OpenBSD 5.1 Released

[LightBit]

I'm sorry, but I don't understand what you're trying to say. Linux 2.5 is experimental branch.

Windows XP was released in 2001. There was no atheros driver for Linux at the time.

[kraftman]

It's better to do neither on a Desktop. Windows doesn't kill anything, it enlarges the page file. Much better than OOM roulette.

Nope, it's better to do what Linux does, because you will fully use your memory. Windows uses swap without a reason and wastes memory. If Windows doesn't kill anything then it crashes in some scenarios. Btw. did you mean swap file?

[curaga]

Windows XP was released in 2001. There was no atheros driver for Linux at the time.

Hear hear. Me, a linux user, humbly admits that wireless support used to suck back then. Happy?

[kraftman]

Windows XP was released in 2001. There was no atheros driver for Linux at the time.

That's true, but the main point is it's better when OS has control of its drivers. Windows lives mainly, because it's supported by third party members. Linux was in much worse situation, because it rarely had such support (today it is much better, but Windows still rules in this case). In theory Windows can be left alone without support in the future and this is impossible to do with Linux. Nobody will steal its drivers.

[LightBit]

No, it's not. Default is 0, so memory overcommit is not always being done. It's fully tunable. You have proven you don't know much about Linux and your basing on not trustworthy data that usually comes from Linux competitors.

Ok not 100%, but mostly.

Who cares? It was killed by Solaris even without OOM, am I right? It's better to kill some process or to crash?

I believe Linux also has protected memory, which means that process will be killed, if it does something stupid on memory.
There are also other limits.
At least it doesn't kill random process.

I'm interested in mobiles in this case. I don't care about user land, but about kernels.

I'm not interested in mobiles and I do care about userland. GNU is big shit.

Don't you see this is from 2007 and don't you think that current Linux kernel documentation is better place to find information about what Linux does?

Yes, and Linux documentation says: "in MOST situations"

No, package size doesn't matter, because you're not running every file system and driver that package contains. Linux supports many more drivers and file systems than OpenBSD, so the package must be bigger. This should be obvious. Btw. I didn't know that not having modules is something good. I also didn't know having dozens of drivers and file systems is bad. What did you check? vmlinuz, perhaps?

I don't need 100 file systems. What I don't need and is there I consider as a bloat.
I checked vmlinuz which doesn't have any driver since they are in modules as you say, and it is bigger than OpenBSD with all drivers ...
Modules are good, but OpenBSD beats Linux in that aspect without using them.

[LightBit]

Hear hear. Me, a linux user, humbly admits that wireless support used to suck back then. Happy?

No, I'm sad.

[LightBit]

That's true, but the main point is it's better when OS has control of its drivers. Windows lives mainly, because it's supported by third party members. Linux was in much worse situation, because it rarely had such support (today it is much better, but Windows still rules in this case). In theory Windows can be left alone without support in the future and this is impossible to do with Linux. Nobody will steal its drivers.

At that point I completely agree. Hardware should have a good documenatation and public domain driver for at least one os.

[LightBit]

Nope, it's better to do what Linux does, because you will fully use your memory. Windows uses swap without a reason and wastes memory. If Windows doesn't kill anything then it crashes in some scenarios. Btw. did you mean swap file?

If OOM kills web server, server is down anyway. It doesn't matter if kernel still works.

[kraftman]

I'm not interested in mobiles and I do care about userland. GNU is big shit.

Ok, but I was interested. I like GNU much more than not GNU user land.

Yes, and Linux documentation says: "in MOST situations"

Yes, but we were talking about memory over committing and not about killing a process - you agreed above it's a good thing when the process is killed when it does something stupid.

I don't need 100 file systems. What I don't need and is there I consider as a bloat.
I checked vmlinuz which doesn't have any driver since they are in modules as you say, and it is bigger than OpenBSD with all drivers ...
Modules are good, but OpenBSD beats Linux in that aspect without using them.

Yes and I don't need 100 file systems neither and that's why I'm using just one at the same time. As far as I know Arch' vmlinuz have compiled drivers and file systems in (not all of the drivers are shipped as modules in Arch). For example my vmlinuz in Kubuntu is just 4.7MB. I don't see how OpenBSD beats Linux in this aspect.

[kraftman]

If OOM kills web server, server is down anyway. It doesn't matter if kernel still works.

If it kills a web server it's probably doing this to save system from crash. However, in Linux you can decide what processes can and can't be killed by the OOM killer:

It is the job of the linux 'oom killer' to sacrifice one or more processes in order to free up memory for the system when all else fails. It will also kill any process sharing the same mm_struct as the selected process, for obvious reasons. Any particular process leader may be immunized against the oom killer if the value of its /proc/<pid>/oomadj is set to the constant OOM_DISABLE (currently defined as -17).