Crash terminology

Question about which tools to use, bugs, the best way to implement a function, etc should go here. Don't forget to see if your question is answered in the wiki first! When in doubt post here.
User avatar
Colonel Kernel
Member
Member
Posts: 1437
Joined: Tue Oct 17, 2006 6:06 pm
Location: Vancouver, BC, Canada
Contact:

Crash terminology

Post by Colonel Kernel »

This is very much a non-technical question. :) Which synonym for "crash" do people prefer for describing situations where the kernel has to halt the machine (e.g. -- double-fault). I mean purposefully halt it (i.e. -- lock-ups do not count).
Top three reasons why my OS project died:
  1. Too much overtime at work
  2. Got married
  3. My brain got stuck in an infinite loop while trying to design the memory manager
Don't let this happen to you!
Dreamsmith

Re:Crash terminology

Post by Dreamsmith »

I'm rather fond of calling something like that an "oops", but I probably just say "crash" or "crashed" 80% of the time.
Colonel Kernel wrote:I mean purposefully halt it (i.e. -- lock-ups do not count).
Of course, my OS never purposefully halts. If all else fails, as a last resort, I attempt a reboot. Better a service is unavailable for a few seconds than it being unavailable until someone manually resets the machine...
User avatar
Solar
Member
Member
Posts: 7615
Joined: Thu Nov 16, 2006 12:01 pm
Location: Germany
Contact:

Re:Crash terminology

Post by Solar »

"Guru Meditation" sure is what I grew up with, but it doesn't really fit here as it isn't a "double fault", or an intentional halt. As far as I understood it, the Amiga's Guru Meditation was kind of a ROM-driven "post-desaster" debug message. The kernel didn't "handle" them, and there was - to my knowledge - no way of "catching" them.

That being said, "Panic" would be my choice, and I second Dreamsmith - the system should reboot, not halt.
Every good solution is obvious once you've found it.
bkilgore

Re:Crash terminology

Post by bkilgore »

I had a BSOD on my XP box today for the first time in a long time. It was annoying because I wasn't home, and had been VPN'd into it, and I couldnt do anything until I went home and restarted it. I would have much rather windows just restart for me, and tell me about it in a little window when it comes back up.
User avatar
Candy
Member
Member
Posts: 3882
Joined: Tue Oct 17, 2006 11:33 pm
Location: Eindhoven

Re:Crash terminology

Post by Candy »

Personally, I like a system to be both reliable, always-on and crash-proof. If it thus still crashes, it should display something that's obvious to all who can look at it, including my mom, sister & girlfriend. They don't know what a BSOD would be, they have learned after a while that the blue screen means your computer crashed (and thus reset the computer when I use DOS edit :(). thinking about this as a noob, guru meditation seems very weird, panic seems like something a computer can't do (you've never seen one running around frantically, now have you?). Crash might indicate the good thing, but is a little exaggerated. Your car crashes if it hits a wall. When the engine fails, it's a failure. When you make a mistake you've made an error.

My votes would thus be for system error or system failure.

As a nice sidenote, they're also usually preferred in movies where the point is that most should understand it.

System Failure would be my idea of a system that can't do anything anymore, a System Error is something like a filesystem that can't be read (not the root FS, something like a floppy disk or such) or a network connection failure.
mystran

Re:Crash terminology

Post by mystran »

I believe "crash" and "bluescreen" and "oops" and whatnot is a kernel crash that get's catched by kernel and reported with a register dump while I prefer using "crash" to mean a crash so fatal that not even the normal panic-handler get's called.
IRBMe

Re:Crash terminology

Post by IRBMe »

I had a BSOD on my XP box today for the first time in a long time. It was annoying because I wasn't home, and had been VPN'd into it, and I couldnt do anything until I went home and restarted it. I would have much rather windows just restart for me, and tell me about it in a little window when it comes back up.
Windows XP is set to restart by default is it not? Either way you can change that behaviour:

Right Click "My Computer" -> Properties -> Advanced -> Startup and Recovery.Settings -> "Check" System Failure.Automatically Restart

And yes, I agree restarting is better than halting, and I use the word panic or just crash.
User avatar
Colonel Kernel
Member
Member
Posts: 1437
Joined: Tue Oct 17, 2006 6:06 pm
Location: Vancouver, BC, Canada
Contact:

Re:Crash terminology

Post by Colonel Kernel »

I forgot "bugcheck" -- the "official" term for BSOD. Not very intuitive IMO.

So far, I like "system failure" the best, at least in terms of accuracy. So how would one turn that into a verb? "Crash" and "panic" both have the advantage of being usable as nouns or verbs depending on the context (e.g. -- Cause a kernel panic by calling panic()). FailTheSystem()? ;)
Top three reasons why my OS project died:
  1. Too much overtime at work
  2. Got married
  3. My brain got stuck in an infinite loop while trying to design the memory manager
Don't let this happen to you!
User avatar
Candy
Member
Member
Posts: 3882
Joined: Tue Oct 17, 2006 11:33 pm
Location: Eindhoven

Re:Crash terminology

Post by Candy »

Colonel Kernel wrote: I forgot "bugcheck" -- the "official" term for BSOD. Not very intuitive IMO.

So far, I like "system failure" the best, at least in terms of accuracy. So how would one turn that into a verb? "Crash" and "panic" both have the advantage of being usable as nouns or verbs depending on the context (e.g. -- Cause a kernel panic by calling panic()). FailTheSystem()? ;)
My personal short name will probably become sysfail, it's short & pronouncable, and it still kind of says the same as system failure. Only problem is, most computer nerds will look at you blank and ask you what it is :)
bkilgore

Re:Crash terminology

Post by bkilgore »

IRBMe wrote:
Windows XP is set to restart by default is it not? Either way you can change that behaviour:

Right Click "My Computer" -> Properties -> Advanced -> Startup and Recovery.Settings -> "Check" System Failure.Automatically Restart

And yes, I agree restarting is better than halting, and I use the word panic or just crash.

Yes, it is set to that by default. And mine is still set that way. Yet, when I got home, it was just sitting there with a BSOD. Go figure....
User avatar
Candy
Member
Member
Posts: 3882
Joined: Tue Oct 17, 2006 11:33 pm
Location: Eindhoven

Re:Crash terminology

Post by Candy »

bkilgore wrote:
IRBMe wrote: And yes, I agree restarting is better than halting, and I use the word panic or just crash.
Yes, it is set to that by default. And mine is still set that way. Yet, when I got home, it was just sitting there with a BSOD. Go figure....
How do you test your crash routine? IE, how do you make the OS crash? By definition, it should not be possible by any route you already know, or you're outputting crap. So, in a way, you can't test whether it works :)

In a related story, do you have PM switched on in your bios? that might cause it to not work :).
Dreamsmith

Re:Crash terminology

Post by Dreamsmith »

Candy wrote:thinking about this as a noob, guru meditation seems very weird, panic seems like something a computer can't do (you've never seen one running around frantically, now have you?). Crash might indicate the good thing, but is a little exaggerated. Your car crashes if it hits a wall. When the engine fails, it's a failure. When you make a mistake you've made an error.
Well, no, a computer cannot literally panic. Nor can a monitor contain a desktop (although it may sit on one), nor can a program possibly open a window without some sophisticated robotic equipment attached to the computer. For that matter, a hard drive cannot possibly contain a file or a folder (unless it's a very large hard drive and you've cut a slot in it to stuff the paperwork in), trees cannot grow inside any program, and no one, certain not my computer, can pull itself up by its own bootstraps. My computer isn't even wearing boots!

Given this, I find the argument against the use of "panic" flys in the face of general computer science practice. It is a world of metaphors, none of which make any sense if you take them literally...

And as far as the car goes, yes, when you make a mistake, you've made an error. Which sometimes results in a crash. "An error" to me means I select "italics" from the "Style" menu and my text turns bold. That's an error. If instead, I select "italics" and the program displays a memory dump and gets killed or quits, that's a crash. If the program doesn't keep running, it's more serious than simply an error. I don't think "crash" is over the top in this case, it's a perfectly appropriate metaphor for an error you can't just say "oops" but keep driving afterwards.
User avatar
Colonel Kernel
Member
Member
Posts: 1437
Joined: Tue Oct 17, 2006 6:06 pm
Location: Vancouver, BC, Canada
Contact:

Re:Crash terminology

Post by Colonel Kernel »

I just can't get past the fact that "crash" and "panic" sound so undignified... ;)

Although now I'm leaning towards flagrant system error. ;D
Top three reasons why my OS project died:
  1. Too much overtime at work
  2. Got married
  3. My brain got stuck in an infinite loop while trying to design the memory manager
Don't let this happen to you!
User avatar
Candy
Member
Member
Posts: 3882
Joined: Tue Oct 17, 2006 11:33 pm
Location: Eindhoven

Re:Crash terminology

Post by Candy »

Dreamsmith wrote:
Candy wrote:thinking about this as a noob, guru meditation seems very weird, panic seems like something a computer can't do (you've never seen one running around frantically, now have you?). Crash might indicate the good thing, but is a little exaggerated. Your car crashes if it hits a wall. When the engine fails, it's a failure. When you make a mistake you've made an error.
Well, no, a computer cannot literally panic. Nor can a monitor contain a desktop (although it may sit on one), nor can a program possibly open a window without some sophisticated robotic equipment attached to the computer. For that matter, a hard drive cannot possibly contain a file or a folder (unless it's a very large hard drive and you've cut a slot in it to stuff the paperwork in), trees cannot grow inside any program, and no one, certain not my computer, can pull itself up by its own bootstraps. My computer isn't even wearing boots!
If you ask any given tech support person what the most stupid kind of errors are, it's misunderstanding a metaphor, IE, trying to click a mouse on the screen (click on the button), opening a physical window when asked to open a virtual one, trying to save something to a physical file (and then put it in a filing cabinet), photocopying floppy disks, stuff like that.

The metaphors are to make the idea clear to people that already know what it is. If you place metaphors to people with different backgrounds, they're completely misplaced.

Aiming for most people, including noobs, thus totally removes the option of making anything clearer with a metaphor. Having your computer panic gives me the idea that it's about to rip it's hair out and run frantically around the room (the physical computer box). An oops seems like something that's correctable, if I oopsed on a piece of paper while writing, I can white it out and try again. If/when the Linux kernel oops, it's pretty much dead.

Calling each time the OS goes down a computer crash gives a problem in finding a term for a REAL crash. I've had numerous counts where I had a harddisk that was crashed (physically) and I was given the advice to reinstall windows. The hd never contained windows, nor could it install on it because the system wouldn't recognise it. I couldn't explain to them without using very long sentences what was wrong, because somebody thought using the "crash" metaphor would be a nice idea. The computer did NOT crash if you can fix it without paying money for new parts. If you slip off road you might have a dent in your ego, but if you can get the car back on the road without much rpoblems (say, a tow truck or a second car), and then drive along without any problem, you won't refer to that as a car crash. There goes the metaphor even.
Given this, I find the argument against the use of "panic" flys in the face of general computer science practice. It is a world of metaphors, none of which make any sense if you take them literally...
The world of general computer science is why most people still look at a computer as an unreliable device that talks some alien language, only people transformed to aliens can understand.
And as far as the car goes, yes, when you make a mistake, you've made an error. Which sometimes results in a crash. "An error" to me means I select "italics" from the "Style" menu and my text turns bold. That's an error. If instead, I select "italics" and the program displays a memory dump and gets killed or quits, that's a crash. If the program doesn't keep running, it's more serious than simply an error. I don't think "crash" is over the top in this case, it's a perfectly appropriate metaphor for an error you can't just say "oops" but keep driving afterwards.
You can just switch the computer off, switch it back on and start again. If you car crashes you can't possibly hope to drive it in the same form again, at least not without serious undenting and possibly replacing.

Also being the unvoluntary hardware call-him-when-it-doesnt-work guy, I strongly object against using "crash" to refer to some piece of software malfunctioning. When your head screws up your thinking, you've drunk too much alcohol. When your limbs fall off, you might have crashed or died. There's quite a lot of difference between the two. If I start saying that I crashed when I drank 2 bottles of beer, everybody is going to look at me blank and take away the second bottle, with the idea that I must've had 20 beforehand.
Dreamsmith

Re:Crash terminology

Post by Dreamsmith »

You've actually made a pretty good case for using the word "crash", just using it appropriately. If a program (process, technically) has an error, I might still continue working in it, but if it crashes, it's no good anymore -- the process is destroyed and disposed of, and I need to start a brand new process by rerunning the program before I can continue working. Likewise, if the operating system crashes, I need to reboot the computer, the current OS in memory is no longer salvagable -- it's destroyed beyond repair, generally can't be saved, nothing short of a reboot can get me working again. If the file system crashes, it needs to either be repaired, or the partition reformatted and restored from backup. If the hard drive crashes, it pretty much needs to be replaced. You haven't really presented any good reasons why any of these circumstances shouldn't be called a crash, you've just noted in none of this circumstances should you say your "computer crashed", and I agree completely. Be accurate -- was it a program crash? An OS crash? A filesystem crash? A hard drive crash? Use the term appropriately, and there should be no confusion. Whatever crashes in each case is too munged to continue using, requiring either major repair or replacement. Just like a real crash. As far as metaphors go, this is a pretty good one.
The metaphors are to make the idea clear to people that already know what it is. If you place metaphors to people with different backgrounds, they're completely misplaced.
...
The world of general computer science is why most people still look at a computer as an unreliable device that talks some alien language, only people transformed to aliens can understand.
Indeed. The alternative is to invent new words, but in either case, whether its words the person doesn't know or words they do know but don't understand in this context, it's jargon, and not understandable to them. Computer science isn't unique in this. Just try to discuss the epistemic virtues of particular doxastic attitudes as they relate to a priori or a posteriori propositions with a non-philosopher and see how far you get. Or, even worse, start talking about the color or spin of subatomic particles (that do not, literally, have color, nor do they spin -- these terms mean entirely different things to a quantum physicist).

The point being, all specialities have their own vocabularies, whether it's words specific to the field (e.g. epistemic, a priori, etc.) or words from the general language conscripted for new uses (e.g. color and spin, or window and file), no one who isn't familiar with that speciality's use of terms is going to understand what is being said. Nor, frankly, would they likely understand if you tried to avoid jargon and explained it while trying to avoid the use of such terms. No argument about how this mystifies a field for people is going to be convincing, because simply changing to a new set of words isn't going to make the discussion any less mystifying. You can't suddenly make complex topics simple just by substituting words. No matter what you do, if someone unfamiliar with the use of the terms takes them out of context, the meaning will be bungled. This isn't an indictment of the words used, it's just the way communication works (or doesn't, as the case may be).
Post Reply