diff options
| author | Eelco Dolstra <edolstra@gmail.com> | 2020-02-02 12:41:23 +0100 |
|---|---|---|
| committer | Eelco Dolstra <edolstra@gmail.com> | 2020-02-02 12:41:23 +0100 |
| commit | 94a94da07513a9290cd857844d28843e3895614b (patch) | |
| tree | a2a16cd691b653ec1080ea0bad6c179e18cb6b64 /tests | |
| parent | fad9faf3545f115e33e522eb11802e2825c8c425 (diff) | |
Substitute flake inputs
This improves reproducibility and may be faster than fetching from the
original source (especially for git/hg inputs, but probably also for
github inputs - our binary cache is probably faster than GitHub's
dynamically generated tarballs).
Unfortunately this doesn't work for the top-level flake since even if
we know the NAR hash of the tree, we don't know the other tree
attributes such as revCount and lastModified.
Fixes #3253.
Diffstat (limited to 'tests')
| -rw-r--r-- | tests/flakes.sh | 1 |
1 files changed, 1 insertions, 0 deletions
diff --git a/tests/flakes.sh b/tests/flakes.sh index 141bb9779..c6f9d78df 100644 --- a/tests/flakes.sh +++ b/tests/flakes.sh @@ -235,6 +235,7 @@ mv $registry.tmp $registry # Test whether flakes are registered as GC roots for offline use. # FIXME: use tarballs rather than git. rm -rf $TEST_HOME/.cache +nix-store --gc # get rid of copies in the store to ensure they get fetched to our git cache _NIX_FORCE_HTTP=1 nix build -o $TEST_ROOT/result git+file://$flake2Dir#bar mv $flake1Dir $flake1Dir.tmp mv $flake2Dir $flake2Dir.tmp |