[gentoo-user] Re: Is GWN dead?

[James]

Michael Schmarck <michael.schmarck <at> habmalnefrage.de> writes:


> Well, actually, I never used a Gentoo install CD to install Gentoo. I
> also don't quite understand, why anyone would need such a beast.

Folks new to gentoo, would find it suspicious, for a distro not to have
it's own install..... Other forks of Gentoo have their install methods.

For those new to gentoo, it's like getting married without a honeymoon....
IMHO.


> To install Gentoo, I'd boot my favorite "rescue system" (GRMl nowadays,
> Knoppix back then, but IMO Knoppix is too "fat" for *this* *task*)
> and install from there. No need for an install CD.

OK, fine, then why doesn't of the persons that says it so easy, just take
a GRMl (or whatever)  cd and add the minimal (non gui) stuff to the same 
cd and make a simple to use 'install cd' for gentoo that is unofficial?
Wouldn't it be easy for all of those whose answer this installation 
question over and over and over, to make a basic install cd on top of GRMl
once and be done with it? After all very little would change, except when
the GRMl cd changes..... Time the updates with changes int he 
GRMl cd....


> As far as I'm concerned, the Gentoo install CD could easily be dropped
> without a loss.

Well, I differ with this statement 100%. What, IMHO, needs to happened is the
whole install process be changed to a minimal working kernel and basic tools.
Then you fork the install in the direction as to what the system is to be 
used for: embedded-gentoo, firewall, bridge, managed switch, server 
(mail, web, dns, terminal etc etc) and last the complicated nightmare of 
a workstation  (kde vs gnome vs etc etc).

Of of the best features of Gentoo, is how easy maintaining and managing a 
server is. 99.999% of the issues with updates to gentoo, are related to
the wide variety of packages available for workstations......?
This approach could be used to build a basic installation with support for a
wide variety of hardware, within a particular architecture. Then
as the amount of installation packages are increase, logically break the
installation across multiple (media) CDs. For example something like this

Basic system             complete packaging            workstation
kernel, baselayout...    <needs to be discussed>       X, kde, gnome, 


(very crude idea that needs to be refined.)

It's the graphics and installation of thousands of various gui-packages 
(and using a gui that installs on any machine) that is the nightmare, IMHO.

So yes, drop the (graphical) liveCD 0ption and create  installation
method(s) that begin simple (therefore easy to maintain and update)
and becomes more involved depending on what you are trying to build from
gentoo...... Some of the x86 embedded devices, such a GNAP, use
older versions of compilers and sources. Their install could fork
much earlier, depending on the current state of the architecture.
Or maybe the necessary cross-compile environment would be set up,
along a particular fork.


A robust, well defined installation semantic, is fundamental to
any successful distro, IMHO. Exactly what that semantic entails
should be widely discussed, refined for ease of maintenance and something
that uniquely leverages Gentoo's strengths. 

As processors continue to  shrink and have a lower power consumption, 
the natural migration to mobile (embedded systems) is the future, 
methinks. Gentoo's strength in the embedded space combined with 
being a source code flexible system puts gentoo in the forefront of 
this revolution. However, if installing gentoo, when asked, gives
dozens of different answers, depending on a variety of asymmetrical, 
emotionally charged opinions, then the distro will  continue to 
languish, and be a reclusive club for experts, or those
with very think skin (to which I belong <you pick>)....

For example, when show skiing recently, I met a kid that had a camera
mounted on top of his helmet connecting a coax cable (and power) to
a very small (temperature rated) embedded system. He just replaced
the SD media when he runs out of disk space. That's the kind of project
where embedded gentoo and gentoo workstations need  "seemless"
integration.

The greater Gentoo community should decide what is best for gentoo and 
the installation semantic is the most important piece of 
advertisment/marketing that the  Gentoo organization will ever 
devise, IMHO.


James




-- 
gentoo-user@l.g.o mailing list