Conversation
jopejoe1
force-pushed
the
recurseIntoAttrsWith
branch
from
ff6819d to
0f04718
Compare
nixpkgs-ci
bot
added
10.rebuild-linux: 1-10
wolfgangwalther
left a comment
wolfgangwalther
left a comment
There was a problem hiding this comment.
LGTM overall; just some nits.
I believe the first four commits should be squashed, they form a single logical unit.
This makes it possible to change the recursion of attrsets depending on if we are evaling for search, hydra or eval
jopejoe1
force-pushed
the
recurseIntoAttrsWith
branch
from
0f04718 to
9f1ea86
Compare
Sure thing, also applied your nits. |
| recursionMode = mkOption { | ||
| type = types.uniq ( | ||
| types.enum [ | ||
| "hydra" | ||
| "eval" | ||
| "search" | ||
| ] | ||
| ); | ||
| default = "eval"; |
There was a problem hiding this comment.
Pinging the @NixOS/nixpkgs-ci team, because eval recursionMode will be unused initially. However, if the Eval jobs were to pass recursionMode=eval to override recursionMode=hydra in release.nix, it would be possible to run CI's eval for more packages than those built on hydra.
Not done in here, because this probably needs better reporting of "rebuilds" - not all packages that are evaluated in "eval" mode will actually count as rebuilds for hydra, so the labels would be off initially...
nixpkgs-ci
bot
added
the
12.approvals: 1
3 participants
Made this after reading @wolfgangwalther idea in #481627 (comment). This is my initial implementation of it; it could probably use a lot of improvements.
Things done
passthru.tests.nixpkgs-reviewon this PR. See nixpkgs-review usage../result/bin/.