You should talk about this with someone who has done a carrier landing or who has had to get airborne again from a runway ( "Touch-and-go").
Wildly, I would say.
I don't know what kind of Doppler equipment is at that field, but I haven't seen any mention here yet of "microburst". This comes to mind:
Not a chance...
No, not possible, feasible, or sensible. The pilot must have the ability to abort and attempt a takeoff during landing (touch and go...) for many reasons including runway obstructions.
What does happen is that pilots screw up. Engins screw up too. So do computers.
Exception checking seems to be the pitfall of a lot of software development. If there's one thing I remember from a lot of the programming classes I had it was that you should always maintain exception traps so if something goes wrong, it doesn't make the whole system crash.
Well you have a point, although there are not too many systems that don't need to respond to any interrupts at any point whatsoever. And what when thing2 fails to complete? I am not at all sure you can write reliable code without some supervisory code that could be called an OS. But you could call it something else and I won't argue :-).
I kind of doubt that, they do use thrust reversers and full throttle to help slow the plane down during landing.
JosephKK
The point about a state machine is that is always completes, one way or another, but it never hangs. There's an ever-downward path through each block of code, like a marble in a pachinko machine, and it's written so that it always reaches the end within a guaranteed maximum time. You only need an OS of you have multiple tasks that have no real end, namely pieces of code that must be suspended and resumed, from that same point, later.
I did a couple of recent gadgets that are state-machine driven.
The main loop just runs a bunch of state blocks that do various things. There are two ISRs, a periodic timer and a serial input interrupt. The periodic timer blinks led's and performs some timing services for the mainline logic, mostly decrementing counters if they are nonzero.
The serial interrupt is a little smarter. It grabs incoming characters, does some editing/translation, and loads a line buffer. When a character is seen, it sets a "command line ready" flag, and one of the mainline state blocks, next time it runs, parses the text and does things. From the outside, it sure looks like it's multitasking, but it's not.
John
On a sunny day (Sat, 26 Jan 2008 02:29:21 GMT) it happened Glen Walpert wrote in :
I wrote a complete CP/M (*) clone in Z80 assembler that uses NO interrupts. Zero, zilch, none. That included the BIOS. The floppy drive was polling... I did not add a timeout loop, so if the floppy was removed while reading, the system would wait... days.. years.... and then simply continue once you re-inserted the floppy. A timeout loop could have been added, no need for interrupts.
Reliability 'could' be better, as system behaviour is more easily predicted in a system without interrupts. Not that I hate interrupts, use them all the time.
Microburst killing engines?
-- Keith
At a minimum, each engine controller will be fed with an 'air/ground' signal from separate sensor channels (At least that's the way we did it before the accounting department took over). For critical signals, it will be triple-redundant inputs so that a single fault can be detected and corrected.
This whole incident is going to be a big mess. Everything from equipment redundancy, maintenance and flight procedures to the analysis of things like bird strikes is done to minimize the probability of such an event happening. That it did happen means people are going to have to go back over their design assumptions to see if this was just one of those 'one-in-a-million' occurrences or if something got overlooked.
-- Paul Hovnanian mailto:Paul@Hovnanian.com ------------------------------------------------------------------ At some point it becomes necessary to behead all the architects and begin construction. -- Abi-Bar-Shim (Project Mgr. - Great Pyramid)
[snip]
One at a time. A failure that could shut down both engines violates numerous design criteria.
-- Paul Hovnanian mailto:Paul@Hovnanian.com ------------------------------------------------------------------ 2 + 2 = 5 for extremely large values of 2.
Nope. The closest Windows gets to the flight controls is through various I/O ports used during manufacturing and maintenance procedures.
There is a problem with this. Back when I was at Boeing, we had a hell of a problem convincing some of the engineers and techs that the same laptop they were using to upload firmware updates to various systems SHOULD NOT be taken off the property. Like home, where the kids can goof with it, download virii, spyware, etc.
-- Paul Hovnanian mailto:Paul@Hovnanian.com ------------------------------------------------------------------ A vacuum is a hell of a lot better than some of the stuff that nature replaces it with. -- Tennessee Williams
I have my work laptop at home right now. Of course, I never gave the kids my passowrd.
Or not that close. The Pratt&Whitney guys I know abuse me for running Windows. They use Linux on desktops, and RT Linix for engine testing.
John
Heh, the Woodward guys I've seen program their testing software in VB.
I think some of their fuel controls get on P&W engines...
Tim
-- Deep Fryer: A very philosophical monk. Website @
87C752 by any chance? an engineer I once worked with had this particular problem. 2 RTI's mostly got rid of it, and 3 seemed a good enough fix to ship 10,000 units. He said he really didnt believe it when he found it....
Cheers Terry
LOL! I had a programmer write code for a user interface module. 4x20 LCD, 3 LEDs, 5 buttons and a MODBUS interface (it was a master). Used a
8951 with 64kb flash and 4kb SRAM. He wrote an RTOS, which wasnt a bad start, although not really necessary. but ran out of RAM, and used most of the FLASH, which was supposed to store drive config data - I expected the code to take 4kb max.turns out his C compiler was using 32-bit numbers for BOOLEANS! and it got a lot worse. It was a really fancy RTOS, multi-threaded & multi-tasking. And got binned. I learned a lot about writing clear specs from that debacle. The replacement programmer got all the code in 8kb including the GNU real-time debugger. and used a fraction of the RAM.
Cheers Terry
"T" skrev i en meddelelse news: snipped-for-privacy@news.east.cox.net...
It gets better: Microsoft Excel is used almost exclusively to model financial products - like Structured Investment Vehicles, Credit Defaults Swaps and whatnot. I know someone who does this kind of modelling he says that Excel "sometimes" produces wierd results with the circular references they use. So they run it again and it's "ok". Ah, the optimism!
Subprime reminds me of that Robocop skit: "Bwahahahahahaaaa: I'll buy that for a dollar!"
"Jan Panteltje" skrev i en meddelelse news:fncd1n$hma$ snipped-for-privacy@aioe.org...
Or maybe the French bank held/is holding a lot of American structured investment crap and rather than having to admit that they were: First stupid enough to buy the crap, Second also lost money on it they decided that the "good of all" would be served by having one guy take the rap in return for - perhaps - a Swiss chalet and a nice retirement trust fund in an equally discrete Swiss bank.
"Tom Del Rosso" skrev i en meddelelse news:479911dd$0$6364$ snipped-for-privacy@cv.net...
It's COTS, battle ships are outdated anyway and you get *twice* the budget that way: first to make the mess, then to fix it! Everybody wins - it's the government after all and governments solutions will always be: MORE of whatever it was that did not work in the first place.
Microsoft has the most effektive slides, I would imagine.
After all: "The purpose of Powerpoint presentations is to stop the electrochemical activity of the brain at which point decisions are made". (scott adams, the joy of work)
When I was young and naiive I hired a guy to do a fairly simple parts-list thing. He said "if first I write a database manager, your probem will be trivial and just take a day more." He fiddled with the db manager for months and never finished anything. So it dawned on me that he didn't ever give a damn about my problem, he just wanted to fiddle with databases. Similarly, there are lots of programmers who aren't interested in your dinky buttons and LCD's, they really want to get into context switching and schedulers and dynamic memory assignment algorithms.
Fire them.
John
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here. All logos and trade names are the property of their respective owners.