write intro texts for the various projects
Some checks failed
ci/woodpecker/push/woodpecker Pipeline failed

This commit is contained in:
Jef Roosens 2024-06-06 17:07:01 +02:00
parent 4374c7f280
commit ea1970caa7
Signed by: Jef Roosens
GPG key ID: B75D4F293C7052DB
13 changed files with 90 additions and 72 deletions

View file

@ -8,24 +8,24 @@ params:
url: 'https://git.rustybever.be/Chewing_Bever/rieter'
---
My first goal for this project is to create a convenient all-round repository
server implementation that could be used for everything from self-hosting a
local repository to managing an entire distribution's package repository. It
should be easy to deploy, lightweight, and work with any distribution. It
should support any number of repositories and packages, and work with any
package architecture.
This project is a reimaginging of my Vieter project. While its goal is to
eventually fully replace Vieter, I'm following a different mindset on how to
get there.
The repositories can be populated by manually uploading packages to the server
(e.g. from a CI build), or by mirroring already existing repositories. The
mirroring feature in particular makes it trivial to set up a new mirror for a
distribution, as the server would take care of keeping the mirror up-to-date.
Another usecase for this would be creating a local mirror of your
distribution's repositories, which can greatly reduce your update times
depending on your internet connection.
First and foremost, I want to make a well-designed Pacman repository server
that anyone can set up on any device, be it a Raspberry Pi or a beefy server.
Rieter should be usable for anything from a small personal repository all the
way to a full mirror of a distribution's package server.
Most users however don't need a full copy of a distro's package repository, so
Rieter also provides a "smart mirror" mode. In this mode, a Rieter instance
only syncs packages that have been requested before, e.g. from a previous
system update. This way, your updates will still be a lot faster as the
required packages are cached, but packages you don't use don't get stored,
saving you a lot of storage space.
Something that also fits nicely in this concept is mirroring. Rieter will
support automatically mirroring upstream repositories. This could be used to
support your distribution by setting up a new public mirror, or to speed up
your updates by keeping a mirror of the repositories in your local network.
Only once I've created a robust repository server that can be used on its own
will I start looking towards the package build system. This system will of
course be redesigned from the ground up to (hopefully) eliminate all the
technical debt that's been accumulating in the Vieter codebase over the years.
With these two concepts combined, I hope to create a great ecosystem on which
one can build anything from personal repositories to entire distributions.