Conversation: Gitaly Character Encoding

This is the conversation issue for discussing Gitaly character encoding.

Related Issues:

• Gitaly branch names returning ASCII-8BIT and mixing with UTF-8 in views https://gitlab.com/gitlab-org/gitlab-ce/issues/34156
• @yhl3692319 report on Diff::CommitDiff: #209 (comment 33170406)
• Fragile GitLab-Test Repo: https://gitlab.com/gitlab-org/gitlab-ce/issues/32133#note_30927237
• Encode Gitaly diff patches properly https://gitlab.com/gitlab-org/gitlab-ce/merge_requests/12368/

Work:

• https://gitlab.com/gitlab-org/gitlab-ce/merge_requests/12429

Approach

1. The GitLab-Test repo only contains non-ASCII char encodings on filenames. It does not test for non-ASCII char encoded refs.
2. We cannot change the GitLab-Test repo because it is brittle and any changes will break all branches of GitLab-CE: ref https://gitlab.com/gitlab-org/gitlab-ce/issues/32133
3. We need to add non-ASCII UTF8-encoded ref names to the GitLab-Test repo but can’t.
4. @jacobvosmaer-gitlab is going to work on making GitLab-Tests non-brittle using bare-repo solution proposed in https://gitlab.com/gitlab-org/gitlab-ce/issues/32133#note_30927237
5. Once we can change GitLab-Test, we can add non-ASCII UTF-8 ref names to the repo and ensure that we’ve got a proper fix

Affected Endpoints:

• Diff::CommitDiff #222 (closed)
• Ref::FindDefaultBranchName #220 (closed)
• Ref::FindAllBranchNames #220 (closed)
• Ref::FindAllTagNames #220 (closed)
• Ref::FindRefName #221 (closed)