Re: kernel-{image,headers} package bloat
On Tue, Apr 24, 2001 at 04:41:36PM +0800, zhaoway wrote:
> > Because it teaches them to perform a task they should have to
> > perform, at the right time. At some point, you *have* to stop tying
> > your children's shoelaces, and teach them to do it for themselves.
>
> You obviously didn't read my previous post in this thread.
>
> Your point is ridiculous. You think linux kernel compiling is
> something as fundmental as tying shoelaces. rotfl. sorry.
Yes, but Linux itself is far more complex than life.
Tying shoelaces forms a fundamental part of taking care of yourself.
Compiling a kernel, is a fundamental part of Linux.
In prep, you teach your kids how to take care of themselves. You teach them
to tie their shoelaces.
As they get older, they're able to handle more complex things, like Linux.
So, you teach them the fundamentals of Linux. Including kernel compliation.
(and make-kpkg makes this obscenely simple).
> > > I can see your suggestion help mirrors (The trade-off is
> > > questionable.) But I can't see why it could help users.
> >
> > Questionable?!?
>
> Anything isn't questionable?!? ;)
The thing is, by including all these headers, you're helping break stuff for
everyone. But you're also helping a small percentage of people who don't
want to learn how to compile their own kernels. The scales are tipping very
heavily, and I don't think I need to point out which direction it goes in.
> > Have you seen ftp.au.d.o and mirror.aarnet.edu.au lately?
> > (Admittedly, aarnet sucks and is broken half the time, but has the
> > exact same breakage as ftp.au.d.o, which is normally fine). The
> > mirrors which have trouble keeping up, and have out-of-sync
> > packages, don't need *any* more load, least of all load like this.
>
> Okay I have to agree I'm not very aware of the sufferings of our
> mirrors. I left this for those who know better to decide.
ftp.au.d.o is currently missing several packages, as is
mirror.aarnet.edu.au. Breaks a dist-upgrade (out-of-sync Packages, and
actual packages).
> > > I certainly don't think that make users (even make the task
> > > easier) to compile a kernel can do help to most of the users (That
> > > is *not* a task they should be bothered at all.) And I certainly
> > > disagree that they who don't want compile a kernel don't belong to
> > > our user base.
> >
> > There's a *reason* RedHat exists.
>
> But that's not the reason that Debian can't do better.
You have to look at where Debian is aimed, really.
You may disagree, but I think it's rather obvious that RedHat is aimed at
newbies, Debian at power users. If newbies want Debian, hand them the
Progeny ISO. I started on RedHat, and spent a while there, and I'm thankful
for that. While you have things like /etc/rc.d/init.d vs. /etc/init.d (which
are easily remembered anyway, and FHS will fix this), RedHat gently eased me
into Linux-y stuff. Including kernel compilation, and, later, hacking. As
well as compiling my own stuff.
> > BTW, we're already asking them to understand the installer and
> > dselect (and I have had to help some of my friends, all of whom who
> > were very good at Linux, through this), so why not the kernel? If
> > the figures Craig's throwing around of 110meg per kernel version are
> > true, then that's absolutely unacceptable. Think of: out-of-sync
> > mirrors (they don't need more load, or bloat to fill up the
> > bandwidth and hard drives), and countries that pay by the meg for
> > bandwidth.
>
> If we have no better way to go, we educate users that computer is
> dumb/stupid/dangerous/broken, take care of yourself. But this is *NOT*
> something to be proud of. please.
>
> But yes if the mirrors do suffer from this ~100M kernel packages, we
> may actually have to tell our users the sad message maybe.
No, we say to users, "This is our kernel. You want variations, you make your
own. Deal." Like I said, I'd rather see teaching newbies the (now rather
simple) step of compiling a kernel, than breaking mirrors for all.
> > Sure, I'll reconsider this argument when the new debian-installer is
> > ready (which sounds great already), but right now, no way.
>
> I guess we're not talking about installers.
No, we're not, look what I've said above. "BTW, we're already asking them to
understand the installer and dselect". This is a non-trivial step. If Debian
clearly *was* about newbies, like RH, then the installer would be a breeze,
no?
> The above said. I may actually agree with you, if
>
> o the preformance difference of optimized kernel and general i386
> kernel is so trivial that users can't even notice in their day to day
> work.
It *is* absolutely *minute*.
> and/or
>
> o the hits on the mirror is untolarable technically/financially.
dist-upgrades broken for about a week?
> I won't agree the following though
>
> o users need to be educated/trained to compile the kernel for
> themselves.
>
> the reason i've stated in my previsous posts to this thread.
I don't mind seeing these packages in Progeny, or whatever the next
Debian-with-a-cool-installer-and-KDE-by-default distro is, but the fact is,
clear and simple, that Debian is aimed at power users. If you can't hack
that, grab Progeny or RedHat. Simple as that.
--
Daniel Stone
Linux Kernel Developer
daniel@kabuki.openfridge.net
-----BEGIN GEEK CODE BLOCK-----
Version: 3.1
G!>CS d s++:- a---- C++ ULS++++$>B P---- L+++>++++ E+(joe)>+++ W++ N->++ !o
K? w++(--) O---- M- V-- PS+++ PE- Y PGP>++ t--- 5-- X- R- tv-(!) b+++ DI+++
D+ G e->++ h!(+) r+(%) y? UF++
------END GEEK CODE BLOCK------
Reply to: