Why does Guix have so much in-tree functionality?

[Kai Jellinghaus <kaij@j10-labs.com>]

[-- Attachment #1: Type: text/plain, Size: 1381 bytes --]

Hello all, I've been fiddling with Guix for about two days, and have
managed to make some first modifications.
I'm wondering why Guix has so much built-in functionality? I don't really
see the benefit of having this large number of packages in-tree. Similarly,
machine configs, bootloaders, build systems, and some image configurations
are all kept in-tree, part of every guix build.
I'm not clear on what the reasoning is? I'd highly appreciate some help
here.
I think being able to hack guix is very fun and once I get the hang of it a
bit better still, I can even see it being very useful for all sorts of
things. But it requires modifying, building and ultimately shipping this
large, complicated repository.
Of course channels exist, but they are additive only, while a large number
of packages already exist in-tree. And either the build system of the
package is supported, or it's a huge pain.

For bootloaders the image configuration I'm totally lost.

I've tried my hand at reducing guix to only the absolutely necessary parts,
but it's very hard with packages like `base` and `admin`, and of course I
don't have the knowledge (yet) to know what exactly is critical to
functionality and what is just "nice to have".

Anyways, I'd really appreciate it if someone that has more experience and
might even know the history of how these things came to be would help me
out! :)

[-- Attachment #2: Type: text/html, Size: 1584 bytes --]