| 13 Apr 2026 |
 K900 | I think you'd have to start at nixpkgs | 05:04:08 |
| K900 | And probably structuredAttrs? | 05:04:30 |
 blitz | I almost have a prototype of attic that solves this for now by calling the Nix CLI. Will check gorgon once I have something working. One last obstacle: Is there a stable way to list the signatures of a store path without resorting to nix path-info (which has changed output formats a couple of times...) ? | 07:35:48 |
 Lisanna | Oh for sure, the whole point would be to make nixpkgs do the heavy lifting here | 08:11:59 |
 zoë (she/her) | is there any way to see previous build logs after rebuilding a derivation? kind of like journalctl's --boot option to see previous boot logs instead of the current one. it seems like nix overwrites them now; would there be any obstacle to renaming them to foo.drv-1, foo.drv-2, etc when a new build is launched? | 08:41:12 |
| zoë (she/her) | * is there any way to see previous build logs after rebuilding a derivation? kind of like journalctl's --boot option to see previous boot logs instead of the current one. it seems like current nix overwrites them; would there be any obstacle to renaming them to foo.drv-1, foo.drv-2, etc when a new build is launched? | 08:41:27 |
 boop-north | I did sudo -i nix upgrade-nix and now can't run commands related to nix. `nix/var/nix/profiles/default/bin exists in path | 09:56:08 |
| boop-north | I did sudo -i nix upgrade-nix and now can't run commands related to nix. nix/var/nix/profiles/default/bin exists in path | 09:56:15 |
|  Baum(she/they) | 10:08:03 |
 Lotte (it/its)/Cinny (she/her) θΔ& | interesting. a path can be garbage collected between the time a build gets started and the time the build runs?
at least i saw that earlier today. some rust crate was complaining about its dependencies being missing and not the typical hydra complaints about some specific store paths not being present locally | 11:35:39 |
 raitobezarius | is your daemon still running? | 11:36:42 |
| zoë (she/her) | that sounds like a bug? on the contrary, a few months ago when i tried this (either on cppnix or lix, can't remember), it wouldn't collect/delete dependencies of running builds or open shells | 11:38:09 |
| Lotte (it/its)/Cinny (she/her) θΔ& | unfortunately due to the build having been overwritten by a retry i can’t provide logs of it happening | 11:38:58 |
| zoë (she/her) | or maybe i'm misunderstanding what you mean by "build started" and "build runs" | 11:39:04 |
| zoë (she/her) | haha exactly what i was complaining about a few messages above :^) | 11:39:30 |
| zoë (she/her) | * haha exactly what i was complaining about a few messages above :p | 11:39:42 |
| Lotte (it/its)/Cinny (she/her) θΔ& | https://hydra.chir.rs/build/72679#tabs-buildsteps i guess there’s this | 11:39:49 |
 Sergei Zimmerman (xokdvium) | In reply to @blokyk:matrix.org
that sounds like a bug? on the contrary, a few months ago when i tried this (either on cppnix or lix, can't remember), it wouldn't collect/delete dependencies of running builds or open shells Temp roots should handle this. At least in cppnix we’ve been trying to squash those issues. Not sure Lix fixed all of those | 11:40:01 |
| Lotte (it/its)/Cinny (she/her) θΔ& | i should probably just add hydra to the conflicts of the gc | 11:40:09 |
| Lotte (it/its)/Cinny (she/her) θΔ& | Aborted: [31;1merror:[0m dependency '[35;1m/nix/store/6nar2q94jybd9lgsa1s41zwk6q2d5il6-rustblake2-0.10.6-lib[0m' of '[35;1m/nix/store/01dqlw7c55b7icy4zx69mqvvgv50fxkm-rustlibsignal-message-backup-0.1.0.drv[0m' does not exist, and substitution is disabled
is what i would expect to happen | 11:40:23 |
| Lotte (it/its)/Cinny (she/her) θΔ& | the failed line above was the juicy one | 11:40:28 |
| Lotte (it/its)/Cinny (she/her) θΔ& | because that was full of errors where dependencies couldn’t be found by rustc | 11:40:57 |
| boop-north | I just reinstalled lix. When I ran it, it was. Afterwards I'm not sure | 11:41:32 |
| Lotte (it/its)/Cinny (she/her) θΔ& | which implies at least to me that between the time the build was started and the time rustc was run the store path got deleted | 11:41:51 |
| Lotte (it/its)/Cinny (she/her) θΔ& changed their profile picture. | 14:31:39 |
| Lotte (it/its)/Cinny (she/her) θΔ& changed their profile picture. | 14:31:44 |
| Lotte (it/its)/Cinny (she/her) θΔ& changed their profile picture. | 14:35:24 |
| Lotte (it/its)/Cinny (she/her) θΔ& changed their profile picture. | 14:35:24 |
| Lotte (it/its)/Cinny (she/her) θΔ& changed their profile picture. | 14:38:07 |
 whispers [& it/fae] | we vaguely recall some connection between auto-allocate-uids and cgroups—something about one of them having a hard dependency on the other. we can find no evidence of such a relation in the docs, however. have we mentally made this up or is this a thing? | 19:26:22 |
