From: dr_jeff on 24 Mar 2010 21:58 Jack Myers wrote: > In rec.autos.tech Don Stauffer <stauffer(a)usfamily.net> wrote: > > >> I suspect software errors rather than interference. The auto industry >> is actually fairly good in designing resistance to interference, but in >> my opinion not very good on software design. >> ... design of real-time software is quite difficult. ... > > Just love how the factory engineers and federal regulators can examine > a system for a couple of hours and then call it clean. I've spent > literally weeks on the integration test bench running full bore with > top-notch test equipment to tease out rare failure modes, both > software and firmware. The bugs relating to race conditions, cross- > domain timing errors, and sensitivity to normal component tolerances > are especially entertaining--NOT! I would think that, together, the NSTSA and Toyota engineers have plenty of knowledge and experience with this sort of testing and of Toyota's system. They may know what they are doing when it comes to looking for problems. In addition, the software errors that exist in the system (yes, I am nearly certain there are some) would be able to examined elsewhere (because the software in the system is a copy). The type and amount of testing might be perfectly fine to get the information they need. Jeff
From: chuckcar on 24 Mar 2010 22:04 "Jack Myers" <jmyers(a)n6wuz.net> wrote in news:9eqq77-vri.ln1(a)n6wuz.net: > In rec.autos.tech Don Stauffer <stauffer(a)usfamily.net> wrote: > > >> I suspect software errors rather than interference. The auto >> industry is actually fairly good in designing resistance to >> interference, but in my opinion not very good on software design. >> ... design of real-time software is quite difficult. ... > > Just love how the factory engineers and federal regulators can examine > a system for a couple of hours and then call it clean. I've spent > literally weeks on the integration test bench running full bore with > top-notch test equipment to tease out rare failure modes, both > software and firmware. The bugs relating to race conditions, cross- > domain timing errors, and sensitivity to normal component tolerances > are especially entertaining--NOT! > Absolutely. When I was in college we spend a solid week learning how to design input sets to check for faulty coding. It is in no way trivial. -- (setq (chuck nil) car(chuck) )
From: Obveeus on 25 Mar 2010 07:03 "Jack Myers" <jmyers(a)n6wuz.net> wrote in message news:9eqq77-vri.ln1(a)n6wuz.net... > In rec.autos.tech Don Stauffer <stauffer(a)usfamily.net> wrote: > > >> I suspect software errors rather than interference. The auto industry >> is actually fairly good in designing resistance to interference, but in >> my opinion not very good on software design. >> ... design of real-time software is quite difficult. ... > > Just love how the factory engineers and federal regulators can examine > a system for a couple of hours and then call it clean. I've spent > literally weeks on the integration test bench running full bore with > top-notch test equipment to tease out rare failure modes, both > software and firmware. The bugs relating to race conditions, cross- > domain timing errors, and sensitivity to normal component tolerances > are especially entertaining--NOT! Yep. Hopefully the reason that the NHTSA bought up the computer board is so they can spend the next month or two with it tied into a logic analyzer.
From: Obveeus on 25 Mar 2010 07:05 "Tegger" <invalid(a)invalid.inv> wrote in message news:Xns9D45C8CDDC650tegger(a)208.90.168.18... > "Obveeus" <Obveeus(a)aol.com> wrote in > news:hodfar$8fb$1(a)news.eternal-september.org: > >> >> "hls" <hls(a)nospam.nix> wrote: >> >>> Airplanes have seen the same sort of interference from cell phones, >>> but it doesnt seem to be very repeatable and is normally fleeting >>> with no residual >>> results. >> >> I think that internal software or electronic hardware platform errors >> are far more likely that glitches due to cell phone signals, sun >> spots, etc... > > I think glitches caused by the presence of free-and-easy tort are the most > likely of all. Most definitely. Regardless as to whether or not there are real problems, the trend in reported problems related to news coverage of reported problems / recalls clearly demonstrates that most of the complaints are linked to opportunity.
From: Scott Dorsey on 25 Mar 2010 10:51
In article <4BAAC351.9000906(a)msu.edu>, dr_jeff <utz(a)msu.edu> wrote: > >I would think that, together, the NSTSA and Toyota engineers have plenty >of knowledge and experience with this sort of testing and of Toyota's >system. They may know what they are doing when it comes to looking for >problems. In addition, the software errors that exist in the system >(yes, I am nearly certain there are some) would be able to examined >elsewhere (because the software in the system is a copy). Toyota may, but the NHTSA sadly does not. On the other hand, they may be able to call in some folks from the NTSB, for example, who do. >The type and amount of testing might be perfectly fine to get the >information they need. It's a lot easier to find odd conditions with code reviews and verification than with exhaustive testing. But it's also very, very expensive. --scott -- "C'est un Nagra. C'est suisse, et tres, tres precis." |