Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

2.0.5: gitdb.test.db.test_git.TestGitDB, gitdb.test.db.test_ref.TestReferenceDB, gitdb.test.performance.test_pack_streaming.TestPackStreamingPerformance fail #51

Open
dvzrv opened this issue Aug 27, 2019 · 6 comments

Comments

@dvzrv
Copy link

dvzrv commented Aug 27, 2019

Hi!

I'm currently maintaining gitdb(2) for Arch Linux. During packaging I would like to be able to run the tests, but they fail on me due to relatively undocumented behavior in regards to the test git repository:
python-gitdb-2.0.5-check.log

I'm using the pypi source tarballs for version 2.0.5:
https://files.pythonhosted.org/packages/c4/5c/579abccd59187eaf6b3c8a4a6ecd86fce1dfd818155bfe4c52ac28dca6b7/gitdb2-2.0.5.tar.gz

After build (before installation), I run the tests and setup a git repository for test purposes:

local TEST_TMPDIR=$(mktemp -d)                                                                                                                                                              
( cd "$TEST_TMPDIR" && git init . )
export GITDB_TEST_GIT_REPO_BASE="${TEST_TMPDIR}/.git"
nosetests -vd

We're currently on:

  • git 2.23.0
  • python-setuptools 41.2.0
  • python-smmap 2.0.5
  • python-nose 1.3.7

Any hint/help would be greatly appreciated!

@Byron
Copy link
Member

Byron commented Aug 28, 2019

As in GitPython, an initialised git repository along with some history are assumed for the tests to run successfully.
When looking at the (freshly run) tests on CI, I see the performance tests are skipped there. This might be as they are known to fail on travis as it does check out the repository with low history depth (even though 1000 seems high enough).

I don't think the performance tests are testing anything new or add to coverage, which is why you could skip them as well.

For the other two non-performance failures, there might just not be enough history. If git init . does not create a single commit, then there is no reference either. Please make sure there is some history.

Apologies for the inconvenience, it's clear that unit tests should be isolated and not depend on any state they don't control themselves.

Please let me know if this helped, or if there are any further questions.

@dvzrv
Copy link
Author

dvzrv commented Sep 28, 2019

Why not create a git repository on the fly as an asset?
That kind of decoupling would be really great!

@dvzrv
Copy link
Author

dvzrv commented Sep 29, 2019

This at least works for nearly all tests:

  local TEST_TMPDIR=$(mktemp -d)
  (
    cd "$TEST_TMPDIR"
    git init
    git config user.name "Test User"
    git config user.email "test@user.org"
    for commit in {1..50}; do
      touch "file${commit}"
      git add "file${commit}"
      git commit -m "file${commit}"
    done
  )
  export GITDB_TEST_GIT_REPO_BASE="${TEST_TMPDIR}/.git"
  nosetests -vd -e 'test_pack_writing'

@Byron
Copy link
Member

Byron commented Sep 29, 2019

That looks neat! Do you think you could supply this with a PR? Thanks so much!

@dvzrv
Copy link
Author

dvzrv commented Sep 29, 2019

@Byron I guess it depends on what you understand under pull request.
As part of the test suite (have mainly worked with pytest actually), or as part of the travis setup?

@Byron
Copy link
Member

Byron commented Sep 30, 2019

I think Travis could call a script which contains that code. If I remember correctly, one can set an environment variable for the repository it should use for tests. @dvzrv

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Development

No branches or pull requests

2 participants