Release 0.78.0

.github/instructions/frontend-tests.instructions.md

@@ -183,6 +183,18 @@ expect(consoleError).toHaveBeenCalled()
 // In rare cases, TestingErrorBoundary can be used to test thrown errors.
 ```
 
+**Mutation failures and the global error toast:**
+
+Every mutation failure raises a global error toast unless the call site opts out (`meta: SILENCE_ERROR_TOAST`), and an `afterEach` fails any test that leaves a toast unacknowledged.
+
+```tsx
+import { expectErrorToast } from "@/test-utils"
+// After driving a mutation failure whose intended error surface is the toast:
+await expectErrorToast("Something went wrong")
+// If the component shows its own inline error instead, don't acknowledge —
+// opt the component out of the toast with `meta: SILENCE_ERROR_TOAST`.
+```
+
 ## Troubleshooting
 
 - **"No response specified"** → Mock the API call with `setMockResponse`

.github/instructions/frontend.instructions.md

@@ -8,3 +8,11 @@ applyTo: "**/*.ts,**/*.tsx,**/package.json"
   - `api` contains generated API client code and react-query hooks
 - For reusable UI, use components from `@mitodl/smoot-design`, `ol-components` preferentially
 - Within `main`, use `@/` for root-relative imports
+
+## Mutation error handling
+
+Every mutation failure in the browser shows a global error toast by default (`MutationCache.onError` in `main/src/app/getQueryClient.ts`), so failures are never silent. Tune it per call site via React Query `meta` (typed in `api/mutation-meta`):
+
+- Component renders its own inline error for the failure → pass `meta: SILENCE_ERROR_TOAST` to the mutation hook, or the user sees a double alert. Required even if you catch the `mutateAsync` rejection yourself — catching does not suppress the toast.
+- Custom toast copy → `meta: { errorMessage: "Could not save your changes." }`, or `getErrorMessage(error, variables)` for data-driven copy.
+- Shared `api/` hooks accept `{ meta }` (`MutationHookOptions`) and forward it; the opt-out belongs at the _consumer_, never baked into a shared hook (many hooks are surfaced in one place and silent in another).

.github/workflows/ci.yml

@@ -27,7 +27,7 @@ jobs:
           - 5432:5432
 
       redis:
-        image: redis:8.2.2
+        image: redis:8.10.0@sha256:344e3945a0b431c8ff1eecd58c5573538126bd756f02fc7e218ddf1fc2546366
         ports:
           - 6379:6379
 
@@ -272,3 +272,62 @@ jobs:
         run: |
           diff $GENERATOR_OUTPUT_DIR_CI $GENERATOR_OUTPUT_DIR_VC \
           || { echo "OpenAPI spec is out of date. Please regenerate via ./scripts/generate_openapi.sh"; exit 1; }
+
+  # ONE required status check standing for this whole workflow.
+  #
+  # GitHub's required status checks take an exact list of context names -- no
+  # wildcards, no "all checks must pass" option -- so requiring these six jobs
+  # directly means restating all six names in mitodl/ol-infrastructure, where the
+  # ruleset lives. That list then goes stale in two ways:
+  #
+  #   A renamed job keeps being required under its old name, which GitHub will never
+  #   report again, so every PR waits forever on it. mitxonline hit exactly this when
+  #   `python-tests` became a 4-way matrix and the checks turned into
+  #   `python-tests (1)`..`(4)`.
+  #
+  #   A newly added job is not required until somebody remembers to go and add it, so
+  #   new CI silently cannot block a merge.
+  #
+  # Requiring `ci-gate` instead keeps the list of what must pass in the same file as
+  # the jobs it names, so adding or renaming a job is one edit in one repo.
+  #
+  # `openapi-diff` is absent from `needs` because `needs` can only reference jobs in
+  # this same workflow file, and `openapi-diff` runs in its own workflow (triggered on
+  # `pull_request` rather than `push`). It now has an `--err-ignore` allowlist
+  # (openapi/oasdiff-err-ignore.txt) for intentional breaking changes, so it
+  # should be required directly, alongside `ci-gate`, in the ol-infrastructure ruleset
+  # -- its job name is stable and isn't subject to the renaming/matrix drift `ci-gate`
+  # exists to solve.
+  ci-gate:
+    needs:
+      - python-tests
+      - javascript-tests
+      - build-nextjs-container
+      - build-storybook
+      - openapi-generated-client-check-v0
+      - openapi-generated-client-check-v1
+    # Without this the gate is skipped when a dependency fails, and a skipped required
+    # check leaves the PR pending rather than failing it.
+    if: always()
+    runs-on: ubuntu-24.04
+    permissions: {}
+    steps:
+      - name: Evaluate CI result
+        env:
+          NEEDS: ${{ toJSON(needs) }}
+        run: |
+          set -euo pipefail
+          echo "$NEEDS" | jq -r 'to_entries[] | "\(.key): \(.value.result)"'
+          # Read from `needs` rather than naming each job again: adding a job to the
+          # list above is then the only edit, which is the entire point of this job.
+          # `skipped` passes so that a job legitimately gated behind an `if:` does not
+          # wedge the branch; `failure` and `cancelled` do not.
+          failed=$(echo "$NEEDS" | jq -r '
+            to_entries[]
+            | select(.value.result != "success" and .value.result != "skipped")
+            | .key')
+          if [ -n "$failed" ]; then
+            echo "::error::CI did not succeed: $(echo "$failed" | paste -sd, -)"
+            exit 1
+          fi
+          echo "All CI jobs succeeded."

.github/workflows/openapi-diff.yml

@@ -67,6 +67,7 @@ jobs:
             --volume ${{ github.workspace }}:${{ github.workspace }}:ro \
             -e GITHUB_WORKSPACE=${{ github.workspace }} \
             tufin/oasdiff breaking \
+            --err-ignore head/openapi/oasdiff-err-ignore.txt \
             --fail-on ERR \
             --format githubactions \
             --composed --flatten-allof \

RELEASE.rst

@@ -1,6 +1,23 @@
 Release Notes
 =============
 
+Version 0.78.0
+--------------
+
+- Remove remaining API endpoint N+1 queries (#3845)
+- Update dependency drf-spectacular to >=0.30,<0.31 (#3858)
+- Update dependency ruff to v0.16.3 (#3859)
+- Submit and link each resource under one URL, from learn_url (#3843)
+- Update dependency litellm to v1.96.2 (#3857)
+- Update redis Docker tag to v8.10.0 (#2706)
+- feat(cohort-1): MicroMasters/MITx Online certificate warehouse-pull sync (#3808)
+- ci: add a ci-gate job so one required check can cover the whole suite (#3825)
+- Overridable Mutation Error Toast (#3837)
+- Show Stay Updated based only on the CMS page flag (#3841)
+- staleness penalty for vector search results (#3834)
+- Update dependency Django to v5.2.17 [SECURITY] (#3849)
+- Update dependency social-auth-app-django to v5.6.0 [SECURITY] (#3850)
+
 Version 0.77.15 (Released August 27, 2026)
 ---------------
 

channels/models.py

@@ -69,10 +69,21 @@ def with_detail_relations(self) -> "ChannelQuerySet":
                     "sub_channels__channel",
                     queryset=Channel.objects.annotate_channel_url(),
                 ),
+                # LearningResourceOfferor.channel_url reads
+                # channel_unit_details.first(); ordering the prefetch by pk
+                # lets that resolve from the cache and pick the same row it
+                # would have queried for.
+                Prefetch(
+                    "unit_detail__unit__channel_unit_details",
+                    queryset=ChannelUnitDetail.objects.select_related(
+                        "channel"
+                    ).order_by("pk"),
+                ),
             )
             .annotate_channel_url()
             .select_related(
                 "featured_list",
+                "unit_detail__unit",
                 "topic_detail",
                 "department_detail",
                 "unit_detail",

channels/views_test.py

@@ -7,7 +7,12 @@
 from django.urls import reverse
 
 from channels.constants import ChannelType
-from channels.factories import ChannelFactory, ChannelListFactory, SubChannelFactory
+from channels.factories import (
+    ChannelFactory,
+    ChannelListFactory,
+    ChannelUnitDetailFactory,
+    SubChannelFactory,
+)
 from channels.models import Channel
 from channels.serializers import ChannelSerializer
 from learning_resources.factories import LearningResourceFactory
@@ -16,7 +21,6 @@
 pytestmark = pytest.mark.django_db
 
 
-@pytest.mark.skip_nplusone_check
 def test_list_channels(user_client):
     """Test that all published channels are returned."""
     ChannelFactory.create_batch(2, published=False)
@@ -33,6 +37,44 @@ def test_list_channels(user_client):
         assert response_channels[idx] == ChannelSerializer(instance=channel).data
 
 
+@pytest.mark.parametrize("channel_count", [2, 6])
+def test_list_unit_channels_query_count(
+    client, django_assert_num_queries, channel_count
+):
+    """Unit channels cost the same number of queries however many are listed.
+
+    unit_detail.unit is nested by the serializer and its channel_url is a
+    cached_property, so without the prefetch each row costs two extra queries.
+    """
+    channels = ChannelFactory.create_batch(channel_count, is_unit=True)
+    # An offeror shared with later, unpublished channels: channel_url has to
+    # resolve to the same row the serializer would pick on its own, so the
+    # prefetch behind it can't be an unordered subquery.
+    for _ in range(3):
+        extra = ChannelFactory.create(
+            is_unit=True, published=False, create_unit_detail=False
+        )
+        ChannelUnitDetailFactory.create(
+            channel=extra, unit=channels[0].unit_detail.unit
+        )
+
+    url = reverse("channels:v0:channels_api-list")
+    with django_assert_num_queries(5):
+        results = client.get(url).json()["results"]
+
+    assert len(results) == channel_count
+    assert all(item["unit_detail"]["unit"]["channel_url"] for item in results)
+    # The listing must agree with serializing each instance directly.
+    for channel in channels:
+        listed = next(item for item in results if item["id"] == channel.id)
+        assert (
+            listed["unit_detail"]["unit"]["channel_url"]
+            == ChannelSerializer(instance=channel).data["unit_detail"]["unit"][
+                "channel_url"
+            ]
+        )
+
+
 def test_channel_detail_has_no_is_moderator(client):
     """Channel detail no longer exposes moderator-specific fields."""
     channel = ChannelFactory.create()
@@ -153,7 +195,6 @@ def test_no_excess_list_queries(client, user, django_assert_num_queries, channel
                 assert channel["channel_url"] is not None
 
 
-@pytest.mark.skip_nplusone_check
 def test_channel_counts_view(client):
     """Channel counts should return per-channel resource counts."""
     url = reverse(
@@ -194,7 +235,6 @@ def enabled_view_cache(settings, request):
     }
 
 
-@pytest.mark.skip_nplusone_check
 @pytest.mark.usefixtures("enabled_view_cache")
 def test_channel_detail_cache_is_global(client):
     """Cached detail responses are shared globally across users."""
@@ -210,7 +250,6 @@ def test_channel_detail_cache_is_global(client):
     assert client.get(url).json()["title"] == "Original title"
 
 
-@pytest.mark.skip_nplusone_check
 @pytest.mark.usefixtures("enabled_view_cache")
 def test_channel_by_type_name_cache_is_global(client):
     """Cached by-type-name responses are shared globally across users."""
@@ -229,7 +268,6 @@ def test_channel_by_type_name_cache_is_global(client):
     assert client.get(url).json()["title"] == "Original title"
 
 
-@pytest.mark.skip_nplusone_check
 @pytest.mark.usefixtures("enabled_view_cache")
 @pytest.mark.parametrize("is_authenticated", [False, True])
 def test_channel_counts_view_is_cached(client, is_authenticated):

docker-compose.services.yml

@@ -31,7 +31,7 @@ services:
   redis:
     profiles:
       - backend
-    image: redis:8.2.2
+    image: redis:8.10.0@sha256:344e3945a0b431c8ff1eecd58c5573538126bd756f02fc7e218ddf1fc2546366
     healthcheck:
       test: ["CMD", "redis-cli", "ping", "|", "grep", "PONG"]
       interval: 3s

drf_lint_baseline.json

@@ -2,8 +2,5 @@
   "channels/serializers.py:132:24:ORM002",
   "channels/serializers.py:134:24:ORM002",
   "channels/serializers.py:136:24:ORM002",
-  "profiles/serializers.py:136:31:ORM002",
-  "profiles/serializers.py:196:16:ORM002",
-  "profiles/serializers.py:446:15:ORM001",
-  "profiles/serializers.py:447:27:ORM001"
+  "profiles/serializers.py:205:16:ORM002"
 ]

frontends/api/package.json

@@ -16,6 +16,7 @@
     "./test-utils/mockAxios": "./src/test-utils/mockAxios.ts",
     "./test-utils": "./src/test-utils/index.ts",
     "./mitxonline-hooks/*": "./src/mitxonline/hooks/*/index.ts",
+    "./mutation-meta": "./src/mutations/mutationMeta.ts",
     "./mitxonline-test-utils": "./src/mitxonline/test-utils/index.ts",
     "./analytics-hooks/*": "./src/analytics/hooks/*/index.ts",
     "./analytics-types": "./src/analytics/types.ts",