Skip to content

G
gitlab-ce
Switch branch/tag
  1. 25 Mar, 2019 5 commits
    • Luke Duncalfe's avatar
      Add changelog entry · f51ee0fa
      Luke Duncalfe authored
      f51ee0fa
    • Luke Duncalfe's avatar
      Update ee specs to correctly select MR author · bcb756cf
      Luke Duncalfe authored 
      When ce MR https://gitlab.com/gitlab-org/gitlab-ce/merge_requests/26144
was merged in, it caused failures in ee specs
https://gitlab.com/gitlab-org/gitlab-ee/issues/10567.

The fix for these specs is to select an "author" of the MR by first
filtering merge commits from the commit collection. This mirrors the
behaviour of MergeRequest#authors.

Previously, this test passed as MergeRequest#commits would always return
commits that lacked parent_ids, as they are initialized from data
kept in MergeRequestDiffCommit which does not have parent_ids data. This
meant that all MR commits were considered not to be merge commits, even
though in gitaly some did have parent ids.

However, now, CommitCollection#enrich! is called
CommitCollection#without_merge_commits is called (as it is when
MergeRequest#authors is called), so commit data is fetched from gitaly
and now have parent_ids data. This cause the tests to fail, because the
"author" being chosen from the merge request commits was in fact the
user who had committed a merge commit.
      bcb756cf
    • Luke Duncalfe's avatar
      Handle blank projects in CommitCollection#enrich! · 4844018b
      Luke Duncalfe authored 
      A project is needed in order to fetch data from gitaly. Projects can be
absent from commits in certain rare situations (like when viewing a MR
of a deleted fork). In these cases, assume that the enriched data is not
needed.

See this comment:

https://gitlab.com/gitlab-org/gitlab-ce/merge_requests/26144#note_152191106

"It's led to a dilemma about where to "fix" this in code. I'm going
to fix it by allowing CommitCollection#enrich! to just return unenriched
commits when a project is missing, essentially "silently failing". I
hope this is the right decision. It's going with the assumption that
calls in these situations in the future are not needing the full data.
The alternative would be to allow CommitCollection#enrich! to still
error, but handle it in the methods that call #enrich!, however that
might lead to brittleness in future when working with project-less MRs."
      4844018b
    • Luke Duncalfe's avatar
      Scope out merge commits in MergeRequest spec · ae13abe4
      Luke Duncalfe authored 
      Previously the code for excluding merge commits from the commit
collection (CommitCollection#without_merge_commits) was not working
when the commits had come from a merge request. Now that this has been
fixed, these tests fails. They should always have been written to
exclude merge commits when comparing.
      ae13abe4
    • Luke Duncalfe's avatar
      Enrich commits with full data in CommitCollection · 7d3d34dd
      Luke Duncalfe authored 
      Allow incomplete commit records to load their full data from gitaly.

Commits can be based on a Hash of data retrieved from PostgreSQL, and
this data can be intentionally incomplete in order to save space.

A new method #gitaly? has been added to Gitlab::Git::Commit, which
returns true if the underlying data source of the Commit is a
Gitaly::GitCommit.

CommitCollection now has a method #enrich which replaces non-gitaly
commits in place with commits from gitaly.

CommitCollection#without_merge_commits has been updated to call this
method, as in order to determine a merge commit we need to have parent
data.

Fixes https://gitlab.com/gitlab-org/gitlab-ce/issues/58805
      7d3d34dd
  3. 21 Mar, 2019 7 commits
  5. 20 Mar, 2019 28 commits
GitLab Nexedi Edition | About GitLab | About Nexedi | 沪ICP备2021021310号-2 | 沪ICP备2021021310号-7