Re: Re[2]: project status

On Sat, Jan 15, 2005 at 06:16:11PM -0500, David Longo wrote:
> Auto hardware integration

The higher the standard deviation from "the mean system" is, the more
difficult it will be to have a fully functional automated hardware
integration. 

At this moment, Linux has a very powerful automated hardware configuration.
A fully configured kernel (such as the one default generated by genkernel)
together with hotplug/coldplug and udev, and some manual scripts that udev
starts for exotic hardware, make it possible to have a very decent automated
hardware detection combo.

The difficulty here is that the developers who work on this have a limited
hardware set at their disposal. Every bugreport that gets fixed is based on
feedback and input from others, which always has the possibility that the
feedback is not accurate enough to be fixed correctly. So the process takes
some time.

The best way is to start with your own desktop and hardware, and investigate
how you can get your beloved operating system to load up all the appropriate
drivers. But don't forget that drivers aren't the only thing of a funtional
desktop - you'll need to install the additional tools as well (thinking
CUPS, SANE, ...).

BTW, on the forums, we've had quite a few people saying two things that can
not be merged together. They wanted genkernel to create a kernel that is
optimized for their system, with no additional drivers attached, but also a
desktop that is able to automount an Iomega Jazz driver. They dislike a
fully expanded kernel but want the features of one.

The kernel people are doing their best to have genkernel function
correctly, but more feedback is certainly appreciated. 

> and config file editing

Apart from my other mail on configuration file editing, I have to say that
most of my systems have little need of additional "configuration file
editing" after the system is operational. Most configuration files are
either automatically merged by dispatch-conf and a heavily modified
CONFIG_PROTECT and CONFIG_PROTECT_MASK setting, or are updated manually in
less than a minute per file.

Mind you that Gentoo provides you with the tools you need to have every
single possibility of Gentoo exploitable. Don't ask us to limit this in
order to improve a certain field. Gentoo is a metadistribution. It is a
distribution from which you can derive "Your Perfect System".

For instance, on the Gentoo Chat forums, we had a request to create a
"Gentoo Light", which was based on Gentoo, but provided a vast set of
applications (less choice) that were tightly integrated, a LiveCD
specifically targetting a certain hardware branch (I think the Mac Mini was
proposed), and an update procedure that improved the configuration handling
of those applications.

That's a fine idea, but it takes a lot of resources. You can use Gentoo for
this (as a base product) and there is nothing that we would like to see more
than distributions that take the benefits of Gentoo and exploit them for
their own purposes.

But when you ask Gentoo itself to put a lot of resources in this, taking
more and more space on mirrors (we already ask 90 GB of space for a full
mirror), etc, you should know that a "NO" is easily returned. 

Not because we think the project is bad, but because we want to focus our
attention on improving aspects that benefit our users *without* limiting
others.

Wkr,
      Sven Vermeulen

-- 
  Documentation & PR project leader

  The Gentoo Project   <<< http://www.gentoo.org >>>

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.2.6 (GNU/Linux)

iD8DBQFB6jxtXfqz7M26L9sRAgrXAJ4xbWcg8qmqsE0OXUuPHIfKBVVajwCeOgeq
3gOeS4NEWAtCIkh/kbm0di8=
=c01O
-----END PGP SIGNATURE-----