2005/11/27, Arjan van de Ven <arjan(a)fenrus.demon.nl>:
On Sun, 2005-11-27 at 10:10 -0500, Matthew Miller wrote:
> On Sun, Nov 27, 2005 at 07:37:08AM -0700, Tom Tromey wrote:
> > I don't doubt this. However implicit in this is some idea of what the
> > target Fedora Core user looks like -- some kind of user profile. I
> > think the goal of this thread is to make this idea explicit.
> > Are we targeting developers? Corporate desktop users? People who
> > travel a lot? People setting up LAMP servers? Some combination?
>
> I'm for going all the way -- it should contain no end-user applications,
> just the framework for making those applications run. C'mon, let's make
Core
> be just that.
actually I tend to disagree, not sure how much say I should have in this
though.
I think of it as an union model, where each layer is both self-contained
in terms of package dependencies but also in the requirements it
fulfills in terms of user/market.
interesting thought.
eg
layer 0 of the union
"base" - minimum set to get the machine booting and operating
thats what we call "minimum install" right? ;)
layer 1 of the union
"core" - set of functionality people expect from a consumer oriented
linux distro
i agree here aswell. this should include everything that is needed for
a regular end user to do the daily trivial tasks. from an instant
messaging/irc client to a web browser for adding bugzilla entrys and
find help/info on the net.
layer 2 of the union
"extras" - more "obscure" functionality for example because it's
a
relatively small userbase but also because it can be new and emerging
things. In addition this can be alternative implementations to core
functionality. An example of this could be wu-ftpd or xfce or ..
in my experience before stuff comes there usually its been in a higher
layer before and got some love there already.
layer 3 of the union
"dedicated repos" - very specialist repositories that each target what
is pretty much a niche market and who's requirements are very different
or unique but isolated. Examples could be a beowulf repo, or a "video
montage" repo.
i see that new stuff can be plain pushed faster with "specialist
repos" also as the name you gave it already implies there are people
dedicated to the certain task working on it, which also implies that
those people really "use" the tools they package. thats not
necasserily true with the lower layers. Its a good way to get stuff
ready for a lower layer if specialists of a certain task help with
getting the upstream work going.
again, for me it's essential that each layer of the union is
self-contained in terms of packages (eg no dependencies to higher
layers, lower layers is of course ok) and of functionality (eg if a
layer contains a certain functionality, it should contain in a general
useful matter. this doesn't mean that all optional things should be
there, but enough functionality that most people expect should be
there).
For me the difference between layer 2 and 3 is also a "market segment"
one. Some things will be so specialist that it's better to have a few
experts have their own repo that they maintain as a coherent add-on
function than lumping it all in one big repo.
Well theres a huge difference in my eyes between plain rolling a
current state to rpm and pushing necassery changes and improvements
upstream while beeing in the process of creating a real good solution
for a task. so yes... "task forces" make definitely sense.
regards,
Rudolf Kastl
--
fedora-devel-list mailing list
fedora-devel-list(a)redhat.com
https://www.redhat.com/mailman/listinfo/fedora-devel-list