Slashdot Mirror

← Back to Stories (view on slashdot.org)

Software Glitches Cause Airport Delays in Britain

Posted by michael on from the abort-retry-ignore dept.

bnoise writes "There has been air traffic delays of up to 6 hours today above UK (and this includes north atlantic flights). A BBC News article points out the reasons: a software upgrade. Another article gives more general information about the delays. Companies pin-pointed are IBM (initial development) and Lockheed Martin. If only they were using Open Source Software in the aviation industry... By the way, is there any Open Source project in the aviation sector? A search on Freshmeat gives back 5 projects."

13 of 194 comments (clear)

  1. Interesting news but... by Marx_Mrvelous · · Score: 5, Insightful

    I seriously doubt that open source is the solution to this problem. Honestly, there are glitches in OS projects, too, that get by review. I don't like this spin put on this story... OS is *not* the holy grail of software development!

    Oh well, time to burn some karma for a neede rant ;)

    --

    Moderation: Put your hand inside the puppet head!
    1. Re:Interesting news but... by medcalf · · Score: 3, Insightful

      And if this had been developed open source, how many contributors would there have been? This kind of software is more like an internal development project for an organization, albeit outsourced, than it is like a general release application. The companies developing this gain none of the benefits of open source development (wide expertise available to the project, many eyes to review, etc) whilst giving up all of their development effort to their competitors bidding for the next, similar, project for some other country.

      --
      -- Two men say they're Jesus. One of them must be wrong. - Dire Straits
    2. Re:Interesting news but... by dthable · · Score: 4, Insightful

      I think someone needs to step out of the glass box. This comment wasn't written by anyone with any amount of development experience. The number of programmers working on a system still don't solve all of the problems. The root cause of this defect isn't listed. If it is a code mistake, then it might have been caught by a different group of developers. What if the defect cause wasn't a mistake, but a lack of understanding about the requirements? External system feed was incorrect? Does your open source model prevent this? I could take the best programmers in the world and give them incorrect requirements and guess what? You'll get incorrect software. I guess then you'll be touting the benefit of closed source then.

  2. I'm a big Free Software fan but... by evilpenguin · · Score: 5, Insightful

    Open Source isn't a magic bullet for this kind of thing. Software Engineering is the solution to this kind of thing, and no one has a monopoly on that. The amount of crap code in the Open Source world and proprietary world is, in my experience, roughly equal. (Actually, I think there is a bit more crap code in Open Source, but it doesn't get used much). The difference is that with Open Source/Free Software you know what you are getting and with closed/proprietary you don't.

    1. Re:I'm a big Free Software fan but... by artg · · Score: 2, Insightful

      But since this is a custom development, not a shrinkwrap, the customers almost certainly do have access to source.

      This is a testing problem : the new terminal software has generated server traffic of a volume that wasn't expected. Might be a terminal software bug, might be a system design problem, might be a capacity thing that they weren't expecting.

      Either way, adequate testing would probably have avoidede it - but testing big client-server systems is pretty difficult, as the live system is often the only viable test platform. Credit to the ATC people that they were able to back off the upgrade and continue to run at night-time loading : better than the crash-and-burn that failed upgrades more often cause.

  3. Freshmeat? by QuodEratDemonstratum · · Score: 3, Insightful
    Pardon? There's a world of difference between
    • A flight planning tool for pilots
    • Perl module for processing aviation weather reports.
    • Parses FAA weather briefs into individual NOTAMs/METARs/PIREPs/TAFs/etc.
    • Local weather data accumulation for Web sites
    • A Linux port of the X-Plane flight simulator.
    and an ATC system for trans-atlantic airspace.

    I think we're being trolled!

  4. This may not be the place for OSS... by PoiBoy · · Score: 4, Insightful
    As much as I enjoy using open-source software for many things, I would prefer to fly knowing that the software controlling air traffic was produced by a small number of companies. First, something as critical as air traffic control is probably best developed by very knowledgeable experts with extensive backgrounds in air safety. While many (most) OSS contributors are great programmers, I doubt if many truly understand the needs of air traffic control. Secondly, as many companies and PHB's say about OSS, if someone in my family were in an airplane that crashed due to air traffic problems, I would like to hold someone liable if there was a software glitch that should have been found and fixed before being deployed. Of course, mistakes happen and we shouldn't look to sue everytime one occurs, I'd still feel safer knowing that if there were gross negligence I would have some legal recourse.

    --
    Sig (appended to the end of comments you post, 120 chars)
    1. Re:This may not be the place for OSS... by T5 · · Score: 2, Insightful

      You're missing the point of using OSS in this arena. The issue here is not one of open sourced ATC software, or typical OSS developers creating ATC software (which isn't necessarily a bad thing, judging by the quality of many OSS pieces). It's one of the tools of the trade being OSS.

      The risk of any glitches that may arise as the result of using OSS as the core building blocks of any mission-critical/life-encapsulating project such as ATC may actually be less than the closed source arena. We've got hundreds of examples where errors in OSS are fixed correctly in a more timely manner than many of their closed source counterparts.

      As to the legal liabilities, those rest squarely on the shoulders of the developers of the ATC code, not on the tool creators. The ATC developers, were they to find a heretofore unseen bug in some OSS tool, are in a position to fix it and/or report it to the package maintainers. This would help avoid those nasty little workarounds that lead to nasty code that much harder to maintain/certify.

  5. Open Source Terrorism by FortKnox · · Score: 3, Insightful

    Sorry to make this sound "Katz"-ish, but try and follow.
    You make an air-traffic control program open source. An airport decides to use it. A quality hacker, yet terrorist, jumps into the project (honestly, how difficult is it to get into an Open Source project? I haven't heard of one needing a background check). His code is quality for a long time and gets put into the program. He becomes a trusted member. He pulls a "DirectTV" hack (pieces of code, in several different packages that work once the package is complete) that causes many deaths.

    Yes, this could happen in the software company that creates the software now, but it would be a lot easier for a terrorist to get into an open source project...

    Just another example/reason that Open Source isn't the answer for everything (don't get me wrong, I'm an open source advocate myself, I just know some of its limits).

    --
    Good quote, too many chars. Seriously, the slashdot 120 char limit sucks!
  6. Mistake #1 was a bad assumption by Art+Popp · · Score: 4, Insightful

    From the article:
    "But software is advancing at a tremendous pace, so it becomes obsolete every 18 months."

    Um, no, it's hardware that doubles in speed every 18 months. The approval rate for new aircraft technologies (at least in the states) is unbearably slow. This is clearly a weak excuse for the correctly identified problem:

    "The basic stumbling block was not to get off-the-shelf components and software"

    Maturity couldn't be a more critical issue to this kind of software. Where half a day's downtime can cause inconvenience to 10% of the population of your island, and ignoring a problem can get people killed, you need a proven winner. Software for managing the traffic over the UK should not even have been considered unless it had been proven for years of service controlling airspace over something noticably less crowded than one of the hubs of global trade.

  7. Ludicrous by dmccarty · · Score: 5, Insightful
    Headline example:
    bnois writes, "The California Highway Patrol has been reporting that during rush hour today several large bridges in San Francisco, including the Golden Gate Bridge, have had sections collapse, sending cars and trucks hurtling to their demise below." If only some qualified engineers had drawn up the plans in their free time and let the general public view them first for errors. Does anyone know of some plans like that on Fresharch?

    Linux is a great example of the open-source mindset at work. And there are other great examples of open source projects that work. But the idea that Open Source is the cure-all for all projects big and small is ludicrous. Whoever wrote "If only they were using Open Source Software in the aviation industry" has obviously never been involved in a 100-person project that spanned years and was responsible for critical operations.

    Declaring Open Source to be a cure for all ills is like treating every disease with the same pill. It just doesn't work that way. Open Source software is great when people can unite for a common cause (usually against a common competitor, which Microsoft convienently happens to be) and produce a good product. But thre's no evidence that an Open Source project would have worked where this upgrade failed.

    Closed source might not be your model of choice, but it solves the same problem. Software engineers writing code which is never released to the public don't do their jobs any worse because of it. You might think that the purity of the code is flawed by company management bent on releasing buggy products for profit, but the open source alternative is a Mozillian, buggy product that is years behind schedule and never quite ready. Don't assume that just because a model you don't like has a failure the model that you do like would have worked.

    --
    Have fun: Join D.N.A. (National Dyslexics Association)
  8. Open Source and Software Validation? by blair1q · · Score: 4, Insightful

    The economic value of the open source development model is that directed validation is unnecessary.

    The code is released, and the horde of developers does trial-by-fire validation for you. They run it in real-world usage and report bugs itinerantly for others to fix or sign-off on.

    That's not feasible for programs where using the code means implementing it in an embedded system responsible for safety. The downloaders won't have the hardware to test it on, and putting it in use to test it misses the point of validating it.

    But it's not as though the validation systems in use today are much better. Simulators and debugger-controlled code exercisers create sort of a chicken-and-egg problem. Recursive review decreases the probability of certain kinds of errors, but not to nil.

    --Blair

  9. Why open-source really wouldn't work here by FearUncertaintyDoubt · · Score: 2, Insightful
    One reason is this: the difference between air traffic control software and an open source MP3 encoder or web browser is that anyone can run the latter 2 examples in a real-world situation. I can encode MP3s or browse the web and really put the software through its paces quite easily. I can make changes, compile, then test. I can participate easily in the development process.

    How do you suggest the average coder puts his copy of OpenATC to the test? Start controlling planes from his bedroom? Maybe have all the kids in the neighborhood clear their bicycles and bigwheels for takeoff? I wonder if the testing phase for ATC software is a bigger effort than the actual development.