Goal: Increased Modularity?

[Richi Plana]

Hi,

I've been reading past posts and am starting to understand the goals of
the Fedora project much better. If I'm not mistaken, the primary goal
isn't to target a specific set of users (ie. desktop vs. server) or
platform (x86_64 vs. ppc, notebook vs. desktop), but to come up with a
distro that's easy to re-spin for people who DO want to customize Fedora
for a particular purpose.

If that's the case, would I be correct in assuming that Fedora would
prioritize increasing the modularity of its packages? That Fedora is
making it easier to add or remove while minimizing the affect it would
have on other packages?

For instance, if I desired to come up with a spin that doesn't have
Sendmail, why must I give up fetchmail, mutt or tor? Why is it that so
many packages can't stand alone without libvorbis? I know that some of
the packages NEED libvorbis, but for many, shouldn't it be optional and
something that isn't required to be compiled against (think dlopen(3)
instead of ld(1)) like gstreamer-plugins-* (which all seem to require
libvorbis)?

I guess I'm just stating a principle which should be a guiding one if
Fedora wants to come up with a system that's really flexible for people
who want to come up with their own spins. So currently, is it (a guiding
principle to make packages modular and depend less on others)?

Note that I'm not asking if this is to be a target, just if it's written
somewhere as something to keep in mind when packaging things. I
understand that Fedora is dealing with a huge system that's been around
for ages. Some people's complaints about Fedora being too bulky isn't
even Fedora's fault. They've just been dealing with a relic that it has
to modernize. Towards that goal, Fedora has done a lot and continues to
improve on it (next-gen init, boot-up process, etc.).
--

Richi Plana