Errors in codes - software

[rowingengineer]

Having some time to my self, I have been working thru my codes to update them to the lastest revisions, in this case the cold formed steel code for US and Aust. I have noticed some of the notes on errors in equations have not been update in these revisions, even thou I have highlighted them to the code body. Most of these are obvious to an exp engineer but not to software programmer, I decided to review some of the software that is for the analysis of said codes and they have full code compliance, but are technically 5 percent off true result.

Should there be a software board that reviews and provides advice/ratings? Should there be a place the traces these errors that can be seen by working engineer's?

http://www.nceng.com.au/

"Programming today is a race between software engineers striving to build bigger and better idiot-proof programs, and the Universe trying to produce bigger and better idiots. So far, the Universe is winning."

 

[Agent666]

I'm not aware of any software suppliers that actively promote or discuss errors in their software (if they are even aware of them), they fix them in their own time. Kind of goes against selling more software if you're promoting in a way the lack of QA.

I wish it was different and they were more open to communicating errors and jumping all over them until they are fixed. I've noted a few errors in commercial software over the years myself.

Also noticed an increasing reliance on the use of software, almost blind devotion to the fact it must be right....

 

[271828]

Which equations in the AISI Specification are erroneous?

 

[JoshPlumSE]

When I worked at RISA (under the old ownership, not the new), we originally had a spot on our website where we posted any known bugs that would cause unconservative errors. Not ALL bugs. If there was something so esoteric (like AISC flexural torsional buckling of members with slender elements), then we didn't post a notice. But, our background (Bruce, Roger and me) was with a company (Fluor Daniel) with a strict QA / validation procedure. So, we posted the kind of stuff that we would want to know about if we had still been with Fluor.

Later, it changed a little. Instead of posting that ahead of time, we waited until we had fixed the problem then posted information about the problem in our release notes.

Later still (after some former STAAD sales / marketing folks became higher up in the company), even this started to go away. I was sometimes pressured to "downplay" any bugs that were talked about in the release notes. To avoid using the term "bug", or "error" or such. When I was in charge of the release notes, I would try to put in virtually all "issues" (which became the preferred term) that we had fixed. Now, I'm pretty sure they list the major bugs fixed with each release. But, the minor ones, or the ones few users encountered may be left out.

 

[JoshPlumSE]

Should there be a software board that reviews and provides advice/ratings? Should there be a place the traces these errors that can be seen by working engineer's?

My previous post (about my experience at RISA) was to highlight why I don't believe we can trust the software companies, in general, to self report issues.

RISA started out with the best of intentions.... To act similarly to what we'd done with in-house software at Fluor Daniel. Self report so that all engineers who are using the software know what to be cautious of. But, the industry is extremely competitive. Competitors will use that self reporting to claim (maybe falsely) that the software has a ton of serious bugs compared to their own.

Maybe one day, one of us (independent engineers) could set up a public page with "verification" problems for various codes and analysis examples which can be tested in the various programs. Maybe even with example models that show those problems.

If some of the code bodies (AISC, AISI, et cetera) could provide some funding for that, great. Though I wouldn't hold my breath. More likely, they would just get out of the way, and allow the site to use some of their published examples as part of the verification problems.

 

[BUGGAR]

"Maybe one day, one of us (independent engineers) could set up a public page with verification problems for various codes"
In a way we have this under "Engineering Failures". The thoroughness of the contributors should flush this out.

 

[Guest090822]

I've never tested a program that didn't have issues. Check the disclaimer on any program you open. They all pretty much state that the software isn't perfect and you as the engineer take the final responsibility. At my former place of employment, we had a database with "known" software issues and the work-arounds we came up with. I was actually responsible for "support" of a few programs. I reached out to the vendor on several occasions to point out errors which were then fixed fairly quickly.

If are worried about 5% in structural engineering I'm hoping it's because you design aircraft or spaceships!