Subject: Bootdisks and Bootable CDs.
wulfworks at nanux.com
Tue Sep 12 04:54:24 PDT 2000
----- Original Message -----
From: Jesse Tie Ten Quee <highos at highos.com>
To: <alfs-discuss at linuxfromscratch.org>
Sent: Monday, September 11, 2000 10:13 PM
Subject: Re: Subject: Re: Bootdisks and Bootable CDs.
I must admit that I was somewhat taken aback by your responce, it seems that
you and I are actually eye-to-eye on most of this? Anyway, allow me to
** Note - possibly nothing more than a long, off topic diatribe **
> And we aren't doing that?
I would have to say that - given the amount of discussion on this topic - this
is not clear? It seems that the path is not well charted.
I understand and appriciate the Open attitude of ALFS development, but do keep
in mind that someone has to be in charge (remeber, a giraffe is a horse
desigend by committe)... Sure Gerard is the ultimate head of the project - it
is, after all, his baby. But each seperate faction of the design and
development also needs a person to head up that area. OK, I do not spend a lot
of time following LFS/ALFS (I have little spare time), but I am not aware of
any documentation that deliniates the "team" structure? I checked the web site
and the sourceforge page and found nothing? Although, I did note that the
sourceforge page has had over 50,000 page views in the last three months! Looks
like other people are looking too?
So, if you are the person responsible for this part of development, just let me
know? It would also be great if you published a project document (on the source
forge site?) that defined the subproject goal, milestones and a list of those
things you want input on -- as well as a list of things that are already
decided and will not change?
> No, all you need is the simple boot disk that is on the CD (usually in
> /boot/boot.img) it just bootups and hands control over to the cd.
I wish were that simple, I know of no magic single image that can be expected
to handle the booting process for the majority (not even thinking ALL) of the
systems that it may encounter? Between old interfaces, SCSI, DMA/66-100 and
whatever else... it requires (in my experience) more that one image to be able
to make the rt/bt disks. Granted... the El Torito specs provide for a 2.88MB
image and that is large enough to include suppport for probably 90% of the
systems?... But when you start talking about rawriting an image to a floppy, I
still think you will need an "image farm"?
I agree that writing a "live CD" is not that great of a feat (I have manged to
make a dozen or so different version), but it is also not really simple... to
build a full Linux OS that runs from CD *reliably* can be a little tricky, but
no big deal really... just look at var type stuff and how it is handled,
check those links!, can user configs be adjusted? What about swap space? Does
the system run only in RAM? How about transfering userspace over to the HDD
> Another reason i want a pure floppy install also (which so far, no one
> has answered my original question)
I thought I made my feelings clear on this? If not, I am sorry... here goes
(again, this is biased to the way that I did my CD): The CD is a *total*
solution... everything you can think of is addressed -- if the user wants to do
an build *entirely* from the floppy drive... then he/she just writes out the
disk sets from the CDROM. If they want to just boot up with network support and
build the system from a network connection... the image is there to do it. FTP,
same thing... etc, etc...
> Besides, it is going to be documented anyways, i like the way LFS is
> done that way... you can learn alot from a simple document.
And where is this documentation? On the CD that can not be mounted yet? I would
hope that the project is arranged so that the user does not *have* to have
another computer in the same room! The worst case should be that the user need
to write out a rt/bt floppy set... in that case, they simply need to find
another system *NIX or Win* and make the disk set! That is what I like about
the "Live CD"... once it boots, the user fires up X and opens a couple of
| Browser - |
| View documents in this window |
| Term - |
| Work on the build in this one! |
> It would be pretty stupid not todo that.
Unless we can find that magic single image that supports all systems, this is
the *only* way?
> If someone didn't "reinvent" the whell more then once then GNU/Linux
> (and almost every distro) would not be where it is right now.
Just a point of procedure here... the engineer in me coming out. BTW, you know
the definition of an engineer?
***An Optimist says the glass is half full.
***A Pessimist says the glass is half empty
***An Engineer simply says the glass is twice as big as it needs to be.
The term "reinvent the wheel" has a well documented and defined meaning in
enigineering circles and simply states this:
If any given number of people from all over the world where tasked with
developing and designing a wheel from specifications of needs only...
Every example of a wheel, even the ones from those who may have never heard of
or seen a wheel previously, would follow the same mechanical principal - In
that, each wheel would have a circular perimiter that acts as the bearing
surface, that circular perimiter would rotate perpendicular and equidistant to
the design attachment point at the axis and there would be direct relation to
the linear displacement of any attached object and the rotary motion of the
wheel... in other words, wheel go 'round!
Granted, no two wheels would be exactly the same... design, material, technique,
apperance and many other things would vary... but all the wheels would still go
Now, at this point, we would be well served to examine each of the submitted
wheel designs and pick from them the best of each? Best material? Weight?
Durability? Then maybe research other factors of the our wheel... is there an
optimum diameter? How about the size of the bearing surface or "patch"? Can we
move the physical attachment away from the design point and affect performance
by changing caster/camber? But... alas... all of this is just "improvement"!
It is important to look at other's work, evaluate the whole and *improve* the
design! Without improvement Linux would not be where it is today. So, you can
spend all the time you wish trying to "reinvent" the wheel - But, in the end,
your wheel will really just go 'round?
> There would be no Linux... Unix is about a choice.
Gnu is Not Unix...
K.L. Davis - LycanthroLabs
High Performance Computers
systemID (mondo - FreeBSD)
More information about the alfs-discuss