Windows XP, Anyone?

J

Jake Brodsky

As seen in many other forums, the end user license agreements (EULA) from Microsoft are getting a lot more cumbersome, expensive, and difficult to read. Microsoft Software is not the cheap operating system it once was.

The questions I have for this group:

Are you still recommending Microsoft Operating systems for use in your projects? If so, have you changed your policies on what applications you recommend thier products for? If you aren't recommending Microsoft, what alternatives are you proposing?

On a side note: is anyone concerned about how tightly standards such as OPC are tied to Microsoft platforms? Are there any emerging alternative standards?

M

Michael R. Batchelor

Well, CORBA isn't tied to Microsoft, and it's freely available.
It's just no as popular as the DCOM which underlies OPC. I can't think of any technical reason that an OPC-like standard couldn't work with CORBA underneath. And there is no reason CORBA can't work on Windows. It's just that nobody does it.

MB

C

Curt Wuollet

HI Jake
There are no alternatives supported by the big automation vendors. How could one possibly reccommend an alternative? And this _is_ my pet peeve amd a major liability for the whole automation community.

Because they have no real choice, they are bound to whatever Microsoft does as far as licensing and forced obsolecense. Dependence on COM/DCOM and OPC tend to reinforce this but since there is really no choice in your tools, databases, etc., many folks don't see this as a problem.

If there were choice, it would tend to favor cross platform standards and you can't have that, it must be MS or nothing or the Redmond gang retaliates and makes your pricing non-competitive.

I don't think this will change until people start seeing how bad for business and risky having no alternatives really are. It may not even happen then as there are, no doubt, exclusive licensing agreements between these corporations and Microsoft that prohibit selling anything else

> On a side note: is anyone concerned about how tightly standards such as
> OPC are tied to Microsoft platforms? Are there any emerging alternative
> standards?

Alternative object brokers, etc. are merely wishful thinking as long as the companies are in bed with Bill G. Microsoft now specifically forbids using anything open with it's software.
The only way would be to provide a completely seperate system from top to bottom with no MS involved. That is what we are trying to do with the MAT/LinuxPLC project. I still need convincing that these general class mechinisms are the way to do things in automation but at least we can
use Corba or lightweight alternatives.

Regards
cww

B

Brian E Boothe

XP IS SPYWARE, GET AWAY!!!! ILL STAY W/ 2000 thanks...

M

Michael Griffin

On June 20, 2002 05:15 pm, Jake Brodsky wrote:
<clip>
> Are you still recommending Microsoft Operating systems for use in your projects? If so, have you changed your policies on what applications you recommend thier products for? If you aren't recommending Microsoft, what alternatives are you proposing?<
<clip>

We don't specify an operating system. If the designer proposes to use Windows we do specify what version of that to use, but we don't specify using Windows.
Our decision on this was based on the belief that it is ridiculous to make a
fundamental design decision such as this without taking its impact on the overall system design into account. Windows is simply too specialised and inflexible to be used in all applications.
We would expect the system designer to propose an operating system which was suited to the application, and to take complete responsibility for this choice. We would review this part of the proposal the same as we would any other component in the system. The software itself isn't going to "wear out", so it doesn't present the same sort of spare parts issues that valves or relays do.

************************
Michael Griffin
************************

C

Chiron Consulting

> And there is no reason CORBA can't
> work on Windows. It's just that nobody does it.

Sure they do. The following are all portable CORBA-compliant ORBs that run on Windows (as well as other OS's):

TAO ( "http://www.cs.wustl.edu/~schmidt/TAO.html"://http://www.cs.wustl.edu/~schmidt/TAO.html )
ORBacus ( "http://www.iona.com/products/orbacus_home.htm":http://www.iona.com/products/orbacus_home.htm )
eORB ( "http://www.vertel.com/products/eorb_c.asp":http://www.vertel.com/products/eorb_c.asp )
Orbit ("http://www.labs.redhat.com/orbit/":http://www.labs.redhat.com/orbit/ )
Borland's VisiBroker ( "http://www.highlander.com/products/index.html":http://www.highlander.com/products/index.html )

eORB and VisiBroker are commercial; the others are Open Source. And there are others.

But Michael makes a valid point when he says that CORBA isn't as popular (i.e. widely used) as DCOM. And "nobody does it" does apply to the
observation that there's no OPC-like standard built on top of CORBA. But there's no reason there can't be.

Greg Goodman
Chiron Consulting

R

Ralph Mackiewicz

> Well, CORBA isn't tied to Microsoft, and it's freely available.
> It's just no as popular as the DCOM which underlies OPC. I can't
> think of any technical reason that an OPC-like standard couldn't
> work with CORBA underneath. And there is no reason CORBA can't
> work on Windows. It's just that nobody does it.

The Object Management Group has developed such a standard. Its called: Data Access for Industrial Systems (or DAIS). You can get a copy of this specification via the web at:

"http://cgi.omg.org/docs/dtc/01-07-03.pdf":http://cgi.omg.org/docs/dtc/01-07-03.pdf

It is modeled after OPC but the interface is platform independent.

Products are available.

Regards,
Ralph Mackiewicz
SISCO, Inc.

D

Davis Gentry

Michael Griffin wrote:
> The software itself isn't going to "wear out",
> so it doesn't present the same sort of spare parts
> issues that valves or relays do.

I would take issue with this statement. While the
operating system itself (whichever you should choose) will never "wear out" the hardware attached to the system will. Let us assume that you put together an HMI system with a Matrox video frame grabber card back in 1994. You may well have chosen Windows 3.1 as the operating system (I am not arguing for or against any
Windows system here - just pointing it out as a very possible choice) - yesterday your frame grabber went bye-bye, and today you discover that you cannot under any circumstances purchase one with Win 3.1 drivers. So you look to upgrade your operating system, and find that you now need all new hardware. Then you find out that the HMI software you used back in 1994 will not run under any current operating system. And that the
custom code which used the frame grabber for defect detection is going to have to be completely
re-written. So the $275 replacement of a simple frame grabber is now going to cost you$15k worth of hardware, software, development, and integration time, and that your machine is going to be down for four to six weeks.

At no point in the original design process did anyone make any bad decisions, and the systems were obviously adequately implemented (they've been running 8 years now). In spite of that you have to go to management and explain all of this and they are not happy. I used a frame grabber as an example, but it could just as easily have been a modem or ethernet card, or a video card (the ones I have seen the most problems with).

This issue is in no way seen only with PCs. Try
getting a PLC-2 system up and running after major
problems today. It's worse with PCs, however, because of the accelerated life cycles in consumer goods, and the problem has gotten much worse since PCs became a ubiquitous item in American homes.

There are companies which rescue old cards and sell them to people like us who are in the position of having to choose between shutting down a production line and purchasing a ten year old card to replace the one that just blew.

Anyone out there have a long term solution for this problem? I don't know of one yet.

Davis Gentry
Application Engineer
Delta Tau Data Systems

J

Jiri Baum

Michael Griffin:
> The software itself isn't going to "wear out", so it doesn't present
> the same sort of spare parts issues that valves or relays do.

Unless it's leased or requires activation after hardware replacement. In either case, the vendor may refuse to renew / not exist down the track.

Jiri
--
Jiri Baum <[email protected]> http://www.csse.monash.edu.au/~jirib
MAT LinuxPLC project --- http://mat.sf.net --- Machine Automation Tools

P

PETERSON

Who are you worried is spying on you?

Every Windows system gives the network administrator spying rights on anyhting you do or store on the computer you use. Thats a big part of what SMS is. But when you use a computer and network belonging to someone else, you have to abide by the rules of the owner.

I really suspect you think XP is some big spy thing for MS, and probably thats in some small way true. But, so what?

M

Michael Griffin

On June 24, 2002 04:22 pm, Davis Gentry wrote:
> --- Michael Griffin wrote:
> > The software itself isn't going to "wear out", so it doesn't present the same sort of spare parts issues that valves or relays do.

Davis Gentry:
> I would take issue with this statement. While the operating system itself (whichever you should choose) will never "wear out" the hardware attached to the system will.
<clip>

Well yes, the hardware will wear out - I never said it wouldn't. The software isn't wearing out though so I need to stock spare hardware, not
spare software.

> and today you discover that you cannot under any circumstances purchase one with Win 3.1 drivers. So you look to upgrade your operating system, and find that you now need all new hardware. Then you find out that the HMI software you used back in 1994 will not run under any current operating system. And that the custom code which used the frame grabber for defect detection is going to have to be completely re-written.
<clip>

I am familiar with this situation. We have some test systems which were designed about 8 or 9 years ago. The boards are obsolete, and using the
replacement products for them will require re-writing the software. The original software was written in 'C', it used LabWindows for DOS to create the user interface, and DOS 6.2 as an operating system. I am looking at implementing the same test method in new software, new computer hardware, and new boards.

This isn't an urgent project as nothing is down and we still have spare parts. However, I am aware of what you are referring to. I have noticed though that the turn over in data aquisition boards is at a much slower rate than for the computer hardware and operating systems. Generally, I am quite pleased with the long term availability and support of hardware we had from
National Instruments.

Note though, that we are still using the same software today that we used originally. The software isn't wearing out - the hardware is (actually may - the data aquisition boards are holding up very well, unlike the computers).
However yes, new boards means new drivers which means new application software which means new development software, etc.

I would consider an 8 year (or more) turn over period to be acceptable for this class of equipment though. A one year turn over (as postulated in the original question) would not be.

I am by the way, considering using Borland Delphi/Kylix as the new development software for this project. I don't know if this is practical in this application, but the fact that it is intended as a cross development system for both Windows and Linux is rather attractive. This may allow us to more easily hedge our bets when it comes to selecting an operating system.

--

************************
Michael Griffin
************************

V

MS try to turn the community to new model of software selling. MS try to legalize the "endless upgrades policy". Policy of total dependence of users from MS. Serious users understand the danger and many of them reject the XP... I very frequently meet such statements in forums. No need to say I personally do not plan to use XP in
any circumstances and more and more closely look at Linux. (sometimes we use NT for UI on a separate process-independant computer)

J

Jiri Baum

A partial solution would be to demand the source for *all* parts of the system, which would minimize the amount of re-development. You'd still have to back-port the new drivers to the old OS, reintegrate and retest, but it should come to much less work/cost than upgrading everything.

Jiri
--
Jiri Baum <[email protected]> http://www.csse.monash.edu.au/~jirib
MAT LinuxPLC project --- http://mat.sf.net --- Machine Automation Tools

C

Curt Wuollet

And I might add the the full solution would have been to use OSS in the first place and archive the drivers you use as they are upgraded and ported to new kernels. That's what I've been doing. Lately, the V4L (Video for Linux) is a normal part of the distribution and it just works.

Regards

cww

C

Curt Wuollet

That doesn't bother me so much, after all that's what Microsoft does. The part that saddens me and makes it sickening, is that people will meekly bend over and let them do it as long as they don't have to learn anything new.

Regards

cww

V

Hello Curt,
Yes. There is a story about a man who goes to buy a rope when he see that all around him plan to hang. Mainstream. There is the believe that mainsteam is the right and safety decision... because the croud is right. Complex question, but not in the case when the croud is a flock leaded by the herd instinct provoked with the mass media, ads.

M

Michael Griffin

On June 24, 2002 05:08 pm, Jiri Baum wrote:
>Michael Griffin:
>> The software itself isn't going to "wear out", so it doesn't present
>> the same sort of spare parts issues that valves or relays do.

Mr. Baum replied:
>Unless it's leased or requires activation after hardware replacement.
>In either case, the vendor may refuse to renew / not exist down the
>track.
<clip>

Consider this however in the context of what else I said:
"We would expect the system designer to propose an operating system which was suited to the application, and to take complete responsibility for this choice. We would review this part of the proposal the same as we would any other component in the system."

An operating system which was "suited to the application" (or at least any of the typical applications I am familiar with) should not be subject to these problems. This is a piece of production machinery
- not something that sits on someone's desk.
If in the course of reviewing a proposal where we asked someone to "take complete responsibility" for selecting an operating system, and they selected something which presented these difficulties, we would have to conclude that this person did not know what they were doing. This is why the operating system would need to be reviewed "the same as ... any other component in the system."

The reason why I pointed out that the software isn't going to wear out, is that if someone proposed using an operating system which we didn't already use, we wouldn't dismiss it soley for that reason. It's not like a valve, or a PLC where we would have to stock extras on the shelf for when it wears out or breaks.
Instead we would have to review questions such as "how do you back it up?", "what happens when I need to replace hardware?", "is it compatable with our preferred computer hardware?", etc. If we got reasonable answers to these questions, we would give it serious consideration.

As a customer, we need to be careful that we don't specify something that can't meet our expectations. The consequences of specifying a particular operating system aren't necessarily obvious at first glance. We don't want to find ourselves compromising on the capabilities of a system for the sake of an operating system that we weren't particularly attached to to begin with.

************************
Michael Griffin
************************

J

Jiri Baum

Michael Griffin:
...
> An operating system which was "suited to the application" (or at least
> any of the typical applications I am familiar with) should not be
> subject to these problems. This is a piece of production machinery -
> not something that sits on someone's desk.

> If in the course of reviewing a proposal where we asked someone to
> "take complete responsibility" for selecting an operating system, and
> they selected something which presented these difficulties, we would
> have to conclude that this person did not know what they were doing.

Which would have to mean that anyone that recommends Windows XP doesn't know what they're doing; and as MS phase out other versions, anyone that recommends MS doesn't know what they're doing (if they ever did).

> The reason why I pointed out that the software isn't going to wear
> out, is that if someone proposed using an operating system which we
> didn't already use, we wouldn't dismiss it soley for that reason.

Right. No worries.

Jiri
--
Jiri Baum <[email protected]> http://www.csse.monash.edu.au/~jirib
MAT LinuxPLC project --- http://mat.sf.net --- Machine Automation Tools

M

Michael Griffin

On July 2, 2002 09:20 pm, Jiri Baum wrote:

<clip>
(My original statements were quoted)
> > If in the course of reviewing a proposal where we asked someone to
> > "take complete responsibility" for selecting an operating system, and
> > they selected something which presented these difficulties, we would
> > have to conclude that this person did not know what they were doing.
>
(Mr. Baum replied)
> Which would have to mean that anyone that recommends Windows XP doesn't
> know what they're doing; and as MS phase out other versions, anyone that
> recommends MS doesn't know what they're doing (if they ever did).

I will qualify my own statement (in the first paragraph) by saying that this opinion applies to systems which:
a) are directly integrated into a production line,
b) contain various data aquisition and other special boards,
c) must maintain operation with minimal down time.

That is, this describes production systems such as automated test systems, or computerised control systems in "standard" OEM machines. These systems are very expensive to shut down, and solutions such as hot-backups or a quick swap of the entire computer are not realistic.

Non-critical computer SCADA and MMI systems may not have these same restrictions. I won't try to voice an opinion on these types of systems as I don't have enough operational experience with them to judge.

Given the above qualifying statements, I would be a bit hesitant in saying "anyone that recommends Windows XP doesn't know what they're doing". I would say rather that they would have to be prepared to make a very strong case in
favour of it before I would be convinced.
I would not however be overly surprised if the person recommending the Windows XP operating system did in fact not know what they were doing when it came to OS problems. Windows is a very complex operating system, and there are very few people anywhere who really know much about it. There are unfortunately, quite a few people who know just enough about it to be
dangerous.

************************
Michael Griffin
************************

C

Curt Wuollet

More distressing is that many people don't even know there are more viable alternatives. These are the same folks who believe PC _hardware_ is unreliable because it crashes all the time.

Regards

cww
--
Free Tools!
Machine Automation Tools (LinuxPLC) Free, Truly Open & Publicly Owned Industrial Automation Software For Linux. mat.sourceforge.net.
Day Job: Heartland Engineering, Automation & ATE for Automotive Rebuilders.
Consultancy: Wide Open Technologies: Moving Business & Automation to Linux.