Stories
Slash Boxes
Comments

SoylentNews is people

posted by martyb on Tuesday March 12 2019, @12:23PM   Printer-friendly
from the the-times-they-are-a-changin' dept.

The American Radio Relay League (ARRL), the national association for amateur radio, has a reminder about older GPS receivers which may hit a wrap-around bug on April 6th this year.

The GPS network will encounter a small millennium bug of its own in April when the network's "week number" rolls back to zero. This known issue especially could affect those who use GPS to obtain accurate Coordinated Universal Time (i.e., UTC). In the GPS network, the number of the current week is encoded into the message the GPS receives using a 10-bit field. This allows for weeks ranging from zero to 1023. The current period began on August 1, 1999. On April 6, 2019, the week number rolls over to zero and starts counting back up to 1023.

This should not affect later-model GPS receivers that conform to IS-GPS-200 and provide UTC, [...]


Original Submission

This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
(1)
  • (Score: 0) by Anonymous Coward on Tuesday March 12 2019, @12:51PM (6 children)

    by Anonymous Coward on Tuesday March 12 2019, @12:51PM (#813214)

    pshaw! how is this news?
    just wait for that armaggedon bug when we run out of numbers!

    • (Score: 3, Touché) by coolgopher on Tuesday March 12 2019, @01:14PM (3 children)

      by coolgopher (1157) on Tuesday March 12 2019, @01:14PM (#813222)

      The numbers you can count are not the real numbers.

      • (Score: 0) by Anonymous Coward on Tuesday March 12 2019, @06:42PM (2 children)

        by Anonymous Coward on Tuesday March 12 2019, @06:42PM (#813410)

        You've been waiting for a long time to use that one, haven't you? Sorry I don't have mod points.

        • (Score: 2) by maxwell demon on Tuesday March 12 2019, @08:50PM (1 child)

          by maxwell demon (1608) on Tuesday March 12 2019, @08:50PM (#813461) Journal

          Note that it's been in my signature since I've registered on this site (and even before, on the green site). So coolgopher wasn't being very original here.

          --
          The Tao of math: The numbers you can count are not the real numbers.
          • (Score: 2) by coolgopher on Wednesday March 13 2019, @01:47AM

            by coolgopher (1157) on Wednesday March 13 2019, @01:47AM (#813545)

            Oh absolutely not original. I think I first came across it on Usenet many many moons ago. It stuck because it was so neat :)

    • (Score: 3, Interesting) by driverless on Tuesday March 12 2019, @01:51PM (1 child)

      by driverless (4770) on Tuesday March 12 2019, @01:51PM (#813243)

      It's OK, I found some spare ones hiding out past the decimal point.

      • (Score: 3, Funny) by bob_super on Tuesday March 12 2019, @04:55PM

        by bob_super (1357) on Tuesday March 12 2019, @04:55PM (#813352)

        Are those a bunch zeros ?
        If they are, please return them to me. I had a whole bunch, but they fled the other day when some smart ass mentioned a bit too loud how you can get rid of them all.
        I miss them.
        I had great plans to push them just past the decimal point, though my boss keeps saying they should then take a leading role.

  • (Score: 0) by Anonymous Coward on Tuesday March 12 2019, @01:35PM (11 children)

    by Anonymous Coward on Tuesday March 12 2019, @01:35PM (#813231)

    The current period began on August 1, 1999.

    So they had the "Y2K bug" coming up (all over the news at that time)... and could not foresee this being an issue in 20 years? Well, not really an issue as it seems to be fixed already in new gear... Yay for planned expiring goods.

    • (Score: 0) by Anonymous Coward on Tuesday March 12 2019, @03:01PM (6 children)

      by Anonymous Coward on Tuesday March 12 2019, @03:01PM (#813290)

      Yeah, they baked it in on purpose. You own nothing. Certainly nothing that can't be taken from you.

      Defective by design. Love it.

      • (Score: 2) by bob_super on Tuesday March 12 2019, @04:51PM (5 children)

        by bob_super (1357) on Tuesday March 12 2019, @04:51PM (#813351)

        You're totally right. It couldn't have been that 10 bits what what they had available when all critical tradeoffs were considered, and something they knew SW could deal with.

        Must have been planned obsolescence. Everyone is evil.

        • (Score: 3, Interesting) by urza9814 on Tuesday March 12 2019, @05:41PM (4 children)

          by urza9814 (3954) on Tuesday March 12 2019, @05:41PM (#813377) Journal

          Well yeah, software CAN deal with it, but some manufacturers didn't plan for the future and designed their software with limited lifespans. This isn't an unknown issue. This isn't the first time it's happened. And according to the linked documents, this is something which *most* manufacturers were able to get right. The rest are incompetent at best and at worst they intentionally designed a limited lifespan into their products. Any devices created before 1999 would have already been through this once, and presumably already patched, so this would probably only be an issue for early 2000s era products. There is no way in hell that it was a hardware limitation that prevented them from handling this. We certainly had the technology to build a halfway decent digital clock in 2000.

          • (Score: 2) by c0lo on Tuesday March 12 2019, @09:15PM (1 child)

            by c0lo (156) Subscriber Badge on Tuesday March 12 2019, @09:15PM (#813471) Journal

            The rest are incompetent at best and at worst they intentionally designed a limited lifespan into their products

            I will LMAO if/when I hear about a bunch of car models, witt it as extra for some $xxxx, have their GPS navigator bricked and are out of warranty.
            That will be a good reminder for the UNIX philosophy: never think of your car as a smartphone.

            --
            https://www.youtube.com/watch?v=aoFiw2jMy-0 https://soylentnews.org/~MichaelDavidCrawford
            • (Score: 2) by urza9814 on Wednesday March 13 2019, @11:21AM

              by urza9814 (3954) on Wednesday March 13 2019, @11:21AM (#813669) Journal

              Eh, the car would probably be ten or fifteen years old by the time it happens, and if you had the money for such extras then you probably wouldn't still be driving a car that old. So it's probably the guy who bought it used who gets screwed rather than the original owner, which kinda sucks. On the plus side, it's not like you can't just swap out the GPS unit itself -- AIUI a lot (although certainly not all manufacturers) of these things DO kinda follow the Unix philosophy -- there's standards for size and placement and it's usually not too difficult to pull one unit out and drop another one in, as long as it's one of those embedded into the dashboard itself...although on a car that old most people wouldn't consider it to be worth the effort and they'd just use their phone anyway. It's also possible, if there's enough such vehicles out there for the manufacturer to give a damn, that it could be fixed by sticking a CD in the player to update the firmware. Just like they usually do map updates.

          • (Score: 1, Informative) by Anonymous Coward on Tuesday March 12 2019, @10:55PM (1 child)

            by Anonymous Coward on Tuesday March 12 2019, @10:55PM (#813506)

            The common patch was to add 1024 to the week number, since it wasn't going to happen again for another 20 years.

            Worse is this one:
            if week number manufacturing week { add 1024 to week }

            Or this one:
            while week number manufacturing week { add 1024 to week}

            I'm not kidding. Both have been used by major GPS vendors.

            • (Score: 0) by Anonymous Coward on Tuesday March 12 2019, @10:58PM

              by Anonymous Coward on Tuesday March 12 2019, @10:58PM (#813507)

              Edited for >

              The common patch was to add 1024 to the week number, since it wasn't going to happen again for another 20 years.

              Worse is this one:
              if week number greater than manufacturing week { add 1024 to week }

              Or this one:
              while week number greater than manufacturing week { add 1024 to week}

              I'm not kidding. Both have been used by major GPS vendors.

    • (Score: 0) by Anonymous Coward on Tuesday March 12 2019, @05:05PM (3 children)

      by Anonymous Coward on Tuesday March 12 2019, @05:05PM (#813360)

      GPS has always had the ~20-year wraparound of the week counter and has always had provisions in place to deal with it.

      However a lot of devices -- especially very early consumer GPS devices -- likely implemented it incorrectly.

      The real mistake in the GPS specification is making this happen every 20 years. This is too infrequent to discover any problem via testing. In hindsight, it would have been better to make the counter smaller -- say 4-5 bits or so -- so that the wraparound occurs every few months and is more likely to be noticed during a product's development cycle.

      • (Score: 2) by urza9814 on Tuesday March 12 2019, @05:49PM (1 child)

        by urza9814 (3954) on Tuesday March 12 2019, @05:49PM (#813385) Journal

        The real mistake in the GPS specification is making this happen every 20 years. This is too infrequent to discover any problem via testing. In hindsight, it would have been better to make the counter smaller -- say 4-5 bits or so -- so that the wraparound occurs every few months and is more likely to be noticed during a product's development cycle.

        Presumably it needs to be infrequent enough for the device to stay calibrated though....The only way I can see that the device would know how many times it's rolled over is if it's counting them and storing that in some other memory location. So if it rolls over every few months, and you only use your GPS for a long trip you take once or twice a year, you'd have to manually reset the date every single time you turned the thing on, right? Best case it calibrates against an RTC and lasts until the RTC battery dies, but then the poorly designed devices would have a non-removable RTC battery and might be useless in four or five years instead of twenty...

        • (Score: 0) by Anonymous Coward on Tuesday March 12 2019, @06:48PM

          by Anonymous Coward on Tuesday March 12 2019, @06:48PM (#813412)

          Six bits is enough for a year. For an infrequently used device, telling it what year it is isn't a large hurdle.

      • (Score: 0) by Anonymous Coward on Wednesday March 13 2019, @01:37PM

        by Anonymous Coward on Wednesday March 13 2019, @01:37PM (#813708)

        This is similar to an issue we had at my last job. We were testing devices in bulk and had a system to resume testing if we had a power outage. Sometimes, however, the later tests required data that was saved improperly and after the power-loss it wouldn't exist anymore. Every outage we would get a handful of devices that needed to be retested because they looked for data that was missing.

        My solution? Purge the data at every resume point (where testing would resume after a power-loss) so that instead of it being rare, it happened all the time. Then the test developers figured out how to deal with it.

        Admittedly, this was part of a system re-write and I actually fixed the problems in place when I made the change, but no new ones came in. Overall, we stopped seeing reports of issues after power-loss.

(1)