HTB.init is a shell script derived from CBQ.init that allows for easy setup of HTB-based traffic control on Linux. HTB (Hierachical Token Bucket) is a new queueing discipline which attempts to address the weaknesses of current CBQ implementation. It is not yet part of the official Linux kernel tree and requires patched version of iproute2 utilities.
Re: Demand #1: Are you being fair?
> In the Windows world, you pay for you OS
> and you demand your hardware vendor to
> provide drives for it. In the Linux
There is a little flaw in what you say. You pay
_Microsoft_ for the OS and you demand drivers
from _vendor_ whom you paid for the hardware.
If you made a mistake and bought hardware from a
vendor who is not going to
support you with your different OS with at least
documentation, it's time to stop
buying hardware from him. It's that easy. If you
are dying for that
hot new piece of wiring, use the OS the vendor
claims to be supporting.
Hmm, and about those proprietary things... We all
know hardware X
does Y. We might not know how it does the feat,
but telling us
"how to tell board X to do Y" probably won't
reveal the secret.
We'd like to know the black-box view on the
hardware. You just pull this
string, and that thing over there does that.
> And now comes the flame war: (some)
> hardware vendors say they will provide
> drivers as soon as Linux has stable
> interfaces. Linus says Linux will not
> have stable interfaces in the short
> term, because they constraint the
> evolution of the kernel.
Hm, yes. Depends on view. I think that we are just
humans, we make wrong decisions and we make buggy
code. By changing the interfaces between the
_major_ versions, linux has a chance to make up
for wrong decisions - iptables from 2.4 are just
one of the examples. Come on, you can't get
everything straight for the first time - things
usually evolve and converge to "low-energy" states
and I'm glad linux (the kernel) has got the balls
to do such thing.
Well, you have stated several interesting demands.
I personally am quite happy with several things as
they are now, eg. I don't want to run linux in
linux just to be able upgrade the kernel. I'm glad
I can upgrade modular things without rebooting -
that's enough for me.
As for the drivers, I don't know who is this
demand addressed to. I believe the only reason why
there are no drivers for bleeding edge computer
appliances is the fact that vendors either don't
provide them or they refuse to release the
documentation. As Alan Cox once mentioned, open
source sofware is always late - as it is with
linux kernel now, first there is hardware and then
might come support. But if there are no docs,
there won't be any support.
And no, let's not talk about binary-only modules.
This is a matter of faith in what is good and what
is bad and I believe that source code drivers are
good. The solution is in vendors to understand
this and provide either the driver or
documentation. If they don't understand this, I'm
not going to buy their hardware. I know they won't
care because they have Windows users to pull money
from. I perfectly understand them as economic
subjects and I often repeat to myself: "This
company is not here to provide me with services,
it's here to make money".
But there is (and will be) place for vendors who
are interested even in the relatively small linux
userbase and make money on it. And I'll happily
stuff my few cents in their pockets as long as
they either provide proper drivers or
documentation or, ultimately, both.