Google's bouncyballs
-
@tdb said:
Assuming "many" is at least 3, that works to around one in ten billion (10^10) bugs being due to the OS.
One in ten billion crashes, but yes, sounds about right to me. @tdb said:Somehow I doubt your sample size is large enough to draw that kind of conclusions. It would probably take something like the entire tech support history of the whole world to manage that.
No, I very much doubt it. I don't think 10 billion crashes happen in a day, but we're talking a few years rather than decades. I may have overstated the case slightly - 'many' was probably unjustified - but the sample size for my personal experiences must be of the order of ~1,000 cases, without a single OS-caused-crash among them that I can think of. What is it with people being stuck a decade in the past? We're not talking about Win 95 any more, y'know.
-
@evilspoons said:
I believe this is the only reason why people thought Windows Me and Windows Vista were shit. Yes, Vista had some deficiencies that were addressed by the various updates (and are now included in the Service Packs). But... both of these operating systems coincided with a shift in driver model.
I half agree. Vista was shit anyway. It was probably more usable once it crashed.
-
@evilspoons said:
I believe this is the only reason why people thought Windows Me and Windows Vista were shit. Yes, Vista had some deficiencies that were addressed by the various updates (and are now included in the Service Packs). But... both of these operating systems coincided with a shift in driver model. (VXD to WDM for Windows Me, can't remember the names for Vista). The manufacturers provided absolutely terrible drivers that crashed the operating system, and users blamed Microsoft.
I seem to remember reading a study that showed something like 30% of BSODs on Vista in the first year were caused by Nvidia drivers. 30%!!!
I recall the major arguments against adopting Vista being unavailability of drivers for many less common pieces of hardware, poor quality of existing drivers (especially 64-bit) and the heavy weight of the OS. While the driver problems may not be Microsoft's fault, they still make the OS a bad choice - what good is an OS if you can't run it on your hardware? The heavy weight was Microsoft's problem, and was fixed in Windows 7, which is much less demanding.
-
@davedavenotdavemaybedave said:
@tdb said:
Assuming "many" is at least 3, that works to around one in ten billion (10^10) bugs being due to the OS.
One in ten billion crashes, but yes, sounds about right to me. @tdb said:Somehow I doubt your sample size is large enough to draw that kind of conclusions. It would probably take something like the entire tech support history of the whole world to manage that.
No, I very much doubt it. I don't think 10 billion crashes happen in a day, but we're talking a few years rather than decades. I may have overstated the case slightly - 'many' was probably unjustified - but the sample size for my personal experiences must be of the order of ~1,000 cases, without a single OS-caused-crash among them that I can think of. What is it with people being stuck a decade in the past? We're not talking about Win 95 any more, y'know.Can't really prove or disprove the amount, as I have neither the tools, time nor interest to figure out whether any particular crash is Windows's fault. Particularly since I don't even use the OS that often. And I do admit that Windows has gotten better over time, with possibly the most important improvement being the adoption of the NT kernel in all editions. But with that sample size you can only say that the proportion of OS-induced crashes is somewhere between 0% and 0.1%.
There's a lot of grey area here too - how exactly do we define a crash caused by the OS? An invalid memory access or other terminal error happening inside the OS or its core libraries? Neglecting to check a system call parameter and crashing on invalid input? A system call returning slightly bogus data that causes the application to access non-existent memory? A security hole letting in malware that hooks an API function and causes crashes? Eliminating invalid memory accesses, divisions by zero and other similar bad stuff is relatively easy, as there are automated tools to check your code for potential occurrences. However, I dare say that there are other much more serious faults an OS may have that are much harder to detect in advance.
-
@pkmnfrk said:
It works for me, but only if I click on the "Go to Google.com" link on google.ca
I somehow missed this post.
It looks quite nice. More usage will be required to see if it's really awesomer, because I normally press Return which is almost as close to zero time as an interface interaction can get, i.e. I never "click the button"
Good to see they've still got the suggestion box.
-
@davedavenotdavemaybedave said:
Vista was shit anyway. It was probably more usable once it crashed.
My work machine, with Vista, has never crashed on me.
Perhaps a pre-built, pre-installed computer from Dell is less prone.Who knows. It's my only Vista experience.
My current XP machine also never crashed. My previous XP machine crashed once with a BSOD, which due to a USB driver gone rogue for some reason. Repair install -> fix.
I still hold that bad systems are the result of bad management. Exceptions granted, obviously. To put it more profanely: System is fucked? Stop fucking your system.
-
The only thing I've ever had hard-'crash' a Win7 system was a Logitech webcam. Its app went south and refused to quit; ending the process didn't kill it, logging out didn't kill it. I finally told the computer to shut down, and it just sat there - waiting for the applications to close. For whatever reason, Logi's just wouldn't. Had to hard reset it.
That said, I think there was hardware at work there, too: If the webcam is plugged in when the computer boots, it won't post; it hangs on "enumerating USB devices". Unplug it and it continues on its merry way.
-
@PeriSoft said:
The only thing I've ever had hard-'crash' a Win7 system was a Logitech webcam. Its app went south and refused to quit; ending the process didn't kill it, logging out didn't kill it. I finally told the computer to shut down, and it just sat there - waiting for the applications to close. For whatever reason, Logi's just wouldn't. Had to hard reset it.
Yeah, that's an app stuck in kernel call... and 99% of such lockups are caused by buggy drivers.
-
@davedavenotdavemaybedave said:
I half agree. Vista was shit anyway. It was probably more usable once it crashed.
I never had Vista crash or BSOD on me. Granted it was a serious resource hog but it never crashed. I firmly believe the vast majority of crashes are user caused (bad management of the system) or driver issues. Just my 2¢
-
@PsychoCoder said:
@davedavenotdavemaybedave said:
I half agree. Vista was shit anyway. It was probably more usable once it crashed.
I never had Vista crash or BSOD on me. Granted it was a serious resource hog but it never crashed. I firmly believe the vast majority of crashes are user caused (bad management of the system) or driver issues. Just my 2¢
Since the switch to the nt kernel in windows XP, that's been my experience as well: Shitty Hardware -> Shitty Drivers -> Problems.
