Initial support for file upload by minitriga · Pull Request #71 · opsmill/nornir-infrahub

minitriga · 2026-04-10T10:42:07Z

Initial support for file upload

Summary

Add CoreFileObject support to nornir-infrahub with two new Nornir task functions:

upload_file_object — Upload a local file (or raw bytes) to an Infrahub node that inherits from CoreFileObject. Creates the node if it doesn't exist, updates it if the file changed, and skips the upload entirely when the SHA-1 checksum matches (idempotent). On update, data attribute changes are persisted even when the file content is unchanged.
download_file_object — Download file content from a CoreFileObject node. Returns base64-encoded binary plus a UTF-8 text view for known text MIME types. Optionally writes to a local save_to path; skips the write when the local file already matches the server checksum.

Mirrors the CoreFileObject support added to the Ansible collection in opsmill/infrahub-ansible#317.

Changes

nornir_infrahub/plugins/tasks/file_object.py — new module with the two task functions, helpers (_validate_file_object_kind, _lookup_existing_object, _resolve_upload_source), and the TEXT_MIME_TYPES constant. Idempotency and checksum comparison are delegated to the SDK via node.upload_if_changed(source, name) and node.matches_local_checksum(source).
nornir_infrahub/plugins/tasks/__init__.py — export the two new functions.
nornir_infrahub/utils.py — small get_client(task) helper shared between the file_object tasks (ready for artifact.py to adopt as a follow-up).
tests/unit/test_file_object.py — 28 unit tests covering create, update, idempotent skip, attribute-only updates on the skip path, rejection of relationship and unknown keys in data on update, invalid kind, missing file, explicit object_id miss, ambiguous filename lookup, text/binary MIME download, save_to dir-vs-file handling, local-match skip, and overwrite-on-mismatch.
pyproject.toml — bump infrahub-sdk to >=1.20.1,<2 (first release shipping upload_if_changed / matches_local_checksum / UploadResult from opsmill/infrahub-sdk-python#963); scope PLR0911/PLR0912/PLR0913 ignores to the tasks directory; keep the three ty rule disables that cover broad SDK-dynamic-type patterns and drop the two that inline # type: ignore comments already cover.
docs/ — auto-generated plugin reference for the two new tasks plus generator fixes the rebase surfaced (pipe-escaping in type names, multi-line description flattening, corrected relative path in _plugin_index.mdx, MD012/MD060 disables folded into the canonical .markdownlint.yml and the redundant .markdownlint.json removed).

Usage

from nornir_infrahub.plugins.tasks import upload_file_object, download_file_object

# Upload a file (forwards **kwargs to client.create — e.g. allow_upsert, timeout)
result = nr.run(
    task=upload_file_object,
    kind="AvdHostvarFile",
    file_path="/path/to/hostvars.json",
)

# Download a file (save_to writes to disk; binary/text always returned)
result = nr.run(
    task=download_file_object,
    kind="AvdHostvarFile",
    object_id="abcd-1234-efgh-5678",
    save_to="/tmp/downloads/",
)

Behavior notes:

**kwargs are forwarded to client.create and only apply on the create path; they are not used when updating an existing object.
On update, the data dict only sets attributes; passing a relationship key or an unknown key fails the task instead of being silently dropped (use the SDK directly to change relationships on existing nodes).
A typo'd explicit object_id fails the task rather than silently creating a new node; upsert-on-miss is still in effect for hfid and the filename fallback.

Test plan

Unit tests pass (45/45 — 28 new file_object + 17 existing inventory)
ruff check / ruff format --check pass
pylint 10.00/10
ty check passes
markdown-lint / vale / docs build pass on CI across Python 3.10–3.13
Integration tested against a live Infrahub via test: add integration test suite for InfrahubInventory (NORNIR-5) #79's testcontainers harness. tests/integration/test_file_object.py adds a Testing.TestFile schema inheriting from CoreFileObject and covers upload lifecycle (create → idempotent skip → update), download round-trip, download-skip on local match, metadata-on-skip persistence, explicit-object_id not-found failure, and unknown-data-key rejection. 6/6 pass locally against testcontainers in ~70s; excluded from CI by test: add integration test suite for InfrahubInventory (NORNIR-5) #79's -m 'not integration' addopts (Docker-only). Run with pytest tests/integration/test_file_object.py -m integration.

cloudflare-workers-and-pages · 2026-04-10T10:42:52Z

Deploying nornir-infrahub with Cloudflare Pages

Latest commit:	`9de8e3b`
Status:	✅ Deploy successful!
Preview URL:	https://f9a351b0.nornir-infrahub.pages.dev
Branch Preview URL:	https://001-object-file-support.nornir-infrahub.pages.dev

View logs

wvandeun · 2026-04-19T08:50:07Z

We should not use the term node but object

wvandeun · 2026-04-19T08:55:26Z

+)
+
+
+def _get_client(task: Task) -> InfrahubClientSync:


We should probably move this a utils module. This could be useful in other contexts?

Moved to nornir_infrahub/utils.py as get_client(task) and imported from file_object.py. Good for the artifact.py tasks to adopt too; left that for a follow-up to keep this PR focused. (commit b35f894)

wvandeun · 2026-04-19T08:58:04Z

The SDK provides a capability to create a FileObject from a bytes object as well. We should probably add support for this https://docs.infrahub.app/python-sdk/sdk_ref/infrahub_sdk/node#upload_from_bytes

wvandeun · 2026-04-19T09:33:34Z

+    data = data or {}
+
+    # Look up existing node
+    existing_node = None


Do we need to implement this here? AFAIK the SDK behaves in a idempotent way already when using allow_upsert=True. Ofcourse this requires that you have setup a proper unique constraint and/or HFID

allow_upsert=True avoids the duplicate-create, but it still re-uploads the file bytes on every call and doesn't let us report changed=False to Nornir — the whole point of the client-side SHA-1 compare is (a) skip the upload entirely when the content hasn't changed, and (b) drive Nornir's changed flag correctly (matches how nornir_utils.write_file behaves). I'd like to keep the checksum-skip for those reasons. That said, allow_upsert is now reachable via the new **kwargs passthrough, so callers who do want upsert semantics on create can opt in.

wvandeun · 2026-04-19T09:36:35Z

+    data: dict[str, Any] | None = None,
+    node_id: str | None = None,
+    hfid: list[str] | None = None,
+    branch: str | None = None,


Should we support kwargs here as well, to pass arguments to client.create

Added **kwargs passthrough on upload_file_object. It forwards to client.create(...), so callers can pass allow_upsert, timeout, request_context, etc. New test test_upload_forwards_kwargs_to_create covers it.

wvandeun · 2026-04-19T09:37:09Z

+def upload_file_object(
+    task: Task,
+    kind: str,
+    file_path: str,


Use Path for file_path?

Done. file_path now accepts str | Path and we normalise to Path internally.

wvandeun · 2026-04-19T09:57:43Z

+    kind: str,
+    node_id: str | None = None,
+    hfid: list[str] | None = None,
+    dest: str | None = None,


Use Path here? Also we could come up with a better name for this argument that is a little bit more descriptive?

Also we can probably make this behave in a idempotent fashion? IE the task result should report only if a change was made when a file really changed? Similar to how the write_file in nornir_utils behaves?

All three addressed:

Path: save_to accepts str | Path.

Rename: dest → save_to (more descriptive — "where to save the file locally").

Idempotent: download_file_object now compares the local file's SHA-1 to the server checksum before downloading. If they match the write is skipped and changed=False; otherwise we download, overwrite, and changed=True. Same model as nornir_utils.plugins.tasks.files.write_file. Covered by new tests test_download_skips_write_when_local_matches and test_download_overwrites_when_local_differs.

minitriga · 2026-04-20T11:14:40Z

@wvandeun — review feedback addressed in b35f894. Summary:

Top-level comments

"object" not "node" — node_id → object_id, internal node/existing_node/new_node → obj/existing_obj/new_obj, and docstrings/messages updated throughout.
upload_from_bytes support — upload_file_object now accepts either file_path (disk) or content + file_name (raw bytes). The two are mutually exclusive; the SHA-1 idempotency and **kwargs passthrough work for both paths.

Inline comments — replied on each thread; in short:

_get_client → nornir_infrahub.utils.get_client.
file_path and save_to accept str | Path.
dest renamed to save_to.
download_file_object is now idempotent (skips the write and reports changed=False when the local file's SHA-1 matches the server checksum).
**kwargs forwarded to client.create — allow_upsert, timeout, request_context all reachable.
Kept the client-side checksum-skip on upload; explained in the thread (TL;DR: allow_upsert=True still re-uploads bytes and always reports changed=True, which is not what we want here).

Tests: 38/38 pass. Ruff, ruff format, ty, pylint all clean.

cubic-dev-ai

1 issue found across 19 files

_{Reply with feedback, questions, or to request a fix. Tag @cubic-dev-ai to re-run a review, or fix all with cubic.}

wvandeun · 2026-04-22T07:19:45Z


 ```python
 from nornir import InitNornir
+from nornir.core.plugins.inventory import InventoryPluginRegister


these 2 imports should not be required

wvandeun · 2026-04-22T07:19:55Z

+from nornir_infrahub.plugins.inventory.infrahub import InfrahubInventory

 def main():
+    # Register the custom InfrahubInventory plugin


this is not required

wvandeun · 2026-04-22T09:33:28Z

If we need the idempotent behavior for both uploading and downloading, then it should not really be part of the nornir plugin, but it should be provided by the SDK itself. If not we are going to have to re-implement this in all integrations.

…ical .markdownlint.yml PR #71 originally added a root .markdownlint.json, but the repo's canonical config is .markdownlint.yml (docusaurus-aware: MD033/MD007/MD029/MD047 etc.). markdownlint-cli prefers .json over .yml, so our file shadowed the canonical one and #55's guide/topic docs failed markdown-lint after the rebase. Remove the .json and add the two rules the generated plugin reference tables need (MD012 multiple-blanks, MD060 table padding) to .markdownlint.yml. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

cubic-dev-ai

5 issues found across 16 files

_{Reply with feedback, questions, or to request a fix.

Re-trigger cubic}

lancamat1 · 2026-05-27T11:17:51Z

Testing status

This refactor (moving the CoreFileObject upload/download tasks onto the new infrahub-sdk primitives upload_if_changed / matches_local_checksum, requiring infrahub-sdk ≥ 1.20.1) is currently covered by unit tests only (mocked SDK boundary). It has not yet been integration-tested against a live Infrahub.

Integration coverage is intentionally deferred until #79 (the testcontainers-based integration harness) merges to stable. Once it lands, I'll rebase this branch onto stable and add tests/integration/test_file_object.py — a CoreFileObject-inheriting schema fixture exercising upload-create → idempotent-skip → update → download → download-skip against a live Infrahub.

Holding off on merge until that integration pass is done.

minitriga

Code Review

Overview

Adds two new Nornir tasks (upload_file_object, download_file_object) for Infrahub CoreFileObject nodes, mirroring opsmill/infrahub-ansible#317. Includes a shared get_client() helper, 28 unit tests, doc-template fixes, and an SDK bump. Overall well-built: idempotency semantics are correct, error handling follows the Nornir Result(failed=True) pattern, and test coverage is thorough. SDK interaction details verified below.

Verified correctness ✅

SDK methods exist and behave as assumed. upload_if_changed() / matches_local_checksum() are absent in 1.19.0 and present in 1.20.1 (checked both), so the bump to >=1.20.1 is genuinely required.
Attrs + file both changed is safe. Attributes are staged via setattr before upload_if_changed(), and the SDK's upload_if_changed() calls self.save() on upload — staged attrs persist. The explicit save() on the skip path covers the other branch. Tests cover both (TestUploadAttrsOnSkip).
Download checksum guard is correct. matches_local_checksum() raises ValueError when there's no server checksum, but the code guards with server_checksum and ... first.

Issues

1. Relationships in `data` are silently dropped on update (medium)

The update path in file_object.py only applies keys found in existing_obj._schema.attribute_names:

if attr_name in existing_obj._schema.attribute_names:

On create, relationships in data flow through client.create(data=data) fine. On update, they're silently ignored. The PR's own usage example passes data={"artifact": artifact_id} — presumably a relationship — so the documented example works on first run and silently no-ops on subsequent runs. Either handle relationship_names too, or document that data only updates attributes on existing objects (and ideally warn/fail on unknown keys instead of skipping).

2. Nonexistent explicit `object_id` falls through to create (medium)

_lookup_existing_object returns None on NodeNotFoundError regardless of how the object was identified. If a user passes a typo'd UUID, the task creates a new node instead of failing. Upsert-on-miss is defensible for hfid or the file_name fallback, but an explicit UUID that doesn't exist is almost certainly an error — consider failing there.

3. PR description is stale (low)

Says bump to >=1.19.0; the diff bumps to >=1.20.1 (correctly).
Names upload_from_path(), download_file(), is_file_object() as the motivating methods; the code actually depends on upload_if_changed() and matches_local_checksum() (1.20.x-only).

4. Global ty rule disables are too blunt (low)

pyproject.toml adds project-wide invalid-return-type = "ignore" and invalid-assignment = "ignore", while file_object.py also carries inline # type: ignore[return-type] / [assignment] comments for the same spots. The inline ignores make the global disables redundant — and the global ones silence those rules for all future code. Prefer keeping only the inline ignores.

5. Minor nits

_lookup_existing_object only catches NodeNotFoundError; the file_name__value fallback will raise an unhandled IndexError from the SDK if multiple nodes share a file name (Nornir still marks the task failed, but with a raw traceback rather than a clean message).
str(save_to).endswith("/") won't detect a trailing \ on Windows — save_to_path.is_dir() covers existing dirs, so this only affects not-yet-created directory paths.
download_file_object holds full content in memory and embeds base64 in every Result — Nornir aggregates results across hosts, so this can get heavy for large binaries across many hosts. Maybe worth a docstring note.
**kwargs is forwarded only on the create path, not update — documented that way, just noting the asymmetry (e.g. timeout won't apply to updates).

Code quality & conventions 👍

utils.get_client() matches the existing node._client access pattern in artifact.py — a follow-up refactoring artifact.py onto the helper would be nice.
Pylint per-file ignores (PLR0911/0912/0913) are scoped to tasks/** with explanatory comments — appropriate.
Doc template fixes (pipe-escaping union types, newline flattening in table cells, empty-params guard) are real bug fixes for the generator, and the broken ./references/plugins/ link prefix in readme.mdx.j2 is correctly fixed for where _plugin_index.mdx lands.
Tests are mock-based but assert the right seams (create kwargs passthrough, no-save on no-change, overwrite vs. skip on download). Gaps: no test for the file_name__value fallback lookup, and no test pinning the relationship-on-update behavior from issue 1.

Verdict

Approve with minor changes. Issues 1 and 2 are real behavioral surprises worth fixing (or at minimum documenting) before merge; the rest is polish. The core upload/download logic, idempotency handling, and SDK usage are correct — verified against infrahub-sdk 1.20.1 source.

🤖 Generated with Claude Code

Add upload_file_object and download_file_object Nornir task functions for managing files on Infrahub nodes that inherit from CoreFileObject. Upload supports SHA-1 checksum idempotency (skip re-upload when file is unchanged). Download returns base64-encoded binary content with UTF-8 text decoding for text MIME types, with optional local save. - Bump infrahub-sdk lower bound to >=1.19.0 - Add 13 unit tests covering create, update, skip, download, and error cases - Add plugin reference documentation and sidebar entry - Fix markdownlint, yamllint, and vale for CI compliance

- rename node_id/dest to object_id/save_to (use "object" not "node") - accept str | Path for file_path and save_to - add upload_from_bytes support via content/file_name args - forward **kwargs to client.create (allow_upsert, timeout, ...) - download_file_object is now idempotent: skip write when local SHA-1 matches server - move _get_client into nornir_infrahub.utils for reuse Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

- plugin.mdx.j2 now escapes `|` in type names and flattens newlines in descriptions so types like `str | Path` and multi-line descriptions render as valid markdown tables (fixes markdown-lint MD055/MD056). - .vale/styles/Infrahub/sentence-case.yml: update stale "CoreFileObject node" exception to "CoreFileObject object" to match the renamed heading. Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

#55 reworked the docs generator and added _plugin_index.mdx. Regenerate all plugin reference docs with the current template (escapes pipes in type names, flattens multi-line descriptions) and add the CoreFileObject section to the manually-maintained readme.mdx index. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

…al_checksum Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

…ical .markdownlint.yml PR #71 originally added a root .markdownlint.json, but the repo's canonical config is .markdownlint.yml (docusaurus-aware: MD033/MD007/MD029/MD047 etc.). markdownlint-cli prefers .json over .yml, so our file shadowed the canonical one and #55's guide/topic docs failed markdown-lint after the rebase. Remove the .json and add the two rules the generated plugin reference tables need (MD012 multiple-blanks, MD060 table padding) to .markdownlint.yml. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

- _lookup_existing_object now catches only NodeNotFoundError, so real API errors surface instead of silently triggering a create (review #1) - upload_file_object persists data attribute changes even when the file checksum matches, instead of dropping them on the skip path (review #2) - document **kwargs as (Any, optional) so generated docs show the right type and optionality (review #3) Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

- mark schema_mappings/group_mappings optional in the InfrahubInventory docstring so the generated table shows Required=No, matching the Optional[...] = None signature (review #4) - generate _plugin_index.mdx links relative to its own location instead of prepending references/plugins/, which produced broken doubled paths (review #5) - regenerate plugin reference docs; file_object_tasks.mdx now shows the **kwargs row as type Any / Required No (review #3) Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

- reject relationship and unknown keys in 'data' on update instead of silently dropping them; document that 'data' on update only sets attributes (review #1) - re-raise NodeNotFoundError when the caller passed an explicit object_id, so a typo'd UUID fails instead of creating a new node (review #2) - convert the SDK's IndexError on ambiguous filename lookup into a clean failed Result naming the conflict (review #5a) - detect Windows trailing separator on save_to (review #5b) - docstring notes: download memory/base64 footprint (review #5c) and **kwargs being create-only (review #5d) Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

pyproject's [tool.ty.rules] disabled invalid-return-type and invalid-assignment project-wide while file_object.py also carried inline # type: ignore[...] comments at every relevant spot — the globals silenced those rules for all future code unnecessarily. Drop the two globals (keep the three that cover broader SDK-dynamic-type patterns) and move the existing return-type ignore on _validate_file_object_kind from the def line to the actual return line so ty honours it (review #4). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

upload_file_object's create path passes data={} to client.create when the caller supplies neither data nor **kwargs, but the SDK rejects that with 'Either data or a list of keywords must be provided' — a bare upload fails before upload_if_changed gets to set the real file_name. Seed data['file_name'] from the upload source when there is nothing else to send. upload_if_changed refreshes file_name from the actual content before save, so the seed is invisible in the persisted object. Discovered by the new CoreFileObject integration tests. Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

Adds tests/integration/test_file_object.py running against a live Infrahub spun up by #79's testcontainers harness. Loads a TestingTestFile schema that inherits from CoreFileObject plus a label attribute, and exercises: - upload_lifecycle: create → idempotent skip → update on content change - download_returns_content: bytes round-trip via base64 - download_skip_when_local_matches: second download is a no-op - data_attr_persists_on_skip: label is saved even when the file content is unchanged (verifies the metadata-on-skip behaviour) - explicit_object_id_not_found_fails: typo'd UUID fails cleanly - unknown_key_in_data_fails_on_update: bogus data key is rejected Run with 'pytest -m integration' (requires Docker). All 6 pass locally against the testcontainers Infrahub in ~70s. Closes the integration-testing follow-up referenced in the testing-status comment on PR #71 (post-#79). Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

lancamat1 · 2026-06-04T10:27:53Z

Integration tests landed

Following up on the testing-status note — now that #79 is merged, this PR has been rebased onto stable and tests/integration/test_file_object.py was added in 9de8e3b, extending the testcontainers harness with a Testing.TestFile kind that inherits from CoreFileObject.

Six integration tests, all passing locally against testcontainers in ~70s:

upload lifecycle (create → idempotent skip → update on content change)
download round-trip via base64
download skip when the local file already matches the server checksum
data attribute persisted on the file-unchanged skip path (review Add CI files #2)
explicit object_id not-found fails cleanly (review Add CI files #2)
unknown data key on update fails cleanly (review Copy project from main repo #1)

The run also surfaced a real pre-existing UX bug: client.create(data={}) is rejected by the SDK, so a bare upload_file_object(task=…, kind=…, file_path=…) failed before upload_if_changed could set the real file_name. Fixed in b18c4f2 by seeding data['file_name'] from the source when both data and **kwargs are empty (upload_if_changed refreshes it before save, so the seed is invisible).

Run integration tests with: pytest tests/integration/test_file_object.py -m integration (requires Docker; not in CI by design).

minitriga requested a review from a team as a code owner April 10, 2026 10:42

minitriga force-pushed the 001-object-file-support branch from a9bccf9 to e155a78 Compare April 10, 2026 12:21

wvandeun reviewed Apr 19, 2026

View reviewed changes

wvandeun requested changes Apr 19, 2026

View reviewed changes

cubic-dev-ai Bot reviewed Apr 20, 2026

View reviewed changes

Comment thread .vale/styles/Infrahub/sentence-case.yml Outdated

wvandeun requested changes Apr 22, 2026

View reviewed changes

minitriga mentioned this pull request Apr 22, 2026

feat: add SHA-1 idempotency primitives for CoreFileObject opsmill/infrahub-sdk-python#963

Merged

5 tasks

BeArchiTek requested review from lancamat1 and removed request for lancamat1 May 19, 2026 13:32

lancamat1 force-pushed the 001-object-file-support branch from 585dcf2 to 982962f Compare May 27, 2026 10:24

lancamat1 force-pushed the 001-object-file-support branch from 1043710 to 8c45a30 Compare May 27, 2026 10:31

cubic-dev-ai Bot reviewed May 27, 2026

View reviewed changes

minitriga commented Jun 4, 2026

View reviewed changes

minitriga and others added 7 commits June 4, 2026 12:14

build: require infrahub-sdk>=1.20.1 for upload_if_changed/matches_loc…

1de3834

…al_checksum Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

refactor: drive upload_file_object through SDK upload_if_changed

da3a911

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

refactor: use SDK matches_local_checksum in download_file_object

1b74a3a

Co-Authored-By: Claude Opus 4.7 <noreply@anthropic.com>

lancamat1 and others added 7 commits June 4, 2026 12:15

lancamat1 force-pushed the 001-object-file-support branch from c19c27e to 9de8e3b Compare June 4, 2026 10:24

Conversation

minitriga commented Apr 10, 2026 • edited by lancamat1 Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Initial support for file upload

Summary

Changes

Usage

Test plan

Uh oh!

cloudflare-workers-and-pages Bot commented Apr 10, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Deploying nornir-infrahub with Cloudflare Pages

Uh oh!

wvandeun commented Apr 19, 2026

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

wvandeun commented Apr 19, 2026

Uh oh!

wvandeun Apr 19, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

minitriga commented Apr 20, 2026

Uh oh!

cubic-dev-ai Bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

wvandeun commented Apr 22, 2026

Uh oh!

cubic-dev-ai Bot left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

Uh oh!

lancamat1 commented May 27, 2026

Testing status

Uh oh!

minitriga left a comment

Choose a reason for hiding this comment

Code Review

Overview

Verified correctness ✅

Issues

1. Relationships in data are silently dropped on update (medium)

2. Nonexistent explicit object_id falls through to create (medium)

3. PR description is stale (low)

4. Global ty rule disables are too blunt (low)

5. Minor nits

Code quality & conventions 👍

Verdict

Uh oh!

lancamat1 commented Jun 4, 2026

Integration tests landed

minitriga commented Apr 10, 2026 •

edited by lancamat1

Loading

cloudflare-workers-and-pages Bot commented Apr 10, 2026 •

edited

Loading

wvandeun Apr 19, 2026 •

edited

Loading

cubic-dev-ai Bot left a comment •

edited

Loading

cubic-dev-ai Bot left a comment •

edited

Loading

1. Relationships in `data` are silently dropped on update (medium)

2. Nonexistent explicit `object_id` falls through to create (medium)