1. 18 Apr, 2019 4 commits
    • georg's avatar
      Merge branch 'master' into master-to-release-4.0 · 0216e8ff
      georg authored
    • paz's avatar
      Merge branch 'ci-make-code-coverage-usable' into 'release-4.0' · c6a75553
      paz authored
      CI: make code coverage actually usable
      See merge request !287
    • georg's avatar
      CI: codespell: ignore code coverage dir leading to false positives · c60e18ec
      georg authored
      Due to the introduced artifacts to store the code coverage report, and
      to "hand it over" from one stage to the next, these artifacts are now
      downloaded as well before the codespell job. This lead to false
      positives, spelling errors within this directory, we're not really
      interested in. Therefore, ignore this directory if running codespell.
    • georg's avatar
      CI: Deploy code coverage overview as HTML to make it actually usable · 2f779414
      georg authored
      Code coverage was added to the CI some time ago. Up until now, a part of
      the result was printed at the end of the rspec test jobs, accessible via
      the individual job logs, barely usable.
      This commit introduces deploying the HTML result and making it
      accessible via a dedicated page, allowing to search within and filter
      the data, in a much more convenient and appealing way.
      While I'm not hunting for 100 percent code coverage, I think having at
      least a better insight into this topic is quite valuable. I hope this
      leads to a better understanding of our test suite, possibly identifying
      weak points (which then can be improved) and in the end to a better code
  2. 17 Apr, 2019 2 commits
  3. 16 Apr, 2019 1 commit
    • georg's avatar
      CI: Add job to check spec filenames for trailing '_spec' part · 4f7c0ae5
      georg authored
      Some time ago we added some spec files, however, the filenames were
      missing the trailing '_spec' part, so these were never executed.
      This commit adds a new CI job to check for this problem and to catch it
      earlier next time. For now, this job allows errors. Lets revisit this in
      the future to probably make the result fatal, that is, block a merge
      request based on the result.
      This job is limited to changes made in the relevant spec directories,
      however, due to the nature GitLab works currently, if pushing new
      branches, this job will be executed regardless of which files were
      changed. However, as this job is quite "light" regarding resources, this
      shouldn't be of any concern.
      Some tests were done via intermediate commits to ensure this works as
      Relates !284
  4. 14 Apr, 2019 1 commit
  5. 08 Apr, 2019 1 commit
    • georg's avatar
      Drop pin_keys (originally introduced to support v2 to v3 migration) · 0eb91fdd
      georg authored
      Remove obsolete code, which looked for subscriptions without an
      associated key, and tried to find a distinctly matching key. Originally
      this was implemented to help with a shortcoming of code which handled v2
      to v3 migration. As the support to migrate v2 lists to v3 is dropped in
      v4, remove the pinning code as well.
      Closes #411
  6. 07 Apr, 2019 9 commits
  7. 06 Apr, 2019 1 commit
  8. 05 Apr, 2019 1 commit
  9. 02 Apr, 2019 2 commits
  10. 01 Apr, 2019 2 commits
    • georg's avatar
      Drop deprecated x-listname keyword · 2345deb3
      georg authored
      Closes #374
    • georg's avatar
      CI: Enforce rubocop, disallow errors · 231f8278
      georg authored
      We've introduced rubocop some time ago, to help with code linting. We
      still don't like the name, but it might improve code quality in the
      future. Due to some errors in the beginning, the results weren't
      reliable. This seems fixed since a while, therefore, disallow errors and
      close the corresponding issue.
      Closes #382
  11. 31 Mar, 2019 1 commit
  12. 30 Mar, 2019 4 commits
  13. 26 Mar, 2019 1 commit
  14. 01 Mar, 2019 1 commit
  15. 25 Feb, 2019 1 commit
    • paz's avatar
      Re-define associations between Account, Subscription, and List. · 02032fa5
      paz authored
      This is required to make them work with non-persisted accounts.
      We need non-persisted accounts because when the authorization is done on behalf
      of a keyword_handler there might be no account yet (because email
      authentication is bassed on signatures, not the account), and as we can't set a
      password we can't save the account. (And we don't want to save the account with
      a random password to avoid brute force opportunities.)
  16. 22 Feb, 2019 5 commits
  17. 20 Feb, 2019 3 commits