bit registers and a 24 bit program counter, fabbed it in CMOS and retained the original 6502 instruction set. New instructions were added to take advantage of the enhanced architecture, but all of the old instructions were retained for backward compatibility.

Although Linux running on a 16 bit CPU at 20 MHz won't be quite as speedy as it would running on one of our Opteron screamers that we sell to discriminating clients, the idea of Linux running on a 35 year MPU design fascinates me.



Wed Oct 21 16:22:02 2009: 7327   TonyMcMahon

gravatar
i would love to move our company away from SCO 5.0.x too. but I can't see it happening. we have a large informix database (that's SE v5 - REAL ancient too) plus our own custom applications with about ~3million lines of 'C' code (embedded SQL etc, etc). It seems to me like it would be a massive undertaking to port to LINUX and another/upgraded database (And I only have a team of 3 developers). So, I'm researching new hardware to install a fresh 5.0.7 OS on. /sigh



Wed Oct 21 16:25:34 2009: 7328   TonyLawrence

gravatar
Yeah, it's a tough place to be. You really should consider virtualization - VMware now supports SCO 5.0.7



Wed Oct 21 18:09:16 2009: 7330   BigDumbDInosaur

gravatar
So, I'm researching new hardware to install a fresh 5.0.7 OS on.

5.0.7 will run on most AMD server platforms (I can't vouch for Intel hardware since we don't build anything with their stuff). You may not be able to find device drivers for the embedded hardware (e.g., Ethernet adapters or SATA controllers, the latter which isn't really suitable for server use), however, plug-in stuff will take care of some of that. According to Adaptec's website, there is at least one SCSI host adapter with OpenServer support that plugs into a PCI-E slot, which is found on all current server boards (see (link) for info). You may still find some Opteron boards with PCI-X slots that accept the Adaptec 29160 HA, which also is supported in OSR5. Those boards will be socket 939, but I am told matching MPUs are still available. So there are ways, but it may take a little thinking to put something together.

Obviously, the long-term solution is to port your stuff to Linux, but, yes, that will be labor-intensive.



Thu Oct 22 15:19:10 2009: 7340   Tony

gravatar
the best thing is that i have to get PCI-DSS compliant. with that regard, does anyone have any idea if there is a cisecurity.org benchmark equivalent for hardening a SCO Openserver install that i can get hold of? I've looked at the AIX benchmark (and its effectively a walkthrough - specific commands, scripts etc), and if there was a SCO one, I'd be a lot, lot happier.

------------------------
Kerio Samepage


Have you tried Searching this site?

Unix/Linux/Mac OS X support by phone, email or on-site: Support Rates

This is a Unix/Linux resource website. It contains technical articles about Unix, Linux and general computing related subjects, opinion, news, help files, how-to's, tutorials and more.

Contact us





UNIX does not allow path names to be prefixed by a drive name or number; that would be precisely the kind of device dependence that operating systems ought to eliminate. (Andrew S. Tanenbaum)

Computers are like Old Testament gods; lots of rules and no mercy. (Joseph Campbell)







This post tagged: