p.robinson at mmu.ac.uk
Wed Mar 10 03:30:48 PST 2004
Bringing it back to where all this began if I may. This HAS been
discussed before. Lots of times. Lots and lots of times. So, I figured,
it might be worth just re-capping some of the consensus reached on
> I may be part of a startling minority within the *nix/BSD
> community when I believe that they are (even Linux) FAR from
> being ready for home usage by, say, Joe Sixpack, or whatever
> cliche home user image you have in mind. There are a few
> reasons for this that I think we would, the Free Software
> Community as a whole, have a HARD time resolving.
There are no issues resolving this. There is however a split attitude
over how this should be done, or even if it should be done at all.
Some people do not want Unix in the desktop market at all. Some people
don't mind it, but don't want it to change. Others want it work just
like Windows because they want to compete with Microsoft.
The first camp are content with making sure Unix is the best server OS
in the world. The second camp are the ones who bring us where we are
today, and why we have KDE, reasonable USB support, etc. The last camp
are confused, overwhelmed and are kind of hoping the WINE guys hurry up.
> Joe Sixpack becomes fusterated, even angry, when he learns
> that HIS OS is the reason he can't use some program that he
> wants. He doesn't think of it the same way we do though. It's
Generalisations are always stupid.
Did you see what I did there? Funny, no? The point is, there is no ideal
user. There are trends, and those people with money to spend on market
research understand them very well. If you want to help Unix in the
desktop market, produce an unbiased piece of market research.
> realize that it won't run on their computer). Second of all,
> I'm not sure many people who buy these computers fully
> understand what Linux is, and without that understanding they
> are getting themselves in somewhat of a jam.
If you're spending $500 on something, it pays to do some research and
find out what it is you're buying. This is another area where you can do
something directly. Go out and build a website helping complete newbies
considering Unix work out what it does and doesn't do. BUT, you must be
objective. If you don't, they'll see through you and it becomes
> Workstations at corporations are different
As others have pointed out many times before, workstations are not the
same as desktops. This is important to understand when comparing things.
> I've realized this, I don't know how many others have. But
> when a user is told he should switch, say, OSes because his
> OS sucks, he tends to react with anger. You aren't just
"Grown-ups" have understood this for years. It is why FreeBSD Advocacy
is more about communicating what we do rather than slating the other
guys. The Linux/Slashdot crowd refer to themselves as being "better" and
windows being "inferior technology". This just doesn't work in the real
If BMW put out an advert just saying "Mercedes $UX00RRZZ!!", BMW is
likely to win fewer sales compared to the alternative strategy of
pointing out the engineering excellence of BMWs.
> The Desktop version would assume more on install, have a
> graphical installer, and let you choose GUI that you want on
> start (along with xdm/kdm/gdm). Possibly even its own theme,
> which would MAKE a lot more of a difference then you would
> think at first.
[BANGS HEAD ON WALL]
You know, I used to think the installer was the answer to everything. I
then realised it wasn't, and it was much, much, much more complicated
than that. I have over 400 pages of notes stashed away somewhere that
I've been threatening to organise into some website for the last two
years. You may have provoked me into actually doing something about it.
> We wouldn't have to make our own GUI installer from scratch,
> what's wrong with RedHat's anaconda? Modify it, make it use
> pkg_add. Bam, we're in buisness.
Because it does 20% of what an installer is supposed to do. And it's
GPL, not BSD. And the graphical bit is a red herring - that isn't the
good bit about an installer. I really do need to write up these notes at
> A lot of work involved in making two images? Not really, the
> only binaries that I envision being different would be (aside
> from the installer) the kernel. Which REALLY should have
> dummynet and ipfw in by default with 'allow all' by default
> (Only for desktop). Other than that, you wouldn't really be
> adding that much overhead to the whole process.
You see, you've already poisoned it with your own opinions. What if I
don't want dummynet and ipfw in there? What if I want the snoop device
and the ess solo1 sound driver in? What works for you does not work for
> It would be work yes, but I'm willing to BET that it'll
> generate FreeBSD a whole new GROUP of users. There are a lot
> of Linux users who are put off by the FreeBSD installer/etc.
If they're put off by the existing installer, they will get put off by
the fact that we expect them to edit some config files by hand, that we
expect them to read documentation, etc. These are people who do not care
about the OS, as you've pointed out. The learning curve and benefit to
them of changing is too low. Their machine came with Windows on it, it
works, what is the benefit of switching?
> By the time the get good enough to handle that sort of thing,
> they are usually already settled in and they have spent so
> long working on a solution that 'just works' for them that
> they aren't interested in switching or trying BSD out.
In which case, they should be allowed to get on with it in piece. Like I
say, just going around talking about how everybody should switch is
likely to make them less inclined to switch.
I see where you're going, but really, this bikeshed has changed colour
at least 20 times in the last year. If you're interested in innovative
installer design, take a look at DragonFly's approach due in the next
year or two.
More information about the freebsd-advocacy