Conversation
Moving all of the custom dependency magic into the package allows us to move it to the by-name hierarchy a lot easier. This however also means breaking any overrides used downstream for the affected dependencies: - fmt - lua - arrow-cpp The last one is now also exposed via passthru for overriding purposes. All expressions evaluate to the same values though. Signed-off-by: benaryorg <[email protected]>
Moving the ceph client to a separate output allows moving the entire package to the by-name structure more easily. Signed-off-by: benaryorg <[email protected]>
Signed-off-by: benaryorg <[email protected]>
|
Tests are currently running. |
nixpkgs-ci
bot
added
10.rebuild-linux: 1-10
|
Tests are happy. diff --git a/pkgs/tools/filesystems/ceph/default.nix b/pkgs/by-name/ce/ceph/package.nix
similarity index 100%
rename from pkgs/tools/filesystems/ceph/default.nix
rename to pkgs/by-name/ce/ceph/package.nixDoes this need a changelog entry? cc @nh2 |
nh2
left a comment
nh2
left a comment
There was a problem hiding this comment.
Looks good to me, I left 2 minor comments, and will push a fix for them on top of this PR (that does not change the eval).
From my perspective this can me merged independent of whteher @benaryorg wants to implement those.
I built it and tests pass.
This PR indeed greatly reduces the diff to #443671, so it will make that one easier to review.
Side remark: I noticed this output:
/nix/store/av6vjncjgylmdakdxgf2ni91vqmnd34z-wrap-python-hook/nix-support/setup-hook: line 84: warning: command substitution: ignored null byte in input
Rewriting ELF>0Y@���@8@$#@@PPPSSH�>H�>�>�>�>q�q�@A@A@A..@ho@ho@ho���2h�sh�sh�s��\o�\o�\o@�\o�\o�\o @ho@ho@hS�td�\o�\o�\o@P�td�OK�OK�OK��Q�tdR�td@ho@ho@ho�w�w/nix/store/daamdpmaz2vjvna55ccrc30qw3qb8h6d-glibc-2.40-66/lib/ld-linux-x86-64.so.2�h#!WKF!X�A�F7m`��\�H
Vz M�M�]��7�R+BQ$y 3�J~�8�! to #!/nix/store/sm8xkb2i0b6dbsll92iyfavy0b0cw0p6-python3-3.11.14
That's independent from this PR, but might be good to understand why such binary gets rewritten to a text hashbang, or if this is due to my terminal, how that comes to be.
nixpkgs-ci
bot
added
12.approvals: 1
I think that's fine without, any such breakage will be at eval time and those things are a hack anyway; if somebody overrides those, they know exactly what they are doing and which package is involved. |
@benaryorg Which ones (beyond #443671) does it conflict with? Just we we know what we need to ping / clean up. |
|
I click "Merge when ready". If something blocks the automerge (e.g. ofborg not being able to build it due to resource use, or it taking long on Darwin), any committer please merge, or ping me here and I will force-merge. |
|
Why was this merged with these commit messages??? |
@dotlambda Do you mean that they start with The rest of the commit messages looks correct to me (they describe what they do, and the commits are correctly split into the independent refactorings we desired to be extracted from #443671). |
Yes. |
Potential conflicts I was aware of:
Unless I missed some it seems nothing unexpected had any conflicts then.
I apologise for that (and at least this embarrassment will serve to help me remember the correct commit message conventions). |
3 participants
Three commits as follows (chronologically):
ceph-clienttoceph.clientThe two earlier commits serve to minify the last one.
I held off on this PR until the package itself was fixed so we'd be able to ensure the package and tests still build.
This conflicts directly with several PRs to the package, however I believe that moving this over now reduces maintenance burden in the future.
Big thanks to #443671 which I used as a rough reference.
Things done
passthru.tests.nixpkgs-reviewon this PR. See nixpkgs-review usage../result/bin/.Add a 👍 reaction to pull requests you find important.