Commit 46303c8
docs: record compatibility audits of 115 GitPython dependents
The five automated downstream profiles do not describe GitPython's wider
runtime, development and documentation consumers. Record four completed audit
batches of 31, 30, 39 and 15 distinct projects selected from September 2026
PyPI downloads plus known Git-focused users. The inventory covers every
identified current declared consumer within the top-15,000 snapshot and
348,530,835 monthly distribution requests, including extras. Downloads are
neither unique users nor attributable GitPython installations; the additions
and metadata limits also prevent a strict popularity top-115 claim.
Retain all four reports and JSON manifests under `test/downstream/`, linked
from its README. Preserve release and archive provenance, commands, import
proofs, backend results, baseline comparisons and coverage limits at their
original tested revisions. Sources, environments, logs, JUnit XML and
supplemental runners remain in ignored audit caches; a fresh environment must
reconstruct the disclosed harness. The five-profile CI runner remains the
maintained automation. Upstream tests, mocks and expectations are unchanged;
selected helpers, unused declarations and omitted application stacks are
identified explicitly. Gix selections retain documented Git CLI fallbacks.
Twelve separate fixes restore immutable index readers, global `-C` selection,
separate-Git-directory context, the `GitDB` selector, local tracking branches,
revision tree/path diagnostics, explicit literal-path construction, editable
version reporting, built-wheel version isolation, historical `Union` and
`Optional` imports, and nonempty configuration-section assignment.
Keep the remaining compatibility work visible. Kas 5.5 has three SSH setup
failures because `add_section()` cannot create an empty section through the
supported native APIs; downstream can use `set_value()` directly, while full
compatibility requires a native section-edit API. sbxloop 2.1.36 has five
constructor-subclass failures because native discovery validates metadata
before its overridden `config_reader()` runs; restoring that view requires
supported discovery/include controls without bypassing Git format validation.
Index mapping mutation and automatic environment expansion remain unsupported.
Airflow's ambiguous bare/mirror fetch results expose a separate GitPython
adapter gap: Git `fetch --porcelain` supplies full destination identities for
a future CLI adapter fix; Gix fetch remains a fallback until per-ref outcomes
are exposed. Other retained failures are downstream defects, baseline behavior
or explicitly recorded harness/coverage limits, not passing results.
Historical final full suites after the last production fix recorded 1,832
CLI passes and 1,934 Gix passes, with 162/60 skips, one xfail and 59 passed
subtests each. The last batch adds 117 upstream and 187 supplemental passes,
one baseline-inherited Dstack failure and two Jira skips per backend; its
separate MkNotebooks Binder build passes with both backends and the release.
Those historical results are not reruns on this rewritten stack.
Consolidate the four `docs:` record commits at the stack tip while retaining
the feature commit and twelve fixes in their original order. Resolve only
changelog context conflicts, keeping implementation entries with their fixes
and the subclass record here. Verify an identical final tree, unchanged
non-documentation patches, fourteen final commits with one top `docs:` commit,
115 distinct manifest projects, recorded CLI/Gix parity, local report links
and `git diff --check`. No new tests or hosted CI were run and no push occurred.1 parent 3640db4 commit 46303c8
11 files changed
Lines changed: 24955 additions & 4 deletions
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
@@ -51,6 +51,24 @@ variables literal; `expand_vars=True` is unsupported. Callers wanting expansion | |||
| 51 | 51 | must perform it before passing the path. Discovery still uses Gix/Git after | |
| 52 | 52 | ordinary input-path canonicalization. | |
| 53 | 53 | ||
| 54 | + Repository construction validates storage and configuration through Gix/Git; | ||
| 55 | + overriding `Repo.config_reader()` does not replace that native discovery parser. | ||
| 56 | + sbxloop 2.1.36's `_MetadataRepo` subclass relied on the released constructor's | ||
| 57 | + permissive Python discovery and overridden no-includes reader. Five upstream | ||
| 58 | + tests still differ: malformed includes prevent opening, malformed local config | ||
| 59 | + is rejected before its custom `ProvisionError`, and version-0 repositories | ||
| 60 | + with version-1 storage extensions are rejected before its format check. Released | ||
| 61 | + GitPython 3.2.1 passes those five tests. This is a breaking subclass contract, | ||
| 62 | + not a demonstrated Gix-versus-Git mismatch: both backends reject the metadata. | ||
| 63 | + GixPython `OpenOptions` has strict/lossy configuration controls and overrides, | ||
| 64 | + but no supported option to open a repository while omitting its local include | ||
| 65 | + graph. A replacement must expose separately controlled metadata discovery and | ||
| 66 | + literal local configuration access with Git-compatible format validation. | ||
| 67 | + Do not bypass validation or reconstruct discovery/storage in Python to restore | ||
| 68 | + the old constructor. Downstream must construct its trusted metadata view before | ||
| 69 | + opening it with `Repo`, and handle native discovery errors explicitly; its | ||
| 70 | + original tests remain failed in the audit. | ||
| 71 | + | ||
| 54 | 72 | `git.GitDB` and `git.db.GitDB` remain compatibility aliases for | |
| 55 | 73 | `GitCmdObjectDB`, so legacy `Repo(..., odbt=GitDB)` selectors use the same | |
| 56 | 74 | managed backend. Aider 0.86.2 uses that selector for every repository open. | |
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
@@ -61,6 +61,12 @@ API changes | |||
| 61 | 61 | * Editable/source imports expose the release version in ``git.__version__``, | |
| 62 | 62 | matching built distributions. Salt's GitPython provider checks can parse the | |
| 63 | 63 | version instead of rejecting the unstamped ``"git"`` build placeholder. | |
| 64 | + * ``Repo`` construction validates repository configuration through Git/Gix. | ||
| 65 | + A subclass override of ``config_reader()`` no longer controls discovery or | ||
| 66 | + suppresses its include graph. Malformed configuration and invalid repository | ||
| 67 | + extensions can fail before downstream custom validation runs. sbxloop's | ||
| 68 | + metadata-reader subclass exposes this remaining compatibility break; callers | ||
| 69 | + must prepare their trusted metadata view before opening it with ``Repo``. | ||
| 64 | 70 | * Revision root trees such as ``Repo.rev_parse("HEAD:")`` retain their empty | |
| 65 | 71 | path and can be iterated. Missing paths within a valid tree revision retain | |
| 66 | 72 | ``Tree.join()``'s ``KeyError`` and missing-component message, restoring MCP | |
| Original file line number | Diff line number | Diff line change | |
|---|---|---|---|
@@ -1,5 +1,31 @@ | |||
| 1 | 1 | # Downstream compatibility | |
| 2 | 2 | ||
| 3 | + The [2026-10-22 audit](audit-2026-10-22.md) covers 31 projects, including major | ||
| 4 | + runtime, development and documentation users. Its [manifest](audit-2026-10-22.json) | ||
| 5 | + records releases, source hashes, commands, results and coverage limits. The runner | ||
| 6 | + below automates five CI profiles; it is not the complete audit inventory. | ||
| 7 | + | ||
| 8 | + The [second 30-project batch](audit-2026-10-22-batch-2.md) brings the total to | ||
| 9 | + 61 processed projects, with four further local fixes and explicitly retained | ||
| 10 | + compatibility, downstream and harness failures. Its | ||
| 11 | + [manifest](audit-2026-10-22-batch-2.json) records the next popularity-ranked | ||
| 12 | + selection and results on both backends. | ||
| 13 | + | ||
| 14 | + The [third 39-project batch](audit-2026-10-22-batch-3.md) brings the audit to | ||
| 15 | + 100 distinct projects. It records further local compatibility fixes, three | ||
| 16 | + remaining Kas configuration API failures, and failures inherited from released | ||
| 17 | + GitPython. Its [manifest](audit-2026-10-22-batch-3.json) includes verified release | ||
| 18 | + archives, commands, import proofs and coverage limits. Additional selections are | ||
| 19 | + popularity-ranked; the original known-user additions mean this is a 100-project | ||
| 20 | + audit rather than a strict mathematical top 100. | ||
| 21 | + | ||
| 22 | + The [remaining 15-project batch](audit-2026-10-23-batch-4.md) brings the audit to | ||
| 23 | + 115 distinct projects and completes every identified current declared consumer | ||
| 24 | + in the September snapshot of the top 15,000 distributions. Its | ||
| 25 | + [manifest](audit-2026-10-23-batch-4.json) records released callers, tests, baseline | ||
| 26 | + comparisons and limits. This covers the identified distribution download weight; | ||
| 27 | + it does not attribute GitPython installs or imply complete application coverage. | ||
| 28 | + | ||
| 3 | 29 | Run released projects' GitPython-related tests against this checkout with | |
| 4 | 30 | [uv](https://docs.astral.sh/uv/) and Git 2.52 or newer: | |
| 5 | 31 | ||
@@ -35,19 +61,22 @@ settings, pytest options/plugins, and Python import paths are cleared before | |||
| 35 | 61 | setup and testing. Select a different Git executable by putting its directory | |
| 36 | 62 | first on `PATH`. | |
| 37 | 63 | ||
| 38 | - ## Selection and coverage | ||
| 64 | + ## Automated CI profiles | ||
| 39 | 65 | ||
| 40 | - The selection uses current runtime users, including optional end-user features, | ||
| 66 | + These five profiles use current runtime users, including optional end-user features, | ||
| 41 | 67 | ranked by September 2026 PyPI distribution downloads. Development, documentation, | |
| 42 | 68 | and test-only dependencies are excluded. Downloads are not unique installations. | |
| 43 | 69 | ||
| 44 | 70 | | Project | Distribution downloads | Last tested release | Selected coverage | | |
| 45 | 71 | | --- | ---: | --- | --- | | |
| 46 | 72 | | LangChain Community | 27,882,881 | 0.4.2 | 2 upstream GitLoader tests: real clones, commits, checkout, tree traversal, ignored paths, and remote validation | | |
| 47 | - | MLflow (`mlflow-skinny`) | 25,850,354 | 3.17.0 | 47 upstream tests: 31 repository/project/model-versioning cases plus 16 Git context and credential-redaction contract cases | | ||
| 73 | + | MLflow (`mlflow-skinny`) | 25,850,354 | 3.17.2 | 53 unique upstream repository/project/model-versioning, Git context and credential-redaction cases in the expanded audit | | ||
| 48 | 74 | | Bandit | 24,935,372 | 1.9.4 | 12 upstream baseline CLI tests: real repository creation, commits, branches, resets, discovery, and dirty state | | |
| 49 | 75 | | SWE-bench | 22,942,741 | 5.0.2 | 4 supplemental integration cases for `AutoContextManager`; no upstream tests cover its GitPython callers | | |
| 50 | - | DataHub (`acryl-datahub`) | 5,019,402 | 1.7.0.14 | 7 upstream tests passed, 1 credential-dependent skip: public clone/checkout, SSH timeout, exception/redaction contracts, and configuration | | ||
| 76 | + | DataHub (`acryl-datahub`) | 5,019,402 | 1.7.0.16 | 8 upstream tests passed, 1 credential-dependent skip: clone/checkout, SSH timeout, exception/redaction contracts, configuration and ODCS sourcing | | ||
| 77 | + | ||
| 78 | + The runner automates the original selections. The linked audit broadens MLflow | ||
| 79 | + with `tests/projects/test_utils.py` and DataHub with its ODCS Git sourcing test. | ||
| 51 | 80 | ||
| 52 | 81 | Source: [top-pypi-packages](https://hugovk.github.io/top-pypi-packages/top-pypi-packages.min.json), | |
| 53 | 82 | snapshot updated **2026-10-01 12:40:51 UTC**. Its | |
0 commit comments