Re: Low Latency Patch

Robert Dinse (nanook@eskimo.com)
Sat, 1 Jul 2000 05:28:28 -0700 (PDT)


On 1 Jul 2000, Yoann Vandoorselaere wrote:
>
> Khimenko Victor just explained to you that this patch is broken by design,
> so it isn't acceptable on long term.
>
> If you want to start recoding it the right way, feel free to do it,
> but do not bother people that know it must not be included
> saying that it should be because it work for you.
>
> Understand that even if it improve things for the user it
> may broke things for the kernel.

Do you read what I wrote? I explained my original reason for the post.
I think other people will find value even if they don't only have real time
needs.

I was hoping to encourage people to try it. Because I think if more
people tried it they'd realize the value.

I wish I had the time and competency to recode core portions of the
kernel to fix things like this, but if I did I'd go after the stability issues
first to be frank. However, I don't so it's moot. There is just too much in
the kernel that I don't yet understand. Maybe some day.

I am a heavy user of Linux however, and I hope that the development
community is not disinterested in input from the user community. If that is
not the case, then Linux is dead.

The thing I hate most about Windows is that I can't do anything for an
hour without the machine locking up on me, or at least anything non-trivial. I
also hate the fact that with Windows I'm restricted to one platform, Intel.

I use Linux to get work done, but you know I also still use some antique
Unices here, because:

A) They have support for devices that Linux doesn't support,
most significantly FDDI hardware I have is not supported by Linux.

B) Some antique Unices are more stable on SMP platforms that Linux
is and the applications are such that crashes every 4-5 days are
not acceptable.

If I had an absolute realtime need, as opposed to just liking the fact
that systems aren't hurky-jerky under load with these patches, then I suppose
I'd have to look at a different OS for that.

If you ignore the needs of the end users, pretty soon you don't have a
user community anymore.

But for me alone? Don't fix it, leave it broken if it looks prettier.
But I would suggest that people who haven't tried the patch because they don't
have a need for realtime performance, particularly if they have occasionally
heavily loaded boxes, try it.

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/