An interesting article has been making its way around the internet the past few days, titled "Top 10 Usability Highs Of Mac OS". Mac OS X indeed does some things very, very right, just like many other operating systems and graphical environments do some things very, very right. The issue with the list of the article in question is that many of the items on the list are not exactly examples of "Usability Highs" at all.
To read all comments associated with this story, please click here.
Member since:2005-07-20
Linux is awful about kernel panics while running graphics. Since I run bleeding edge -mm series kernels on my laptop I have seen many of them.
What happens is that the graphical desktop just stops. No mouse movement, no screen updates. The Caps Lock key starts to blink and that is about it. There is no way to switch back to a text console to view the kernel panic.
If you have configured a serial or network console then you get the panic message as output, but that is the only way.
I haven't used it myself but I've read that kdump can get around this. kdump works by kexec'ing a special kernel which is then used to copy the kernel panic and a memory image to disk. This kdump kernel can also reinit the graphics hardware, sometimes.
Edited 2008-08-14 01:21 UTC
Member since:2005-11-14
I've been running Linux for years (guess about 4 years as my primary desktop), and I've never had a kernel panic. If you're using some sort of custom kernel, you can't generalize Linux and say that Linux is very bad about such things. Seems to be more of a problem with your setup than what the overwhelming majority of people do.
Member since:2006-11-13
Just to make sure everybody knows.... kernel panics are expected in bleeding edge software. No matter the OS.
Do you know if Linux has something similar to what I used in OS/2. Back in the days when I used OS/2, I had a watchdog daemon running and it was link to any external switch - I simply used my joystick. If a system crash occured that froze the keyboard and mouse, I could click the joystick button and the watchdog daemon kills the hanging program returning control back to me. This was awesome. At the moment (under Linux), I go to my co-workers PC, SSH into mine and kill the process myself. I would love the watchdog/joystick feature under Linux though!
Member since:2005-12-18
Hey... just use the BSOD of xscreensaver... there you have a number of hang ups of different systems.... including BSODs for different versions of güindous (of course). I think it's the safe's way to see a kernel panic.... other than use -mm (or any other non-stable kernel) like slynx does. :-)
Member since:2005-07-06
ignorance is a bliss, right? ...
And a laundry list of 'codes' helps the end user - how? Microsoft have already demonstrated that they refuse to make available the list of error codes for BSOD's when they occur; heck, Microsoft has gone so far as to have automatic reboots instead of actually showing a BSOD.
The laundry list of Linux codes - that helps the user how? again, there is no standardised error codes, nor is there a single place to look up those error codes.
Screaming about the wonders of actually seeing a dump from a kernel panic may get your loins excited, but for the average end user - it is all Greek to them.
Member since:2006-02-05
Member since:2005-07-06
It doesn't help the end user - but if the end user has to bring in a support tech, they can usually decipher the error code (even if "deciphering" means "search for it in Google", which is usually the case with Windows BSODs).
IMO, the ideal would be a terse "an error has occurred" message - with an option to show more details.
That said, both of my biases are showing. As a result of doing a lot of technical support, I believe that computer software/hardware should be evaluated based on how easy it is to diagnose and fix problems when they occur (and not just how well it works in normal situations).
And as a BeOS devotee, the "ideal" I described is essentially how things work(ed) with that OS: you get a "Welcome to Kernel Debugging Land" message, with a prompt that lets you run a stack crawl, etc, if you need more details.
Member since:2005-07-06
It helps me, and I'm sure lot's of other people that look after Windows systems. Generally speaking, the first point of call for a bugcheck code is support.microsoft.com, then Google.
And the bugcheck code get's written to the event log, even if the system reboots (unless the system really locks, then there is no evidence).
Member since:2005-07-10
has anybody experience a kernel panic when using linux on a desktop? how does it look like?
Well, I've seen it many years ago, but true, it hasn't kernel paniced much for me. It drops to console and writes "Kernel paniced." and then you have to reboot.
ignorance is a bliss, right? ...
In fact, OSX displays a requester after reboot, stating that a serious error occurred. You can then click for more information. You can then study the debug output and send it to Apple if you wish, along with a report of what you did to provoke the error.
WindowsXP lets you do something similar.
This reminds me of the amusing error handling system that Ubuntu had once with such error requesters for crashed programs, which also supports sending debug information back to the developer. Except that it would tell you that the application that crashed was not supported by this error system, so your report would not be sent anywhere. It's been a while, but I hope they fixed that.
Member since:2007-02-17
Linux does this as well as of a few a kernel version ago. Because kernel panics are rather rare on Linux it took longer than necessary to implement something as crucial as kernel dumps. BTW, if I'm not mistaken their is way to configure the kernel panics to send a some sort of message to you framebuffer, I think someone was working on that a while back. It would be nice if when a kernel panic happened in Linux (which is rather rare. The last time that happened to me was in 2004) that it would display an error report similar to OSX with bug buddy or something. I think its an elegant way of handling the issue rather than a blue screen with gibberish on it.
Member since:2005-11-14
Have had Linux kernel panics on several occasions. When using X11 it's just a frozen screen, and, unless you use a wireless keyboard with no leds, a blinking scroll lock key (or was it num lock? I forget). It's almost always due to fglrx or alpha/beta quality ATI drivers.
Also, that list in the article is a cringeworthy applejizzfest. "Informative error reporting" (4)? Total bullshit. When there's something wrong with OS X, you just don't know what to do, and if you ask on a forum you'll be told to "repair permissions" (which never, ever works). This is, of course, due to Apple hiding the "technical details" (5) in ways that make the software totally opaque to the end user (intended to make her/him more of a consumer than a computer user, I suspect). Claiming "the system seems to be more responsive", for OS X, is a blatant lie. OS X is fine, but not quick or responsive.
Member since:2008-07-15
I have had quite a few panics recently in Ubuntu 8.04 connect to Wifi. At least once a day when I use it. The whole system freezes and requires a hard reboot. I have yet to experience a panic in OS 10.4.11 and have been using it for over two years.
Edited 2008-08-15 19:07 UTC
Member since:
2006-11-14
has anybody experience a kernel panic when using linux on a desktop? how does it look like? ..i have been using linux for years now and have never seen a kernel panic ..it hang a few times that required a hard reboot but it has never(so far) dropped me anyway and inform me of a panic ..how does linux handles such situations? do they occur?
A funny but still nice example of Apple’s attention to detail. On the rare occasions when Mac crashes, it still does so in a respectable manner. Usability-wise it’s not perfect, since it doesn’t let the user know what went wrong and only asks the user to reboot the system. Still, beautiful and elegant.
ignorance is a bliss, right? ...