Re: bad experience with Suse 9.1 on Inspiron 8200
From: Kevin Nathan (knathan_at_project54.com)
Date: Sun, 11 Jul 2004 10:49:04 -0600
On 11 Jul 2004 04:46:47 -0700
email@example.com (perfb) wrote:
> however true the case is from the developer's point of view that
> it is an M$ bug, such technicalities are irrelevant to the end user.
> M$ has many such workarounds of its own to correct for 3rd party
> driver bugs as well, and if one of those workarounds failed
> catastrophically, it would be an M$ bug, not a 3rd party bug.
I still don't see it that way. If the driver has a problem (or bug)
that M$ has to 'workaround', then it's a driver bug and *not* an M$
bug _even if_ M$ workaround failed. But, as you said above, this is
irrelevant to the end user, so maybe the point is moot. I, however,
will still put the blame where the blame belongs; just one of my many
character flaws! :-)
BTW, this is similar to the LG CD-ROM issue and Mandrake 9.something
where Mandrake would physically destroy the CD drive. It was not
Mandrake's fault (and I'm no big fan of Mandrake!) but LG's, yet many
people tried to blame Mandrake . . .
> granted, the Linux dual-boot option is a nice feature that Windows
> does not have. But Windows doesn't need it, whereas Linux *requires*
> it, practically speaking.
Sorry, gotta disagree a little with you here, too! :-) Windows has
'needed' a dual boot option almost from the beginning but, since they
were trying to force people to use them and ignore OS/2 and various
DOS clones, they have never *wanted* to add dual boot capability. One
of the (many) reasons I abandoned them four years ago.
> And if it doesnt work in some cases, whose problem
> is it?
This is where I get a little more draconian than many people on this ng
like, but if you have no interest in *learning* an OS and just want to
use it, either stay with what works or have someone help you or do it
for you. The other option, if you're 'just curious', is to set up a
separate computer to test run it. I have *always* strongly discouraged
dual booting for the average user -- there are issues involved that go
far lower than just the OS. One simple example is calling tech support
for some program (or OS) and the techs having you change a BIOS setting
which causes one (or more) of the multi-boot systems to stop working. An
average user would be lost then and would blame the non-working OS,
possibly incurring a hefty service charge to get everything working,
This would be similar to me, who has very little interest in how cars
work as long as they work (average user), trying to install a nitrous
system on my car. If it all goes wrong and my car won't start, it's
*not* the fault of the nitrous system. I would be better off having
someone else do it. And, in either case, it would be sheer stupidity
on my part to do this to my main (or only) vehicle if I need it for
other important reasons.
> And is not just an incidental nicety for Linux, it is a
> practical *necessity* to have such a feature for Linux, for 90% or
> so of the present Linux users.
Ok, I'll tend to agree with you here, but I'd probably put the percent
closer to 50% or less. I still believe that *most* people that try out
Linux are at least a *little* bit curious about how their computer works
and how to tweak it. Most of the 'average users' I know are terrified of
even upgrading Windows to the next Windows version, much less a
completely different OS!
> Maybe, one hopes, in a few more years,
> this will not be so much the case.
I would like to see M$ acknowledge that people use more than one OS and
start *supporting* multiple OSes alongside Windows. Until they do that,
I will continue to blame *them* for problems incurred when installing
another OS (whatever it is).
Using this kernel 'bug' as an example, it really is not unreasonable for
the kernel to report LBA values for the partition table. It's been a
long-standing option in the BIOS and Windows should have supported it
fully long ago. Linux has no problem reading and interpreting the disk
geometry *regardless* of the BIOS setting, and I suspect other OSes do
this similarly. To illustrate, I had a hard drive (my 11GB which is now
my /home partition) which was my only drive at the time and I had to
fiddle with CHS values to get Windows to see the (almost) full capacity
of it. I then installed Linux after having Windows running (and a hidden
MS-DOS boot partition, as well) and it saw the entire drive capacity
*even though CHS values didn't calculate to that size*. I decided to
play with it a little and set BIOS to AUTO. It calculated different CHS
values, slightly lower than I had got working, but Windows refused to
boot, while Linux went merrily on it's way. Then I changed it to LBA and
the BIOS listed the entire capacity. Windows refused to boot and Linux
still booted fine. (BTW, this was about six years ago.)
So, I decided to switch to LBA. I wiped the drive completely and
installed DOS in a 2GB partition (no problem) and then tried Windows
install. It refused to install claiming a problem with the disk (I
no longer remember the exact msg). Installing Linux (this was Red Hat
5.x) had no such problem. So I went back to my tweaked CHS values and
re-did all the installs. I had an 8GB disk in Windows and an 11GB disk
in Linux. Sounds like a Windows problem to me. :-)
> Therefore, imho, the dual-boot is a *critical* feature,
> which *must* work as robustly as possible.
Yes, I will grudgingly agree. The more average users that try Linux,
the better Linux will have to be in 'fixing' other OSes problems. Even
now, Linux has to be a hundred times better than Windows for many users
to say it's even half as good. :-)
> Anyway, I am pretty impressed with how well it does in fact work.
In this, you and I are in total agreement! The work that has gone into
Linux is nothing short of phenomenal and I applaud all those who strive
to make it so, and to make it even better!
> Because Linux has hosed the partition table. And not all BIOS'es
> allow setting the LBA option, e.g. Dell Inspiron.
Thanks for the info -- I'll never buy a Dell then! Of course, this was
never a problem since I do not buy brand-name computers; they are too
much like M$ -- they cripple the BIOS and try to force you into doing
things *their* way. I've had no end of problems with brand-name boxes
while having very few problems with the generic boxes, especially when
I build them. I want *total* control over my computer and will settle
for nothing less . . .
> then all data access could be irrecoverably
> destroyed for both Linux and M$. Ergo *potential* catastrophic data
> loss is a consequence of this bug.
You're right, I didn't take that into account.
> I still dont know what the real cause was
> other than the Suse 9.1 install locking up with the hd on
> continuously, right after downloading all the patches in the normal
> install process. Then once I did the hard reset, it never booted past
> GRUB, and I had to do a FIXMBR in XP recovery. Which, according to
> Suse, might have trashed my hd *if* it was in fact the parted issue.
After a few catastrophes in Windows when running an update right after
installation, (I don't remember them offering it during install on
Win98) I learned my lesson and got everything working first before ever
doing an online update. I also do minimal installs in Linux and only
start adding programs after the hardware is up and running.
I would venture to guess the reboot during install was a big problem and
you should have started the install over at that point. That was even my
experience on Windows if it died during an install: reformat drive and
start over. :-)
> Now M$ is that much closer to being a vestigial wart on my PC!
Shouldn't be long, now -- I've been Windows-free at home for four
years, and couldn't be happier! :-)
-- Kevin Nathan (Montana, USA) Open standards. Open source. Open minds. The command line is the front line. Linux 2.4.20-4GB-athlon 9:30am up 1 day 16:01, 10 users, load average: 0.02, 0.03, 0.01