[Suggestion] Host your own /rpc endpoint for metadata retrieval
Hi,
I'd like to propose going back to an /rpc based approach for retrieving metadata from the AUR:
Instead of directly querying the API on aurweb, you can self-host your own instance of the /rpc service.
This would solve several of the current issues:
-
8.8 MB download of metadata archive is still a huge bandwidth waste.
While the Manjaro users are routed via the manjaro CDN, pamac users on other distros are still fetching it from aurweb, crippling the bandwidth of the aur. -
When pamac loads the metadata file into memory, it's using huge amounts of RAM (+ 500MB).
regards, mo