> I don't think you're correct. They're required to be the primary support
> point for the OS in the licensing for the OEM version.
>
> They may provide crappy support, but they do have to provide support.
They are the primary support. They walk the user through all the obvious
steps - have you run such-and-such a wizard? And it still doesn't work?
Microsoft might be able to give you a better idea of what's going on..."
Or, when all else fails, all they're obligated to do is make sure it
works. So, "run the restore CD." Never mind what you might have done to
customize your system since you took it out of the box. I deal with this
situation 4-5 times a day. Manufacturers put forth a modicum of effort
towards support; little more.
> If PC makers won't support Linux to at least the "drivers for all the
> hardware are on the website, and any gotchas in the installation are
> documented" level, then hardware support is a crapshoot. Manufacturers
> don't care about whether their peripherals work with Linux, and that won't
> change until there's a mass market for Linux-friendly peripherals.
Right. All I'm asking from Dell is a list of exactly what's in the
machine and what I should tell Linux; that can be a static page on a
website or a document tailored for each build. And, fortunately,
manufacturers _are_ getting serious about Linux support. We sell Goldstar
CD-ROM drives that have "Linux" under the "Compatibility" list. We sell
Hauppauge video capture cards that have a link on their website to the
Linux drivers. Et cetera. Sure, it's not everybody. But it's growing.
>
> And *THAT* won't happen until some of the things we've been talking about in
> these very important threads that a few people are trying to call
> "off-topic".
>
> The installation procedures need bullet-proofing. Hardware support needs to
> be less of a crap-shoot.
Something like the "Add New Hardware" wizard of Win95? I'd go for that.
It doesn't always work, but when it does it's nice. What would be
involved in running through an auto-probe? Using various modprobes and
keeping a log of what's already been tried, sync'ing it to disk after
every step... And if the autoprobe somehow locked up the machine, the log
would reflect this and simply step around that module the next time the
install was run, a la Win95's setup? I see that as being very possible.
Maybe configure a GUI for the installation (nothing fancy; 16-color VGA X
Win, and use a relatively quick library - not Tcl/Tk, which seems to be
sluggish even on my 166MMX with a Matrox). Add to the GUI a front-end to
ISAPNP (how do PCI PNP devices get initialized?) so that things like ISA
netcards could be configured. A message would prompt the user to write
down the settings for the various devices, and then s/he could put in, for
instance, specific IRQ and I/O addresses for the NE module, to support PnP
ISA LAN cards that don't get detected by autoprobing. This making sense
so far?
> Installations need to have fewer of these weird gotchas that require
> specialized knowledge. The average person can't be expected to know what
> the hell gated is, much less why it's frying his PPP connections. All he
> knows is that Windows 95 has a nice pretty GUI interface PPP setup that
> handholds him onto the 'net.
Agreed. Have something similar to the RedHat package selector that runs
at setup time, but maybe make it more like the screen you get with Win95's
"Custom" installation option. Add small descriptions. Don't even install
GATED unless you know the user needs it. Indicate that in the
description.
> And he also knows that if he walks into the store and randomly picks the
> latest-and-greatest piece of hardware, it's probably going to work on his
> system, and if he doesn't he can call a toll-free number and talk to
> somebody who knows where to poke and prod (or at least can read a database
> that says where to poke and prod) to coax it into life.
Regularly updated "Linux works with this" lists, and maybe convince
manufacturers (some already do) to have the "Cool! It works with Linux"
sticker?
> Sure, Windows 95 sucks. But so does Linux. It just sucks in a different
> place.
All a matter of opinion; *I* know how to get around the limitations, so it
doesn't suck for me. It might suck for a newbie, maybe...
> My hypothesis is that if you start from a stable foundation and build toward
> Joe Sixpack, you're going to end up with a better platform than if you start
> with Joe Sixpack and build toward stability.
Agreed. (How important is the desktop? Linux excels as a server; not
sure the masses could be weened from MS-Office.)
> The problem is that Linux isn't getting there fast enough. Actually, that's
> not really the problem, Linux is fine where it is. The real problem is that
> people are trying to CLAIM it's at Joe Sixpack's level, and it's not.
There's definitely a learning curve. I think we could easily step around
it. RedHat has laid a lot of the ground work. But there's more that
could be done to make it a very, very polished operating system. And it
wouldn't take that much work. Mostly just intelligent combining of
existing code and technology.
> But a secondary problem is the hardware support crapshoot, and it'd sure go
> a long way toward solving that if somebody who makes decent mass market
> boxes (Dell, Gateway, HP, IBM, Micron, hell even Quantex) would support
> Linux.
Nod.
> I hate Dell, and I don't do business with the jerks, but it'd do me
> MEASUREABLE good if they came out with a Linux distribution of their own,
^^^^^^^^^
No. Let's stick with just three dozen distributions. RedHat, I think, is
the clear winner, although I haven't used Caldera. Debian's DSELECT needs
serious help, and Slackware's appeal (as I see it) is towards tinkerers,
not the mass market.
> offered it preloaded on their systems, and had *EVEN ONE* tech support guy
> who knew Linux from shinola on an OFFICIAL basis. Even if I never used him,
> and never bought another Dell, and never installed Dell Linux, it'd still
> mean that I could know that I could refer people there. And I'd know that
> any manufacturer of PCI Widget Adapters would have to decide "support Linux,
> or guarantee I never get a Dell contract."
Not necessarily. Dell simply wouldn't put that PCI Widget Adapter into
machines built for Linux.
I think we could easily create a kick-ass Linux distribution based on
RedHat 5.x that would perhaps win over some of the fringe. Our numbers
are strong enough now that we could petition companies like Dell, et
cetera, to - if they won't support Linux outright, then at least offer
machines with no preinstalled OS and that contain components known to work
with Linux. (We could perhaps keep a tediously maintained list of which
hardware we want our boxes to have?) And denote such machines in their
catalogues and advertisements as "Linux Ready." This would require no
real investment of money, or even really time, on their part, and would do
a lot towards increasing our visibility and resources available. Thinks?
Chris. | [EMAIL PROTECTED] | http://ebhon.jnst.uor.edu/Users/crank
Give me life, give me pain, give me myself again
- tori amos
--
PLEASE read the Red Hat FAQ, Tips, Errata and the MAILING LIST ARCHIVES!
http://www.redhat.com/RedHat-FAQ /RedHat-Errata /RedHat-Tips /mailing-lists
To unsubscribe: mail [EMAIL PROTECTED] with
"unsubscribe" as the Subject.
