Hello AUR consumers,
This question is for projects who use (or intend to use) the AUR's archive
data. I'm trying to gather the best route for our git repository
archives which will soon be released.
The issue at hand:
- The AUR needs to reduce its bandwidth usage as much as possible,
while still satisfying consumers. Let's say in a world where 5%
of archive consumers need Popularity data, then we would be pretty
much wasting bandwidth serving popularity data to the other 95%.
Ideas for working around this:
- If Popularity data is only needed by a small amount of consumers,
we're thinking about serving out a `popularity.git` repository
and excluding popularity data from `metadata.git`. This means that
clients that need the popularity data would need to consume two
repositories, which complicates the consumption process, but
popularity.git would not be shared with those who don't need it.
Depending on information we gather, we may keep with popularity
data being housed inside of `metadata.git`, as we have a new
popularity calculation method which reduces the amount of changes
drastically.
As a project maintainer which wants to use AUR archives, could you
please answer the following:
1. What's the project?
2. Do you need Popularity?
- Why?
Thanks a lot for your time; any input here helps a ton.
If you want some information about the current devel of the git
repository archives, see https://gitlab.archlinux.org/kevr/aurweb/-/blob/git-archives/doc/git-archiv…
This state does not currently reflect our new popularity strategy,
but it does show the general structure of repositories generated for
these archives.
Regards,
Kevin
--
Kevin Morris
Software & Linux Enthusiast