| 6 Sep 2021 |
 @gtrunsec:matrix.org | David Arnold (blaggacao): please check it out, The latest digga(fup rev) will make devos impossible to use the binary cache. | 19:17:50 |
 David Arnold (blaggacao) | Why? | 19:20:31 |
| David Arnold (blaggacao) | I've also noticed some strange building stuff but didn't attribute it to digga... Maybe a nix issue? | 19:21:11 |
| @gtrunsec:matrix.org |
- https://github.com/divnix/digga/commit/59bb53182b539bc578df2d8dcdfeff08085a35ce, I have not idea about this issue, It even has to rebuild a lot of things
nix flake lock --update-input bud && nix flake lock --update-input digga && nix develop
[4/0/1633 built, 0.0 MiB DL] building libunistring-0.9.10 (configurePhase): checking whether long double and double are t
| 19:23:46 |
| David Arnold (blaggacao) | https://github.com/divnix/digga/commit/e1c8966ef05931da375fe9c88a1f5058b79b6883 might be the offending commit... It is curious though that CI didn't seem to suffer from undue rebuilds | 19:23:50 |
| David Arnold (blaggacao) | so I attributed it to some local stuff on my machine... | 19:24:26 |
| David Arnold (blaggacao) | https://github.com/divnix/digga/runs/3517106739 was rather swift. | 19:24:54 |
| @gtrunsec:matrix.org | I can be sure this problem comes from fup | 19:25:27 |
| David Arnold (blaggacao) | That run was also rather swift... https://github.com/divnix/digga/runs/3519430044 | 19:27:12 |
| David Arnold (blaggacao) | Did nix develop --command -- check-downstream on latest digga/main make those rebuild observable for you? | 19:31:07 |
| David Arnold (blaggacao) | I can now just see endless querying on caches already going on for more than a minute or so p | 19:32:24 |
| David Arnold (blaggacao) | and indeed building busybox hahaha | 19:32:49 |
| @gtrunsec:matrix.org | *
- https://github.com/divnix/digga/commit/59bb53182b539bc578df2d8dcdfeff08085a35ce, I have no idea about this issue, It even has to rebuild a lot of things
nix flake lock --update-input bud && nix flake lock --update-input digga && nix develop
[4/0/1633 built, 0.0 MiB DL] building libunistring-0.9.10 (configurePhase): checking whether long double and double are t
| 19:34:52 |
| David Arnold (blaggacao) | Redacted or Malformed Event | 19:35:15 |
| David Arnold (blaggacao) | gtrunsec: Testing seems to prove you correct, without https://github.com/divnix/flake-utils-plus/commit/ed14e98f77d9c1eb75f2aa86c4f29a284513b7be cache is back to normal. | 19:41:02 |
| David Arnold (blaggacao) | the only thing I can think of is that the config merging changed something... https://github.com/divnix/flake-utils-plus/commit/ed14e98f77d9c1eb75f2aa86c4f29a284513b7be#diff-f5cc39913cfe16126a23b19cce7fa1e584b31dbc8c5211050e2f6d7b7fb76b3cR97-R110 | 19:41:50 |
| David Arnold (blaggacao) | I'll factorize this commit... | 19:42:16 |
| @gtrunsec:matrix.org | Thanks | 19:42:50 |
| David Arnold (blaggacao) | Thanks for reporting!! | 19:43:36 |
| David Arnold (blaggacao) | https://github.com/divnix/digga/commit/7115e02b64f86a78e47477c16a121629269f3988 reverts for now. | 19:46:04 |
| David Arnold (blaggacao) | Honestly, it looks like the bug might be with upstream config merge: https://github.com/NixOS/nixpkgs/blob/d9de842f58b20c35e4e71831c865e308ed00a8d7/nixos/modules/misc/nixpkgs.nix#L17-L32 | 19:55:26 |
 @timdeh:matrix.org | https://github.com/fluidattacks/makes | 20:30:04 |
| David Arnold (blaggacao) | Looks like that is a Colombian company 🚀 | 20:47:28 |
| David Arnold (blaggacao) | Paisas berracos! | 20:48:18 |
| David Arnold (blaggacao) | https://gitlab.com/fluidattacks/product/-/tree/master/makes | 22:07:21 |
| David Arnold (blaggacao) | interesting record of how they came around to do this. | 22:07:40 |
| David Arnold (blaggacao) | I really like how they strictly avoid overlapping namespaces and how they seem to have an a strong anti-corruption layer that fends against the nixpkgs "mess". | 22:19:54 |
| David Arnold (blaggacao) | You can really see a spirit of: "Even cooler than to figure things out is still to get stuff done." | 22:20:37 |
| David Arnold (blaggacao) | Anyone still say, nixkpgs cannot be split... | 22:21:27 |
 tomberek | In reply to @blaggacao:matrix.org
I really like how they strictly avoid overlapping namespaces and how they seem to have an a strong anti-corruption layer that fends against the nixpkgs "mess". https://github.com/fluidattacks/makes/blob/9fc15d3ad0a612e8c17f6c9eb6f6f24dfeb7300d/src/cli/main/main.py#L128-L145 | 22:35:19 |
