Hello,
Reading the doc at https://docs.astral.sh/uv/guides/integration/gitlab/, how do I cache the uv pip install ?
I would have expected the following .gitlab-ci.yml to success.
variables:
UV_SYSTEM_PYTHON: 1 # remove the need to create a venv
UV_CACHE_DIR: .uv-cache
default:
image: ghcr.io/astral-sh/uv:0.5-python3.11-alpine
cache:
- key: test
paths:
- $UV_CACHE_DIR
uv-install:
stage: .pre
script:
- uv pip show rdflib || true # fails as expected
- uv pip install rdflib
- uv cache prune --ci
- uv pip show rdflib
test:
stage: test
script:
- uv pip show rdflib # fails hereBut I have the error where rdflib is not installed in the test stage.
Note: I'm using gitlab-ci-local to test locally
$ gitlab-ci-local
Unable to retrieve default remote branch, falling back to `main`.
The default remote branch can be set via `git remote set-head origin <default_branch>`
parsing and downloads finished in 57 ms.
json schema validated in 220 ms
uv-install starting ghcr.io/astral-sh/uv:0.5-python3.11-alpine (.pre)
uv-install copied to docker volumes in 516 ms
uv-install imported cache 'test' in 274 ms
uv-install $ uv pip show rdflib || true
uv-install > Using Python 3.11.11 environment at: /usr/local
uv-install > warning: Package(s) not found for: rdflib
uv-install $ uv pip install rdflib
uv-install > Using Python 3.11.11 environment at: /usr/local
uv-install > Resolved 2 packages in 14ms
uv-install > warning: Failed to hardlink files; falling back to full copy. This may lead to degraded performance.
uv-install > If the cache and target directories are on different filesystems, hardlinking may not be supported.
uv-install > If this is intentional, set `export UV_LINK_MODE=copy` or use `--link-mode=copy` to suppress this warning.
uv-install > Installed 2 packages in 10ms
uv-install > + pyparsing==3.2.1
uv-install > + rdflib==7.1.1
uv-install $ uv cache prune --ci
uv-install > Pruning cache at: .uv-cache
uv-install > Removed 3422 files (59.5MiB)
uv-install $ uv pip show rdflib
uv-install > Using Python 3.11.11 environment at: /usr/local
uv-install > Name: rdflib
uv-install > Version: 7.1.1
uv-install > Location: /usr/local/lib/python3.11/site-packages
uv-install > Requires: pyparsing
uv-install > Required-by:
uv-install finished in 1.42 s
uv-install exported cache .uv-cache 'test' in 633 ms
test starting ghcr.io/astral-sh/uv:0.5-python3.11-alpine (test)
test copied to docker volumes in 503 ms
test imported cache 'test' in 283 ms
test $ uv pip show rdflib
test > Using Python 3.11.11 environment at: /usr/local
test > warning: Package(s) not found for: rdflib
test finished in 1.35 s FAIL 1
PASS uv-install
FAIL test
> Using Python 3.11.11 environment at: /usr/local
> warning: Package(s) not found for: rdflib
pipeline finished in 3.76 s
Apparently it fails because uv pip installs in /usr/local which is outside the project directory, hence, can't be cached by Gitlab.
To make it work, I have to NOT use UV_SYSTEM_PYTHON: 1 and run ux venv in my .pre stage.
Is the config below supposed to be the correct way to cache uv pip install ?
variables:
UV_CACHE_DIR: .uv-cache
default:
image: ghcr.io/astral-sh/uv:0.5-python3.11-alpine
cache:
- key: test
paths:
- $UV_CACHE_DIR
- .venv
uv-install:
stage: .pre
script:
- uv venv --allow-existing
- uv pip show rdflib || true
- uv pip install rdflib
- uv cache prune --ci
- uv pip show rdflib
test:
stage: test
script:
- uv pip show rdflib # worksOr is there another way to install packages from the pyproject.toml file than uv pip install -r pyproject.toml ?
Thank you,
Pierre
Hello,
Reading the doc at https://docs.astral.sh/uv/guides/integration/gitlab/, how do I cache the
uv pip install?I would have expected the following
.gitlab-ci.ymlto success.But I have the error where rdflib is not installed in the test stage.
Note: I'm using gitlab-ci-local to test locally
Apparently it fails because
uv pipinstalls in/usr/localwhich is outside the project directory, hence, can't be cached by Gitlab.To make it work, I have to NOT use
UV_SYSTEM_PYTHON: 1and runux venvin my.prestage.Is the config below supposed to be the correct way to cache
uv pip install?Or is there another way to install packages from the
pyproject.tomlfile thanuv pip install -r pyproject.toml?Thank you,
Pierre