From cad599f3526b0f1515e323a817fb5821a4a175b0 Mon Sep 17 00:00:00 2001 From: Wolfgang Walther Date: Tue, 21 Oct 2025 11:37:59 +0200 Subject: [PATCH] top-level/by-name-overlay: remove optimization TODO We tried to apply this, but this won't consistently help performance. While it can potentially make full eval a tad faster, we'd pay the price when evaluating single packages. --- pkgs/top-level/by-name-overlay.nix | 4 ---- 1 file changed, 4 deletions(-) diff --git a/pkgs/top-level/by-name-overlay.nix b/pkgs/top-level/by-name-overlay.nix index 9a011439596a..e6fe5b721556 100644 --- a/pkgs/top-level/by-name-overlay.nix +++ b/pkgs/top-level/by-name-overlay.nix @@ -41,10 +41,6 @@ let # if the overlay has to be applied multiple times packageFiles = mergeAttrsList (mapAttrsToList namesForShard (readDir baseDirectory)); in -# TODO: Consider optimising this using `builtins.deepSeq packageFiles`, -# which could free up the above thunks and reduce GC times. -# Currently this would be hard to measure until we have more packages -# and ideally https://github.com/NixOS/nix/pull/8895 self: super: { # This attribute is necessary to allow CI to ensure that all packages defined in `pkgs/by-name`