tester

I, a NON-engineer, used to test software for a Fortune 100 corporation. I worked with 10-15 programmers a day and, in my opinion, engineers are NOT software testers. Engineers are NOT paid to test software, nor, in my opinion, should they be held primarily responsible for the output provided by that software if it can be shown the software has serious "bugs" in it that allow for structures that are NOT code compliant to be generated by this same engineering software.

The FACT is that engineering software will indicate that it meets all requirements of all applicable codes - and in my opinion, that is simply a misrepresentation of the facts. In our particular case, I could show that the software output indicated the specifications met all applicable building codes and yet, we were given a GE truss to be placed next to a cathedral ceiling making the bottom chord BELOW the ceiling level and thus, violating numerous provisions of the building code - in spite of the output saying the truss was "suitable for its intended purpose" and that it met all applicable building code requirements - something that in our case was absolutely NOT true! Indeed, this condition violated the output's own warning notations stating that the bottom chord had to be tied into a rigid ceiling. As such, engineers need to know... the fact that a software package says "meets all building code provisions", etc. on output... DOES NOT NECESSARILY MAKE IT SO!

In my opinion, it is unreasonable for "the industry" and "enforcement bodies" to expect engineers to be held basically solely liable for output from engineering software. Engineers would not intuitively test software to "bend it" ...to see if it will do what it should NOT do... but that is exactly HOW such software should be tested... to "break it"! Engineers would not intuitively include a flaw in their work... that would go against everything they are taught... and as such, in my opinion, although valuable contributors to the overall process, engineers are NOT the best persons to test software if they are not specifically trying to "break it" to make it do those things it should not do!

Software packages do not usually provide a way for the software maker to be FORCED to make a change when a bug is found, nor do they provide a method of documenting what was submitted for review/correction by the software maker by engineers! For all these reasons... I would tell engineers... BEWARE THE SOFTWARE! Push for laws requiring basic testing in matters of appropriate software level access (i.e., do you really want a guy in the lumber yard designing trusses that are up to 120 ft in span, etc.), verification of qualifications upon logon (i.e., input of engineer's number and/or some type of identifier allowing engineers to see if a NON-engineer has provided the output, etc.). Also, reports "output" is often coded separately from the software itself... making it such that if not properly tested, the "printscreen" may NOT be the same as what actually comes out on the printer if "reports"/"output" are coded separately and not properly tested! THIS IS A HUGE ISSUE FOR PERSONS NOT FAMILIAR WITH SOFTWARE TESTING! Without proper checks/balances in the law, engineers, in my opinion, are in for a world of hurt as they become more dependent on software packages in the work they perform and/or review on a daily basis.