Software upgrades

[ThomasH]

Hi all

I'm curious about something after a discussion at our office.

I have worked with a pre/post processor for several years, been through numerous version upgrades and typically I just upgrade. I install the new version and open the file in the new version. A program runs that converts the database and then I keep working. I can't remenber any real problems in that process. Some hickups over the years, sure, but nothing worth mentioning as a problem.

Now we have several other softwares. For some of them the users wont upgrade during a project. Start a project i version 10.0, you complete it in v10.0. Upgrades typically leads to problems, uncontrollable changes in settings, corrupt databases or minor uncontrolled changes that results in mayor problems. The "other" softwares I refer to are typically not FEM - softwares but more of 3d cad etc.

So now my "question", have you had any bad experiences, am I just lucky or am I missing something.

And, do you find it acceptable that upgrades are a problem.

NOTE: I don't want to bash or critizize any particular software. I'm just curious if upgrades generally are considered an issue.

Regards

Thomas

 

[IRstuff]

I'm not sure why you think there would be a generally applicable answer. Clearly, if you do nothing that the developers haven't tested, you probably won't run into problems. But, since no complex software or hardware can be 100% tested under all conditions and platforms, all complex systems probably have hidden flaws and defects, and someone will be unlucky enough to find them for the developers.

Some companies tend to do a better job of testing, but my guess is that they go in cycles, i.e., for a few years, they test the crap out of their software, then, someone complains about the excessive testing cycle, so they cut back, and start getting complaints about untested features, so they increase testing, etc.

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[ThomasH]

@IRstuff

Perhaps I'm a bit unclear. What I'm after is if you generally consider upgrades a problem.

I'm not after the errors in v10.1 compared to v10.0. I'm after the database upgrade from v10.0 to v10.1. Is that a cause for concern? Will the database stay intact or may the properties have changed?

There is no general answer but there could be a simple, "We never upgrade during a project" or the opposite.


Regards

Thomas

 

[IRstuff]

Any perturbation, by definition, is a "problem," if for nothing else, it's a disruption of work, and regardless of how smoothly it might go, it rarely is "seamless." But, the farther back you lag on upgrades, the less support you are likely to get if you need it.

The only plausible option is where you do not upgrade to any initial major revision, but opt to upgrade on the 1st or 2nd minor revision. Ostensibly, each major revision is the mostly likely to have the highest undetected errors, since it's most likely to have gotten the least amount of combined developer and customer testing and usage. After a couple of minor revisions, the most obvious bugs will have been discovered and hopefully repaired.

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[GregLocock]

I wouldn't upgrade during a project except after a major gateway, when I'd have a full backup of the model.

The only exception would be if I had found a job-stopper and the workaround was truly appalling.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[ThomasH]

Hi again

Let me put the question in another way.

Let's assume that you worked on a project in 2000. The files are now in archive but the client wants to make a modification.

Would you:

a) Open the archived files in the current version that you work with and feel confident all the data is ok.

b) Install the 2000 version of the software and do the work in that version.

Sorry for being insistent but I find this interesting.

/Thomas

 

[JoshPlumSE]

Thomas -

I've done it both ways. For most projects, I would use the newest version of the software. But, I would always keep an archived version of my old input file in case I start getting drammatically different results with the new version.

However, I have been on projects which had a tight QA/QC control and they required that the program version remain the same. Unless you wanted to go through a long and arduous QA/QC process to prove that the new version was just as good as the old one.

 

[johnhors]

Thomas,

"b) Install the 2000 version of the software and do the work in that version."

It is quite possible that a 2000 version of the software that ran under say windows98 or win2K will not run under windows7!

Also a lot of software programs limit their backwards capability to just a few versions when reading old databases. So unless you constantly update old databases to the latest standard you run the risk of having unreadable data.

http://www.Roshaz.com

quality, cost effective FEA solutions

 

[tstanley]

Hi Thomas:

I haven't had problems with FEA software, but CAM software has given me problems. A lot of customized post processors are used in CAM software and if you upgrade rather than do a clean installation, it becomes less than transparent which version of the post processor it's using. The vendor highly recommends a clean installation.

It hasn't happened to me, but could it happen with an updated FEA pre and post processor, and the previous versions still active?

Tom

 

[IRstuff]

a) Open the archived files in the current version that you work with and feel confident all the data is ok.

b) Install the 2000 version of the software and do the work in that version.

OK, that seems to be a very different scenario that originally posted. The issues run deeper than just archiving the datafiles. That's assuming that you've got all the relevant files archived, but there may be add-ons, or other script files that might also be necessary to duplicate your original processing environment. A case in point would be Mathcad, whose latest version has dropped backwards compatibility with any files written prior to 2002.

I think that in many cases, original design files may be fubar, and that only a portable format like IGES, possibly DXF, etc., migh be viable. You might still need to have oth the original software, AND a viable computing platform that's compatible with that level of software. There are many programs that will simply not run on Windows 7, because of version specific command usage, or hardware specific compatibility.

At the very minimum, you'll need a computer/software independent copy of your original design, in case you have to recreate it from scratch, which would be the worst-case scenario.

TTFN

FAQ731-376
Chinese prisoner wins Nobel Peace Prize

 

[ThomasH]

Hi again

@IRStuff: The original idea was that the two questions should actually be the same. My focus was the database but in the original post I realized that it wasn't completely clear.

I think it is interesting that the general idea is not to upgrade during a project. I usually work with several parallel projects so a situation when there is a clean break between versions is usually not possible. But, on the other hand, I have never had any serious problems with upgrades. My overall impression is that FEA software is good at this. Maybe I have been lucky or maybe I missed something.

CAD software is unfortunately another story. My collegues won't upgrade during a project. I understand the possible issues but I find it a bit strange that they just accept the situation. If you have a project that runs for several years and spend money on maintenace, don't you want to use the latest upgrades? Shouldn't the vendor make sure that backward compability works?
I can understand that it won't work "forever" but for adjacent versions at least.

Like I said, I find it interesting that the users accept it. On the other hand I have no idea how to push the vendors towards solving the problem. But if the users don't see it as a problem, who am I to argue .

Thanks for you input anyway.

Regards

Thomas