Mailinglist Archive: opensuse (1665 mails)

< Previous Next >
Re: [opensuse] why not openSUSE used grub 2?
On Friday, December 09, 2011 09:40 AM Anton Aylward wrote:
Linux Tyro said the following on 12/09/2011 01:17 AM:
On Thu, Dec 8, 2011 at 7:23 PM, Dennis Gallien <dwgallien@xxxxxxxxx> wrote:
Anton, don't leave out reading the bios hardware map and making bios
calls :)

Everyone needs morel learning in Hardware I guess. Even Anton told, he
doesn't know much of the hardware.

There's hardware and there's hardware.
Anton knows an incredible amount about hardware ... just not *THAT* kind
in _sufficient_ detail to write a bootloader. Only enough to write the
error correcting algorithms for the low level driver.

An important part of Wisdom is recognising the limits to your knowledge
and capability. I can't run a four minute mile or bench press four
times my own weight. If you can, great, but I'd advise you to not
speculate on what other's can't do. You are more likely to be talking
about something that they can do but have more important (or
interesting) other things to do.

Pls don't shoot me if my memory is a bit faulty on this, been a long time . . .
IMHO this thread has gone on way too long. I'd like to try to put a nail in
its
coffin, and maybe pass along a little useful info too.

I found the YaST (IIRC Perl) code structure documentation which showed the
logic
flow, sub-routines and their functions. There I found the steps YaST Boot
Loader
was performing in order to determine what installation instructions to pass to
the grub shell, how menu.lst should be built, what boot loaders were already
installed on the machine, how the partition table was configured, what the bios
boot device/sequence was, etc.

Those of us who manage boot loading manually of course know (or should) all
these things *before* we install grub, change menu.lst or device.map, etc. We
also know how to determine what *grub thinks* about what is where. Some of us
may even know the idiosyncracies such as that the grub shell may come up with
different answers when run outside the OS (it makes bios calls) vs when it runs

under the OS (it uses device.map).

But when most users do a fresh install, or go to YaST for help in setting up
the
boot loader, s/he calls upon YaST to know these things. YaST needs to know not
only what we know in our brain when we do it manually, but also needs to know
how grub thinks so that it can accurately predict what grub will do. If you
look at the grub documentation, it tell you straight out that its installation
script and the shell is making its best *guess*. So just knowing how to
install
is complex in its own right, add to that the complexity of the boot loader
itself, and then there is the wild card - the bios, where there is no standard,
made all the more complicated in recent years by manufacturers adding
proprietary code extensions to accomodate their various undocumented recovery
mechanisms, on-the-fly boot device overrides, booting from a device not
anywhere
in the bios hardware map, etc.

The point of this is to reinforce as forcefully as I can with specifics re
grub,
the broader point Anton has articulated well more than once. No one and I do
mean no one, should be touching this stuff unless s/he really knows what s/he
is
doing. And no one - in particular one person (hint hint) - should be going on
and on about something that s/he so obviously knows nothing about. That
message
has IMO been delivered clearly with a lot of diplomacy and patience - more than
once. To that person who is "learning", may I suggest a focus on better
listening skills.
--
To unsubscribe, e-mail: opensuse+unsubscribe@xxxxxxxxxxxx
To contact the owner, e-mail: opensuse+owner@xxxxxxxxxxxx

< Previous Next >
Follow Ups