News for hackers. Stuff that matters.
Re:Multiple Boeing 787s in China experienced GPS 20 years ro... by labster on Sunday April 07, 2019 @08:48PM (#19601034)
At the [dead] commenter: Timekeeping is an integral part of how the Global Positioning System works, to the extent that relativistic time dilation is a factor. You simply cannot use a GPS receiver with a broken time implementation, at least not without significant accuracy issues.
Re:Multiple Boeing 787s in China experienced GPS 20 years ro... by myself248 on Sunday April 07, 2019 @09:57PM (#19601329)
But internal time is not the same as display time.

Internally, the week counter just rolled over. That's fine; all the almanac data is handled in that same format internally. Some receivers may need a reset when this counter rolls, and may need to relearn the almanac/ephemeris from the sky, but once that's done, Sunday of week 0 is just the same as any other Sunday of any other week 0.

Navigation works just fine because it's all internally consistent.

But display time, ahh, now that's where things get interesting. Sunday of week 0 could be displayed as 2019 April 7, or it could be displayed as 1999 August 22, or it could be displayed as 1980 January 6.

Knowing which epoch you're in is actually considerably trickier than working consistently internally, since there's no information in the satellite signal to tell you this, by design. It comes from applying an external offset. Receivers have lots of ways to do this, and quite a few simply have it baked in and cannot handle a rollover, so as far as they're concerned, week 0 is in 1999, period. They'll navigate just fine but they'll display the date wrong, because they're applying the wrong offset.

Re:Multiple Boeing 787s in China experienced GPS 20 years ro... by joelhaasnoot on Monday April 08, 2019 @06:36AM (#19603894)
The easiest way of doing this is to assume time will never travel and store the latest computed date right? If you've seen 2001, it's never going to be 1999 again, so it must be 2019.

I understand that will require either flash or battery backed RAM, etc to keep this stored, but seems relatively trivial in most applications... (side effect would be that if the backup battery is dead or all batteries are disconnected, you would revert to 1999)

Re:Multiple Boeing 787s in China experienced GPS 20 years ro... by inferiorhuman on Monday April 08, 2019 @08:07AM (#19604409)
Unless someone manually sets the date incorrectly or gets the timezone wrong (assuming the date isn't stored as UTC).
Re:Multiple Boeing 787s in China experienced GPS 20 years ro... by joelhaasnoot on Tuesday April 09, 2019 @05:03AM (#19612922)
Yeah, you can only use the GPS source for that flag.
Re:Multiple Boeing 787s in China experienced GPS 20 years ro... by labster on Sunday April 07, 2019 @11:33PM (#19601768)
Cool, see, I knew this was worth talking about. While I've learned about GPS as a way to gather meteorological data[0] via radio occultation, I never had to deal with the raw data stream myself. Thanks for your expertise.

[0]: https://en.wikipedia.org/wiki/Constellation_Observing_System...

Re:Multiple Boeing 787s in China experienced GPS 20 years ro... by PaulHoule on Sunday April 07, 2019 @10:25PM (#19601460)
Funny enough my Tom Tom made some noise and seemed to boot with a different sequence than usual, in particular it insisted that I remove the USB cable that I use to power it from the car. I did that, and it found the signal in 50 seconds and worked normally, correct date and everything.