[cumbria_lug] New distro advice

[Michael Saunders]

On Thu, 19 Feb 2004, Schwuk wrote:
>
> You said "But that's my point. You can't put a Linux distro on that
> machine that rivals Longhorn in featureset and capability!" - my
> point was that you couldn't put Longhorn (or any Win32 OS execpt 95)
> on that machine anyway, so it was a redundant point to make

My point was that if Microsoft aren't going to make an OS for these
older machines, then we have a huge market for Linux! But instead,
we're making desktops and distros almost as bloated, so we're no
better on that front. We should be providing the upgrade path...

> Yes, but realistically how many people are going to be using the old
> code? There's a reason products are EOL'd - that's one of the
> downsides of OSS: you can't kill something once it's been released.

Hundreds of thousands? Possibly millions? Even tens of millions? RHEL
will be supported until 2008, and as we know RH backports bug and
security fixes to the original code. In 2007, we'll be at GNOME 4 or
5, and nobody but distro staff will be looking at the ancient code.
And as it gets bigger, more complicated and bloated, piled up with
features, it's going to get harder to maintain.

Open source isn't some magic thing that fixes all bugs and cleans up
all security vulnerabilities. Developers only work on what interests
and pleases them, and we don't want a huge, crufty and messy codebase
in thousands of companies throwing up bugs and security holes. But if
we go on like this, we won't have any choice...

> Only so far as it suits them - Microsoft (and other vendors) are
> notorious for requiring bigger hardware to power new software, and I
> have no doubt Longhorn will be the same.

Yes, I think it's very likely that Longhorn will be an enormous beast,
but as I said Microsoft can spin on a sixpence. Up until a few years
ago, they've had no serious competition (and subsequently pressure) to
make major improvements. But as they see chunks of their market being
eroded to Linux, they'll try to beat it in every area. And it won't be
hard to beat Linux desktops on perforance unless things are tidied up.

> > Do you believe that if a modern desktop Linux ran at the speed of,
> > say, BeOS, it wouldn't accelerate takeup enormously?
>
> No. One word - compatibility. Why do you think BeOS failed?

Uhm, I don't know if you're referring to hardware or software compat,
but I still don't see your point. Neither of those have a significant
effect on performance. BeOS failed for a number of reasons; that's
totally irrelevant to this discussion. I'm saying that BeOS attracted
plenty of attention thanks to its speed (both booting and running),
and if Linux performed similarly it'd fare SO much better.

> No, but integration does, and Windows (currently) has the edge on
> integration, and that is the direction Gnome and KDE are heading in.

Yeah, Windows is integrated to such an extent that security holes are
a nightmare to fix, a single bug can have a catastrophic effect on
many aspects of the OS, and it's a disaster zone. Yes, more desktop
integration would be a bonus, but let's keep things neatly layered.

> But a 'desktop' linux doesn't have to target "eighty-five trazillion
> platforms and devices" - just one: the PC.

There is no 'the PC' though. Look on the linux-kernel mailing list,
and you can see hackers battling through all manner of idiosyncrasies
for different motherboards, BIOSes, IDE controllers, SCSI controllers,
ACPI implementations, etc. PCs vary so wildly, it's totally different.

> Who said anything about imitation - I meant do what Apple did: take
> a stripped down UNIX core and build on it.

Well, NeXT, QNX and co. were doing that long before Apple. The single
GUI on a UNIX base is not at all new. Anyway, perhaps we're getting
closer in agreement here: simplification. A standard GUI would
stagnate competition -- something more unified and sane would be good.

> How is it a "different market and userbase" anyway? Do people who
> use Macs use their computers in completely different ways to PC users?

Yeah, they spend all their time drooling over candy layouts and fixing
logic board failures :-) Seriously though, Apple have a few very
specific target markets: SOHO style icons, graphics work and computing
novices. The last has been a problem as Apple kit remains relatively
expensive, but it's entirely different to the corporate desktop Linux
that companies such as Red Hat, IBM Novell are heading for.

> Very true, but don't hold your breath. A company or project can
> change direction - a community the size of Linux can't.

You're right. That's why I'm concerned about this :-)

Mike

-- 
Michael Saunders
www.aster.fsnet.co.uk