Hi, On Tue, 3 Nov 2020 at 15:26, Ludovic Courtès wrote: > 2. Have a Cuirass job running on ci.guix.gnu.org to build and publish > the disarchive-db. One question is: how does the database scale? And only the real world can show it. ;-) > Ludo’, who’s still very much excited about these perspectives! Sounds awesome! On my side, I asked twice on #swh-devel if it is possible to setup a higher rate limit for one specific machine. I have in mind one machine located at my place (Univ. Paris, ex Paris 7 Diderot) because of proximity and because I want to generate (script) some report about how much Guix is in SWH. Whatever! Instead, we could ask for Berlin or for one machine of INRIA Bordeaux, maybe the machine running guix.gnu.org or the one running hpc.guix.info. WDYT? BTW, not related to tarballs and I have not worked so much on (running out of time), but I would like to integrate hg-fetch and svn-fetch with SWH, first to "guix lint -c archival" then second to sources.json. The save seems not so hard, but the lookup needs some experiments with the SWH API. The big picture is to have all the ingestion of the Guix packages done by the automatically generated sources.json file and not via time-to-time "guix lint -c archival" (should be recommended for custom channels). All the best, simon