[Wylug-help] Networking Linux PCs

John Hodrien johnh at comp.leeds.ac.uk
Mon, 2 Dec 2002 11:04:53 +0000 (GMT)


On Sun, 1 Dec 2002, Frank Shute wrote:

> On Sun, Dec 01, 2002 at 05:56:35PM +0000, John Hodrien wrote:

> > But you don't have to do that much to a slackware system.  I didn't know
> > slackware had a reputation for not working out of the box.
>
> I haven't used slackware but I've heard it's not the easiest system to
> maintain. It might work out of the box or it might be semi-broken like
> RedHat distros usually are or completely broken as one was on one
> occasion.

So you don't really know about Slackware then?  Which RedHat are you accusing
of being completely broken, and on which occasion?  Vague slander is really
quite pointless.

> > > > > The RedHat installer is more complicated than the FreeBSD installer
> > > > > & because it's got whizzy graphics and the like, it doesn't really
> > > > > make it any easier to use - newbie or otherwise.
> > > >
> > > > I disagree.  Newbie perceived easiness rates a mouse click as easier than
> > > > right-right-return.
> > >
> > > You've got it right in that yes it's the `perceived' easiness of a
> > > click through menu rather than using cursor keys - I personally don't
> > > think it makes a damned bit of difference.
> >
> > But perceived easiness is really important, since at the end of the
> > install, the newbie thinks that it has been easier.  Anything that makes
> > it feel like it hasn't been an ordeal is a good thing.
>
> It doesn't matter what it `feels' like IMO, does it get the job done
> any better/easier is the question that needs to be asked. My
> experience? It doesn't.

Ah but you're talking about yourself again, when you originally said 'newbie
or otherwise'.  You've not said anything compelling that makes me thing that
the RedHat installer is more complicated than the FreeBSD one.  So what if it
has pretty graphics to make newbies feel more comfortable.  It's not like it
needs them (and can be run without still AFAIK).

> > > > What faults are you pointing at with their kernel?
> > >
> > > It's fairly far from what you'd pick up from kernel.org, so it's
> > > really only been tested by RH beta testers.
> >
> > Only tested by the beta testers + all the RedHat users.
>
> RH beta testers + RH users < linux users

But normal kernel users < linux users.  I don't get your point.

> > > Well my complaints about KDE & Gnome as they stand are too long to
> > > mention here ;) .... but again they've been buggered around with by RH
> > > for the sake of `usability', well to my mind software that's been
> > > buggered around with tends to end up less usable as it's inevitably
> > > got a smaller userbase and has ended up having bugs introduced into
> > > it.
> >
> > So nothing in particular then?
>
> You seem to have problems with the concept that a software release
> that is more widely used and tested is going to be more stable than a
> software release that's not.

But you're implying that RedHat are this evil closed group.  They're merely a
large contributor (who also chooses to fork things slightly to their needs).

You seem to have a problem with RedHat.  What bugs have RedHat introduced into
KDE and Gnome?  What bugs have been removed by RedHat from KDE and Gnome?

> I want release quality software to be release quality, if I wanted to
> bug test then I'd use beta software. RH software is released full of
> bugs in my experience - I used RH for 4 yrs and it never got any
> better.

YMMV I suppose is all I can say to that.

> > > You've got a beta system with hacked kernel and rpms. Upgrading kernel
> > > or rpms is prone to problems in my experience compared to using
> > > cvsup/ports/portupgrade. You can't get away for very long with not
> > > updating kernel/rpms from a functionality and/or security viewpoint.
> >
> > But are you arguing a benefit of ports or simply arguing that BSD is
> > more stable because of the development process?
>
> Both to some extent. RH is largely developed `in-house', BSD is not -
> anybody can contribute. BSD is hence more widely tested resulting in a
> system that's renowned for its stability. That stability is also based
> on it being easy to maintain.

Rubbish.  RedHat is largely developed by all the people who develop the
packages that comprise it.  If it was developed in house it would be nowhere
near where it is now.

> > I still don't see what your complaint with RPM is.  You've not
> > convincing explained why it's so hard with RedHat to stay up to
> > date.
>
> The problem with RPMs is that there are a limited number at RH.com
> that are known to work with some particular distro. One to run qmail
> instead of stinky old sendmail? Go grovelling around the net for a rpm
> that may or may not play nicely with your current distro. Or build
> from source.

Ah, you mean like source RPMS?  Nope, you've just plunged yourself into the
dark again.  If someone makes a source RPM that doesn't work, then it's the
same as if they'd made a port that didn't work, no?

> If you go the rpm route eventually you'll land in RPM hell - you'll
> come across an RPM built against a newer library so you install the
> newer libraries but `oh, dear I've just gone & broken a load of other
> rpms'. So update those rpms and then find they've got dependencies on
> newer/older stuff and other stuff starts breaking when you update
> those....

But you don't *have* to upgrade libraries, you can surely install both?

> But with FreeBSD there are 7000+ ports, the vast majority of these are
> actively maintained and they are known to play nicely with your
> current release and the port maintainers have provided patches and
> Makefiles so you just have to do `make install' and you can keep your
> ports tree current via cvsup and upgrade them with portupgrade. But
> how do I do that with some rpm or tarball that I've just grabbed off
> the 'net from somewhere? You can't.

Agreed.  But if you take something from outside of the ports then you're in
the same state.  You can't just argue scale as being why RPM sucks.  If RPMS
covered all 7000+ ports, what would your argument against it be?  It's not
that hard.

> > > It's easy for me to keep my FreeBSD boxes up to date with the latest
> > > kernel and userland - not that I do, I just fix vulnerabilities - I
> > > just have to run a cron job in the early hours if I want to.
> >
> > As it is with RedHat.
>
> I think not. No cvsup or buildworld equivalent and portupgrade makes
> rpm look lame - which it is of course. Then of course there's those
> pesky rpms/tarballs that you've just grabbed off the 'net.

Ahhh but you're not arguing against RPM now.  You're arguing about one of the
command line tools you've used.  apt-get for RPM, urpmi, up2date?

> There is no proper system in place to keep your system current. cvsup,
> buildworld and ports allow me to keep my system synchronised with how
> RELEASE currently stands ATM.

up2date.  apt-get RPM, urpmi (Mandrake).

> Build a new kernel, base and userland automatically applying patches
> along the way with RH & rpm unattended? I don't think so. Jump
> from RH6.2 to RH8.0 without a major headache? No.

Fair enough.

> See above. As it stands, XP is easier to maintain than RH but that
> doesn't seem to alarm RedHat users....which itself is a cause for
> alarm.

Hahaha.  Sorry, I thought you were a troll before, but now I know you are.

> RedHat's business is currently based on a flawed OS that needs to be
> fixed pretty rapidly if they are to remain in business. Increasingly
> whizzy installation graphics is fiddling whilst Rome burns.

I see no evidence of FreeBSD taking the world by storm.

jh

--
"People demand freedom of speech as a compensation for the freedom of thought
 which they seldom use."
                                                     -- Kierkegaard