Some repositories are wrongly configured and they have symbolic references (e.g.
a HEAD) that points to a non-existent branch. In the general case, we still want
to load these origins; Create a dangling branch for these symbolic references to
point to.
Details
Details
- Reviewers
vlorentz ardumont - Group Reviewers
Reviewers - Commits
- rDLDGee22325a193d: Add a default target for symbolic reference branches
new tox test added
Diff Detail
Diff Detail
- Repository
- rDLDG Git loader
- Branch
- master
- Lint
No Linters Available - Unit
No Unit Test Coverage - Build Status
Buildable 12523 Build 19019: Phabricator diff pipeline on jenkins Jenkins console · Jenkins Build 19018: arc lint + arc unit
Event Timeline
Comment Actions
Build is green
Patch application report for D3188 (id=11330)
Rebasing onto dfd5be7b14...
Current branch diff-target is up to date.
Changes applied before test
commit 0c31a06b94f6f7cccd1db85317afce2032229461
Author: Nicolas Dandrimont <nicolas@dandrimont.eu>
Date: Thu May 28 10:31:14 2020 +0200
Add a default target for symbolic reference branches
Some repositories are wrongly configured and they have symbolic references (e.g.
a HEAD) that points to a non-existent branch. In the general case, we still want
to load these origins; Create a dangling branch for these symbolic references to
point to.See https://jenkins.softwareheritage.org/job/DLDG/job/tests-on-diff/12/ for more details.
Comment Actions
LGTM, but I think it's missing some tests:
- Logging on dangling branch, and absence of logging when there isn't any. (Helpful pytest fixture: https://docs.pytest.org/en/latest/logging.html#caplog-fixture )
- Checking there is no warning if we first get a symref from A to B, then a reference from B to C.
Comment Actions
Yeah, I know of the caplog fixture, thanks. I will not be adding pytest fixtures in the mess of unittest class inheritance that these tests currently are.
- Checking there is no warning if we first get a symref from A to B, then a reference from B to C.
As we don't really control what comes "first" (dulwich is reading the symbolic refs, not us) I don't think this (very) edge case can really be tested for reliably. I don't even know if chains of symbolic references exist in practice.
| swh/loader/git/utils.py | ||
|---|---|---|
| 1 | I am "the pre-commit hook" for copyright headers (I missed those yesterday... ;) | |
Comment Actions
Build is green
Patch application report for D3188 (id=11335)
Rebasing onto dfd5be7b14...
Current branch diff-target is up to date.
Changes applied before test
commit ee22325a193dbc2b52b37496f3374c6bf0c1d1e1
Author: Nicolas Dandrimont <nicolas@dandrimont.eu>
Date: Thu May 28 10:31:14 2020 +0200
Add a default target for symbolic reference branches
Some repositories are wrongly configured and they have symbolic references (e.g.
a HEAD) that points to a non-existent branch. In the general case, we still want
to load these origins; Create a dangling branch for these symbolic references to
point to.See https://jenkins.softwareheritage.org/job/DLDG/job/tests-on-diff/14/ for more details.