[cumbria_lug] Observations

[Michael Saunders]

On Mon, 8 Mar 2004, jenruss+jen at mail.plus.net wrote:
>
> It took a while to boot, but was otherwise quite happy. I do see
> your point, though.

>From Russ' followup, that machine is still very well equipped. Boot
time is a problem that needs fixing: on laptops, slow boots are a pain
(Fedora is particularly weak here -- why start X twice?). I've tuned
my laptop's Slackware 9.1 boot to less than 30 seconds, which even
beats the suspend-to-disk (384M to write), but it's not ideal.

> yet there's no way any modern windows would run on a 486

I should make it clear that I'm not some luddite who thinks the latest
desktop OSes should run on a ZX81, but when machines sold three years
ago can't run modern desktop Linux distros at acceptable speeds,
that's a serious problem. Linux shouldn't have the same hardware
upgrade treadmill as Windows, as it'll severely impact its popularity.

> The other thing that really needs to change is Linux's image. Just
> about everyone that's heard of it thinks it's a geek toy, not really
> an option for me.

I think that stigma is dissolving now as Linux gains various massive
rollouts -- if people are aware that Google's 10,000 boxes are running
Linux, and it's being deployed on 500,000+ upwards systems in China,
and IBM are planning to use it extensively on desktops, it'll change.

> There's only one way I can think of to really change that image, and
> that's to come down of the soap box and give people what they want -
> Windows with fewer bugs and problems.

The more Linux tries to be like Windows, the more it will suffer from
the same problems. There are fundamental design flaws in Windows which
should never be copied. Linux should strive to be a great general
purpose OS, not a Windows clone; ReactOS is available in that case.

Besides, if you want a Windows-like Linux then there are plenty of
projects out there. The GNOME people have done a sterling job of
bolting Windows onto Linux: both have huge system resource demands,
both are bloated, both restrict customisation choice, and both utilise
messy, arcane and barely-readable 'registries' for settings. That's
Linux trying to be like Windows, and the result isn't pleasant.

We're dangeously underestimating Microsoft. Right now, their products
are of generally poor quality NOT because of a lack of technical nous,
but because they don't care -- they're making billions, have a
monopoly and are run my the marketing folks. But they can turn on a
sixpence, and act when it matters, like they did with the Internet.
They improved stability considerably with Win2k. They improved boot
time with XP. They CAN do a lot of work if they feel under threat.

In 10 years time, we don't want to be stuck with the slowest-booting,
most complicated and sluggish mainstream OS around. If Microsoft
offers Longhorn+1 for $30 (which they can easily afford to do), and it
has all manner of cleanups and speed boosts, few people will switch.
And as said before, we may not even have the security advantage; it's
fine when you're running the latest goodies, but someone has to
support this code.

Red Hat released RHEL 3 in October with GNOME 2.2. Just over four
months on, GNOME 2.4 is nearing EOL and GNOME 2.6 is round the corner.
And yet RH have to support 2.2 right into 2008 -- how many people will
be looking at the code then? How many people are looking at KDE 1.1
source now? The more bloat and quickly-coded features we add now, the
harder it'll be to support later.

It's long-term thinking -- there should be more discussion on this.

Mike

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