Slashdot Mirror
Talking With 2.0 Kernel Maintainer David Weinehall
Jeremy Andrews writes "While the recently released 2.6 Linux kernel is all the rage these days, the much older 2.0 kernel is still alive and kicking. KernelTrap has interviewed David Weinehall, the maintainer of the 2.0 Linux kernel. David became the 2.0 maintainer in December of 1999, after Alan Cox moved on to work full time on the 2.2 kernel. In this interview David talks about what's involved in maintaining the 2.0 kernel, who uses it, when we can expect the impending release of 2.0.40, why you should upgrade (if you're still running 2.0.39), and more."
It's like the people that still use Windows 95, sure it's not as new and flashy, but it still runs on the crappy hardware from years ago... And sometimes, that's all you need.
Isn't it great? One of the best things with free software is that anyone is free to maintain and support it for as long as they wish. Compare to say, NT 4.0, which is perfectly capable for some tasks, but users are forced to switch because MS cuts support (read: no more security updates.)
.: Max Romantschuk
Yes, it is. Especially when considering how many untalented suits (posing as programmers) are currently employed out there. And here I see someone maintaining the Linux kernel and he can't find a job...
What would you rather have - some old NT kernel that nobody even thinks about or 2.2 linux kernel actively updated for security issues, even if it is a 14 year old?
I rest my case.
Take the cheese to sickbay, the doctor should see it as soon as possible - B'Elanna Torres, "Learning Curve"
No, sir. By 2.0.40 pre1 the exploit is fixed. So, currently there is no exploits at all... zilch... nada.
Even if what you say is true, theres no reason why a proprietary solution makes more sense. If I'm putting the life in the hands of a computer, I damn well want to be able to know about every piece of code that computer is running.
But really, Linux isn't a consumer OS. Linux is a kernel frequently used in consumer OSes. This means its been tested by a _lot_ of people, and is that much more stable as a result. There is nothing particularly consumerish about the kernel itself (certainly not the 2.0 kernel), and aerospace companies can easily just configure without consumer-oriented features.
I fail to see why you think open source software is somehow less able to protect human life than proprietary software.
I think there is a difference however.
I mean I understand what you are saying, but if you were given the choice to upgrade that old copy of Windows 95 to say Windows 98 or better (I use that term loosely) for FREE (and assuming it still ran on the crappy hardware from years ago) then I am sure you would jump at the chance.
Thats kinda why I find it more surprising that people use the older versions of the kernel, considering its not costing you more than a few minutes time to download the latest tarball from your local mirror, and setup a new kernel!
"Hey! Unless this is a nude love-in, get the hell off my property!!"
and often with an "I don't care" type of lack of concern.
Nope. Actually, many people developing open source software do it BECAUSE they care for quality. They're trying to find the optimal solution for a problem, even if it takes some time. Compare that to someone working at a software company, constantly chased by deadlines and stupid bosses breathing down his neck. Now THAT is an environment which produces "I don't care" attitudes.
And since the original poster said that the open source software they use undergoes the same rigorous tests as the proprietary stuff, I really can't see a problem.
Sorry, no dice there. Win98 had no more stability and was all but Win95b with the bloat of the active desktop required . Win98 for free would be a curse to anyone still running Win95. I hated it when I "upgraded" to Win98 on my p120 because every clock cycle mattered. I went back to Daddy Blue Screen because at least he was faster.
When I tell an object to delete this, am I killing it or telling it to kill me?
Except that 2.0.xx kernel is still getting updated with security patches and bug fixes, thanks to David; while Windows 95 is not secure, and any existing or future vulnerabilities, bugs, leaks, data loss, etc. will not be fixed by Microsoft.
On an unrelated note, David says:
Then I had a hard-disk crash in January (yes, an IBM DeathStar, of course... Heed my advice, never buy one!). While I had backups and of most of my stuff, I didn't have a backup of my latest kernel-tree
Those DeathStars are nothing but trouble. Mine would freeze the system every once in a while (with the HD activity light on), refuse to boot and make funny noises. Looking back, it looks like ReiserFS saved my data quite a few times before I installed a new Barracuda. 0 problems since. Maybe he could port ReiserFS to 2.0? Not that I see any reason to do so.
How come that such a skilled person with enough references can't find a job?
A consumer OS (linux is a consumer OS, BTW) should NOT be controlling fly-by-wire... human life is at risk.
So a non-consumer OS should be used?
Here is an example of custom software causing a crash.
With aircraft systems becopming more and more complex, it does give one pause as to what could happen if there was a "blue screen" or "kernal panic".
IMHO, it doesn't matter what software is used. If it is engineered poorly, it will perform poorly. One would hope that any critical system that relies on software would "fail gracefully", but with budget pressures on a project anything can happen.
A well-tested "consumer" OS can be a lot better than completely untested custom software.
Poor software, like poor concrete, will crumble, no matter where it comes from.
Take the cheese to sickbay, the doctor should see it as soon as possible - B'Elanna Torres, "Learning Curve"
Linux does not do the bare minimum! It does FAR FAR more than a fly-by-wire system would ever need. So,you say, we just disable that when we do "make config". Ah, but what if there are interactions with code which you've just disabled? (It happens. It shouldn't but it does happen).
And don't give me any shit about Linux being tested by millions of people. Yes, it's tested by millions of people, but they are almost all testing different configurations, different code bases (think distribution patches here). Furthermore, Linux contains millions of lines of code. Don't tell me that millions of people testing millions of lines of code = no bugs. The fact is that there are still lots of bugs in Linux.
I would have thought a fly-by-wire system would REQUIRE a hard realtime OS -- we don't want an interrupt stalling (pardon the pun) processing while the system is taking corrective actions to compensate for weather conditions (winds, etc.), do we? Linux is NOT EVEN CLOSE to being hard real time. (Neither is RTLinux).
You mentioned somewhere that Linux is FREE as some great property of Linux. Yes, it's free. However, QNX (which would seem much more appropriate for the application) is a pittance compared to the cost of the hardware it's running on. A plane costs, what, millions of dollars? Who the hell cares if the OS costs $1,000 or even $10,000? It's nothing compared to the total cost of the plane.
HAND.
These things can actually be good things for a programmer though.. Most of the best programmers I know, and we're talking real code wizards here, lack Comp Sci degrees, aren't very social, and have big egos amongst their type.. but they do the work and don't complain.
I disagree. That type of programmer is the kind who has a hard time doing things any other way than his way. If management says "this NEEDS to be done this way", he's the one saying "that's stupid because I wouldn't do it that way. Therefore, no one else would do it that way." You tend to see this attitude a lot on Slashdot, usually when someone points out something Linux can't do but Windows can. To which, the Linux zealot/anti-social programmer will reply: "so what? That's stupid. No one would want to do that anyway."
These people are not necessarily the best people to have on a team.
If I'm putting the life in the hands of a computer, I damn well want to be able to know about every piece of code that computer is running.
No offense, but I'm willing to bet that you've broken this axiom many, many times... possibly without even realizing it. Do you know about every piece of code that the computer in your car is running? Hell, do you even know about every piece of metal that's in your engine? There are tons of possible malfunctions in cars that could easily cause the death of the driver, or at least a major accident.
Not to mention that no one has the time to personally look over the source code to an airplane's systems before they're willing to take that business trip.
It would be nice to know that you have the option to look at all the code if you want, but I'd be willing to bet that if there were any major problems in there that Boeing or Chevrolet's engineers didn't see, you probably wouldn't see it either.... perhaps if you were actively using the code yourself, but certainly not on a cursory glance anyways.
Besides, if there is going to be a mistake, it's probably more likely that it's due to human error than some computer glitch, so I hope you're checking up on the pilots credentials every time you board a plane, and check a driver's license every time you get on a bus or taxi.
-"One machine can do the work of fifty ordinary men. No machine can do the work of one extraordinary man." -EH