Re: FreeBSD-like approach for Debian? [was: Re: Deficiencies in Debian]
hi steve,
thanks for your response. i appreciate the opportunity to discuss this
issue. with you.
>What advantages would those be? I am curious.
i believe the major advantage in making a distinction between a base
operating system and the applications it supports lies mostly in the
modular design of such a system. much has been written about modular
design, but essentially it allows for parts of a system to be easily added,
deleted, or moved with minimum impact or dependencies, allowing
for the greatest flexibility.
> What I am seeing is a clash between the traditional model and the Debian
> model.
yes, in part. but it is not limited to debian. it is true of every linux
distribution
i have worked with. and it is not so much tradition, but experience. when i
used to work at sun we used a subscription based software distribution model.
it was my job to integrate internal and 3rd party software applications for
distribution throughout sun's wide area network (swan). we found that
keeping (as much as possible) the application's resources under one tree
made managing that package much more efficient.
>So it comes down to this, define "base". What is "base" to you? "Base"
>in the traditional model are the bare minimum tools needed for the OS to run
>at a minimal level of functionality....
i can basically agree with your definition. i also liked the definition i read on
jonathan walther's web page: http://mindanaoguide.com/linux-reference.html
as follows:
---------------------------------------------------
What is the Base Reference System?
The Base Reference System is the minimal self hosting GNU/Linux system that
conforms fully to the LSB specifications. It will have an integrated BSD style
source
tree so the entire system can be built and updated from a common repository with
a single command.
------------------------------------------------------
>That is why I ask what advantages you see? I've had this debate with a
>friend of mine off and on for months. He hasn't come up with anything that
>really makes sense other than ideological. Ideological reasons aren't enough,
i enjoy these types of technical/philosophical debates with my friends as well.
i find them interesting and stimulating, usually because they actually force me
to think :) i agree that there are ideological reasons behind my preferences,
but
that is probably true on both sides of the fence, as it where. in my experience
it has just been easier to manage things when they are not strewn about
all over the place. i have often removed software packages, only to find
vestiges still existing in different directories, a library in /usr/lib, or docs
in
/usr/doc, a config file in /etc, and so forth.
i also agree with those people that have voiced concern over package managers
overwriting a local tweak by the sysadmin. i believe this comes down a namespace
management issue that could be handled by something as easy as say /usr/debian
for example.
having said all this, i also realize that making these changes would not be
trivial
work. i am sure you guys are already busy enough, and i do not mean to make you
think i want to add more to your plates. it is just a topic that i find
interesting and
it motivated me to add my .02 cents to the discussion.
thanks,
//daryl
--
Daryl Williams
Network Administrator mailto:daryl@sharewave.com
ShareWave, Inc. Phone: 916-939-9400 x3212
5175 Hillsdale Circle Fax: 916-939-9434
El Dorado Hills, CA. 95762Web: http://www.sharewave.com
Reply to: