| 31 Oct 2023 |
 Foxboron | In reply to @raitobezarius:matrix.org
But structural purity increase the chance of reproducibility I don't think so, there are too many weird issues for stuff to be unreproducible and different domains of "problems" | 13:25:22 |
 Julien | Given than we have the same concept of reproducible build environments, and that most work towards build reproducibility is upstreamed, I'd expect Guix and Nix to have similar performances, but I'd be interested to see | 13:25:26 |
 raitobezarius | In reply to @foxboron:archlinux.org
I don't think so, there are too many weird issues for stuff to be unreproducible and different domains of "problems" While I agree there are many weird issues, we are getting better and better at bounding the class of problems | 13:26:34 |
| raitobezarius | But I know you are not very convinced except if we show you proofs :P | 13:27:01 |
| raitobezarius | So I won't try to convince you until we have proper numbers and testing stories | 13:27:11 |
| raitobezarius | (and you are right to doubt) | 13:27:20 |
| Foxboron | In reply to @raitobezarius:matrix.org
But I know you are not very convinced except if we show you proofs :P Having >80k packages of varying degrees of quality is going to make it a very hard problem :p | 13:28:08 |
| raitobezarius | Fair, but we only need to start by intersecting this set with other distro first | 13:28:29 |
| Foxboron | Guix (and Arch for that matter) get an easier job because we do have a limit of what we decide to support | 13:28:29 |
| raitobezarius | And see if we are already as fine as them | 13:28:33 |
| Foxboron | The ISO targets are nice goals though :) | 13:28:50 |
| raitobezarius | Once we are there, we can start looking at the rest of the observable universe :P | 13:28:53 |
| raitobezarius | I think we need the "Archlinux goal" | 13:29:02 |
| raitobezarius | Or the Guix goal | 13:29:05 |
| raitobezarius | After that, it will be a new type of discussion of "what does it mean to pursue a reproducibility goal for nixpkgs?" | 13:29:18 |
| Julien | debian has really good numbers | 13:29:20 |
| raitobezarius | Any project having a good Reproducible Builds team, I guess :P | 13:29:41 |
| Foxboron | You'll also have inflated numbers because you deal with "paths" and repackage ecosystems. | 13:29:49 |
| raitobezarius | The Repology statistics doesn't include them AFAIK | 13:30:05 |
| Foxboron | In reply to @julienmalka:matrix.org
debian has really good numbers The numbers you are looking at are not real reproductions, fyi | 13:30:05 |
| raitobezarius | I mean | 13:30:07 |
| raitobezarius | Include repackaged ecosystems | 13:30:10 |
| raitobezarius | Yes | 13:30:11 |
| raitobezarius | Not the "paths" though | 13:30:16 |
| raitobezarius | There are too many ways to count what is the artifacts output size of nixpkgs | 13:30:31 |
 atemu12 | In reply to @raitobezarius:matrix.org
Include repackaged ecosystems It does. | 13:30:33 |
| raitobezarius | In reply to @atemu12:matrix.org
It does. Yes yes, I clarified it does include repackaged ecosystems but not the "paths" | 13:30:46 |
| raitobezarius | I think it's fair to include repackaged ecosystems though | 13:30:54 |
| raitobezarius | AUR does have python3-XYZ | 13:30:59 |
| raitobezarius | For many XYZ | 13:31:01 |
