!sBfrWMVsLoSyFTCkNv:nixos.org

OfBorg

165 Members
Number of builds and evals in queue: <TBD>61 Servers

Load older messages


SenderMessageTime
25 Dec 2023
@raphi:tapesoftware.netraphi i think because nix_2_3 is in aliases.nix so it doesn't get picked up in CI
nixVersions.nix_2_3 should work
14:56:25
@trofi:matrix.orgtrofiOh, that's tricky! Thank you!16:24:12
@lily:lily.flowersLily Foster
In reply to @raphi:tapesoftware.net
i think because nix_2_3 is in aliases.nix so it doesn't get picked up in CI
nixVersions.nix_2_3 should work
yeah but commit message should just be for builds iirc. it uses changed paths/attrs from the outpath calculation iirc (though maybe it had other problems with determining the changed path there idk or maybe it is something alias heckery)
16:59:51
@lily:lily.flowersLily Fosterif i remember, i'll double check the code later17:00:15
@lily:lily.flowersLily Foster* yeah but commit message should just be for builds iirc. it uses changed paths/attrs from the outpath calculation iirc for maintainer pings (though maybe it had other problems with determining the changed path there idk or maybe it is something alias heckery)17:00:31
@artturin:matrix.orgArtturin
In reply to @lily:lily.flowers
yeah but commit message should just be for builds iirc. it uses changed paths/attrs from the outpath calculation iirc (though maybe it had other problems with determining the changed path there idk or maybe it is something alias heckery)
The maintainers for all reverse feps aren't pinged
17:01:04
@lily:lily.flowersLily Foster
In reply to @artturin:matrix.org
The maintainers for all reverse feps aren't pinged
i guess that's true, i'm probably just misremembering then
17:03:55
@lily:lily.flowersLily Foster(i really thought it based it on files touched for some reason though rather than commit messages....)17:04:24
@lily:lily.flowersLily Foster* (i really thought it based it on source files touched for some reason though rather than commit messages....)17:04:32
@lily:lily.flowersLily Foster(idk am on mobile and also with family rn so i can't check code)17:04:49
@artturin:matrix.orgArtturinOfborg doesn't use the source names for anything IIRC17:06:03
@trofi:matrix.orgtrofi I changed it to nixVersions.nix_2_3 and ofborg still did not add anyone :( 19:11:02
26 Dec 2023
@lily:lily.flowersLily Foster
In reply to @artturin:matrix.org
Ofborg doesn't use the source names for anything IIRC
i'm looking at the ofborg code and it definitely just does maintainer pings for any changed outpaths that also had a file modified in the PR that corresponds to the derivations' meta.maintainers, src, pname, or version attr positions: https://github.com/NixOS/ofborg/blob/de415d372959b7e6fc6b2f6c95f0c21e5010348d/ofborg/src/maintainers.nix#L56-L64
01:31:56
@lily:lily.flowersLily Foster (i.e. so it requires that 1. the package has an attr, 2. the outpath for that package attr has changed, and 3. the meta.maintainers, src, pname, or version attribute for that package must be defined in one of the files that was changed) 01:32:59
@lily:lily.flowersLily Foster
In reply to @trofi:matrix.org
I changed it to nixVersions.nix_2_3 and ofborg still did not add anyone :(
I'm trying to trace back why that didn't work here, because it definitely should have even with the original commit message
01:33:41
@lily:lily.flowersLily Foster * (i.e. so it requires that 1. the package has an attr on any supported arch, 2. the outpath for that package attr/arch has changed, and 3. the meta.maintainers, src, pname, or version attribute for that package must be defined in one of the files that was changed) 01:34:56
@lily:lily.flowersLily Foster Ah so the nix common function doesn't directly inherit the meta attrset so it doesn't think the attr pos for any of those 4 is in pkgs/tools/package-management/nix/default.nix (just pkgs/tools/package-management/nix/common.nix), it didn't think there were any relevant pings since none of the changed files "corresponded" to the changed attrs 01:39:00
@lily:lily.flowersLily FosterSo I'm not sure this is a bug so much as an unfortunate consequence of ofborg relying on invisible magic and that magic being broken by the nix derivation refactor01:39:33
@lily:lily.flowersLily Foster * So I'm not sure this is a bug so much as an unfortunate consequence of ofborg relying on invisible magic and that magic being broken by the nix derivation refactor that made that common function 01:39:50
@lily:lily.flowersLily Foster

a minor refactor that is something like this would make the magic work again if you really wanted it to trofi:

diff --git a/pkgs/tools/package-management/nix/common.nix b/pkgs/tools/package-management/nix/common.nix
index 0ea47dd7e17c..6d62d8a31bfd 100644
--- a/pkgs/tools/package-management/nix/common.nix
+++ b/pkgs/tools/package-management/nix/common.nix
@@ -5,7 +5,7 @@
 , hash ? null
 , src ? fetchFromGitHub { owner = "NixOS"; repo = "nix"; rev = version; inherit hash; }
 , patches ? [ ]
-, maintainers ? with lib.maintainers; [ eelco lovesegfault artturin ]
+, meta ? {}
 }:
 assert (hash == null) -> (src != null);
 let
@@ -243,10 +243,10 @@ self = stdenv.mkDerivation {
     '';
     homepage = "https://nixos.org/";
     license = licenses.lgpl2Plus;
-    inherit maintainers;
+    maintainers = with lib.maintainers; [ eelco lovesegfault artturin ];
     platforms = platforms.unix;
     outputsToInstall = [ "out" ] ++ optional enableDocumentation "man";
     mainProgram = "nix";
-  };
+  } // meta;
 };
 in self
diff --git a/pkgs/tools/package-management/nix/default.nix b/pkgs/tools/package-management/nix/default.nix
index 4652ddb76a5b..e1d566ecabaa 100644
--- a/pkgs/tools/package-management/nix/default.nix
+++ b/pkgs/tools/package-management/nix/default.nix
@@ -157,7 +157,7 @@ in lib.makeExtensible (self: ({
     patches = [
       patch-monitorfdhup
     ];
-    maintainers = with lib.maintainers; [ flokli raitobezarius ];
+    meta.maintainers = with lib.maintainers; [ flokli raitobezarius ];
   }).override { boehmgc = boehmgc-nix_2_3; };
 
   nix_2_10 = common {
01:53:44
@raitobezarius:matrix.orgraitobezarius
In reply to @lily:lily.flowers

a minor refactor that is something like this would make the magic work again if you really wanted it to trofi:

diff --git a/pkgs/tools/package-management/nix/common.nix b/pkgs/tools/package-management/nix/common.nix
index 0ea47dd7e17c..6d62d8a31bfd 100644
--- a/pkgs/tools/package-management/nix/common.nix
+++ b/pkgs/tools/package-management/nix/common.nix
@@ -5,7 +5,7 @@
 , hash ? null
 , src ? fetchFromGitHub { owner = "NixOS"; repo = "nix"; rev = version; inherit hash; }
 , patches ? [ ]
-, maintainers ? with lib.maintainers; [ eelco lovesegfault artturin ]
+, meta ? {}
 }:
 assert (hash == null) -> (src != null);
 let
@@ -243,10 +243,10 @@ self = stdenv.mkDerivation {
     '';
     homepage = "https://nixos.org/";
     license = licenses.lgpl2Plus;
-    inherit maintainers;
+    maintainers = with lib.maintainers; [ eelco lovesegfault artturin ];
     platforms = platforms.unix;
     outputsToInstall = [ "out" ] ++ optional enableDocumentation "man";
     mainProgram = "nix";
-  };
+  } // meta;
 };
 in self
diff --git a/pkgs/tools/package-management/nix/default.nix b/pkgs/tools/package-management/nix/default.nix
index 4652ddb76a5b..e1d566ecabaa 100644
--- a/pkgs/tools/package-management/nix/default.nix
+++ b/pkgs/tools/package-management/nix/default.nix
@@ -157,7 +157,7 @@ in lib.makeExtensible (self: ({
     patches = [
       patch-monitorfdhup
     ];
-    maintainers = with lib.maintainers; [ flokli raitobezarius ];
+    meta.maintainers = with lib.maintainers; [ flokli raitobezarius ];
   }).override { boehmgc = boehmgc-nix_2_3; };
 
   nix_2_10 = common {
please send the patch; looks good to me :(
05:16:21
@trofi:matrix.orgtrofi I must say I expected $attr.meta.maintainers to Just Work :( 07:50:13
@trofi:matrix.orgtrofiFiled https://github.com/NixOS/ofborg/issues/668 if that makes sense.10:03:50
@trofi:matrix.orgtrofi As a workaround WDYT of adding a pos attribute to the lookup list? I could sprinkle a few pos = allocations for nix edit to do the right thing. 10:36:42
@trofi:matrix.orgtrofi * As a workaround WDYT of adding a pos attribute to the lookup list? I could sprinkle a few pos = annotations for nix edit to do the right thing. 10:36:55
@trofi:matrix.orgtrofi Proposed ofborg change as https://github.com/NixOS/ofborg/pull/669 10:42:59
@trofi:matrix.orgtrofi And proposed nix pos change as https://github.com/NixOS/nixpkgs/pull/276892 10:49:12
28 Dec 2023
@piegames:matrix.orgpiegames changed their display name from piegames to piegames [☎ 9712].01:33:45
30 Dec 2023
@trofi:matrix.orgtrofi Looking at https://github.com/NixOS/nixpkgs/pull/277751 and at https://github.com/NixOS/nixpkgs/pull/277741 would it be reasonable for ofborg to try to eval all of passthru? Just an eval for uncatchable errors a few levels deep (like 2-3). 16:59:31
@raitobezarius:matrix.orgraitobezarius changed their display name from raitobezarius (DECT 2128) to raitobezarius.19:53:26

Show newer messages


Back to Room ListRoom Version: 6