| 2 Apr 2025 |
Martin Schwaighofer | Why do I see this difference between the builders of the same unresolved and resolved content-addressed derivation? 🤔
diff /nix/store/q2nzjyqf4w19w3mgbkn34k2as5hrvwh1-builder.sh /nix/store/ckzrg0f0bdyx8rf703nc61r3hz5yys9q-builder.sh
22c22
< printf '%s ' "${propagatedUserEnvPkgs[@]}" > $out/nix-support/propagated-user-env-packages
---
> printf '%s ' $propagatedUserEnvPkgs > $out/nix-support/propagated-user-env-packages
| 13:10:54 |
Robert Hensing (roberth) | I haven't found time to look into this properly, whether this is a bug or a missing feature, or both. The main constraint for closureInfo etc, is that the info needs to be reproducible, so for example no signatures or other mutable store metadata. ca:fixed: seems like something that should be possible to include | 14:08:20 |
Robert Hensing (roberth) | The --load-db format has come up before. It is entirely forward-incompatible, so we may need to introduce any additions as a new format | 14:09:42 |
p14 | In reply to @roberthensing:matrix.org The --load-db format has come up before. It is entirely forward-incompatible, so we may need to introduce any additions as a new format Glancing at the load db there are signs the signature is there; is it just not being imported properly, I wonder 🤔 | 14:39:37 |
John Ericson | emily ElvishJerricco OK we discussed a bunch and we're liking the sort of compromise you all proposed | 20:30:03 |
John Ericson | 2.28 in 25.05 has Mic92's combo build (or something like it) | 20:30:27 |
John Ericson | 2.28 after 25.05 is componentized | 20:30:36 |
John Ericson | 2.29 in all branches has componentized (2.29 is very unimportant on 25.05 except for new version dogfooders) | 20:31:03 |
John Ericson | Also when we re-introduce nix git, we should use the componentized version for that | 20:32:23 |
John Ericson | how does that sound? | 20:32:27 |
John Ericson | I would like to merge Robert Hensing (roberth)'s open PR right away, for sake of the newer versions and git, and also because I like how it makes a package set for the dependencies, even with the monolithic package for Nix itself | 20:33:19 |
John Ericson | Also, since you two (and others) have such strong opinions about this, it would be great if you signed yourselves up as maintainers in Nixpkgs for this :D | 20:35:00 |
John Ericson | especially because, as ElvishJerricco mentioned, actually the mkMesonPackage stuff and whatnot really ought to not be Nix-specific, but reused for other things like (potentially) systemd | 20:35:32 |
emily | SGTM, especially if we can get a componentized-compatible overrides interface for patching/build flags/env.NIX_CFLAGS_COMPILE on the monolithic one | 20:40:43 |
emily | sorry, I definitely do not have the time and do not feel I understand the packaging well enough to :( | 20:41:03 |