ci(frontend): track bundle size over time with benchmark-action

Superset's frontend bundle size is a recurring complaint, but until now
there's been no ongoing visibility into it -- only periodic manual cleanup
efforts. This adds a `bundle-size` job to superset-frontend.yml that builds
the real production bundle (npm run build), reduces its per-entrypoint
sizes to a few headline numbers, and tracks them over time via
benchmark-action/github-action-benchmark, posting a PR comment when a
change regresses past 110% of the last recorded baseline.

Reuses the already-built CI Docker image (same image sharded-jest-tests /
lint-frontend / etc. already download), so the only new cost is the
production webpack build itself -- and that's cheap after the first run:
webpack's persistent filesystem cache (already configured in
webpack.config.js) makes warm rebuilds ~20s locally vs several minutes
cold, confirmed by hand against this repo's real build.

webpack.config.js gains a BUNDLE_SIZE_STATS env-gated stats override
(mirrors the existing BUNDLE_ANALYZER pattern). The default `stats:
'minimal'` omits per-asset sizes entirely; `--stats=normal` includes them
but also serializes the full ~15k-module dependency graph, producing a
560+MB stats.json for this app -- large enough to exceed Node's max
string length on a plain fs.readFileSync. The env-gated override requests
just `{ assets: true, entrypoints: true }`, verified end-to-end against a
minimal synthetic webpack project (same webpack/webpack-cli versions) and
against real stats pulled from this repo's actual production build.

History storage deliberately avoids the gh-pages branch (benchmark-action's
usual default) since that branch is the live Helm chart index published by
superset-helm-release.yml, not free real estate. Instead uses
external-data-json-path with actions/cache: restored on every run (PR or
push) so PRs get a same-baseline comparison and regression comment, but
only saved back to the cache on push to master, so an unmerged PR's numbers
never become the shared baseline. No gh-pages branch is touched in any
code path, and no PAT/GitHub App is needed -- comment-on-alert only needs
the default per-job GITHUB_TOKEN.

.github/actions/setup-backend/action.yml

@@ -5,6 +5,10 @@ inputs:
     description: 'Python version to set up. Accepts a version number, "current", or "next".'
     required: true
     default: 'current'
+  cache:
+    description: 'Cache dependencies. Options: pip'
+    required: false
+    default: 'pip'
   requirements-type:
     description: 'Type of requirements to install. Options: base, development, default'
     required: false
@@ -39,24 +43,17 @@ runs:
       uses: actions/setup-python@a309ff8b426b58ec0e2a45f0f869d46889d02405 # v6.2.0
       with:
         python-version: ${{ steps.set-python-version.outputs.python-version }}
-    - name: Install uv
-      if: inputs.install-superset == 'true'
-      uses: astral-sh/setup-uv@c771a70e6277c0a99b617c7a806ffedaca235ff9 # v9.0.0
-      with:
-        python-version: ${{ steps.set-python-version.outputs.python-version }}
-        enable-cache: true
-    - name: Install apt packages
-      if: inputs.install-superset == 'true'
-      uses: awalsh128/cache-apt-pkgs-action@553a35bb8ebd9fcabcb1c9451aa4c98e1b4ca8a9 # v1.6.3
-      with:
-        packages: libldap2-dev libsasl2-dev
-        version: 1.0
+        cache: ${{ inputs.cache }}
     - name: Install dependencies
       env:
         INPUT_INSTALL_SUPERSET: ${{ inputs.install-superset }}
         INPUT_REQUIREMENTS_TYPE: ${{ inputs.requirements-type }}
       run: |
         if [ "$INPUT_INSTALL_SUPERSET" = "true" ]; then
+          sudo apt-get update && sudo apt-get -y install libldap2-dev libsasl2-dev
+
+          pip install --upgrade pip setuptools wheel uv
+
           if [ "$INPUT_REQUIREMENTS_TYPE" = "dev" ]; then
             uv pip install --system -r requirements/development.txt
           elif [ "$INPUT_REQUIREMENTS_TYPE" = "base" ]; then

.github/workflows/bump-python-package.yml

@@ -45,10 +45,7 @@ jobs:
           python-version: "3.11"
 
       - name: Install uv
-        uses: astral-sh/setup-uv@c771a70e6277c0a99b617c7a806ffedaca235ff9 # v9.0.0
-        with:
-          python-version: "3.11"
-          enable-cache: true
+        run: pip install uv
 
       - name: supersetbot bump-python -p "${{ github.event.inputs.package }}"
         env:

.github/workflows/check-python-deps.yml

@@ -11,7 +11,6 @@ on:
 permissions:
   contents: read
   pull-requests: read
-  actions: read
 
 # cancel previous workflow jobs for PRs
 concurrency:
@@ -22,10 +21,6 @@ jobs:
   check-python-deps:
     runs-on: ubuntu-26.04
     steps:
-      - uses: Kesin11/actions-timeline@7bf79990b7c09f5dfb570ac30b814ca597bd538e # v3.1.1
-        with:
-          expand-composite-actions: true
-
       - name: "Checkout ${{ github.ref }} ( ${{ github.sha }} )"
         uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
         with:

.github/workflows/dependency-review.yml

@@ -20,7 +20,6 @@ concurrency:
 
 permissions:
   contents: read
-  actions: read
 
 jobs:
   dependency-review:
@@ -51,10 +50,6 @@ jobs:
     # You cannot use a liccheck.ini file in this workflow.
     runs-on: ubuntu-26.04
     steps:
-      - uses: Kesin11/actions-timeline@7bf79990b7c09f5dfb570ac30b814ca597bd538e # v3.1.1
-        with:
-          expand-composite-actions: true
-
       - name: "Checkout Repository"
         uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
         with:

.github/workflows/docker.yml

@@ -213,14 +213,3 @@ jobs:
         shell: bash
         run: |
           docker compose -f docker-compose-image-tag.yml up superset-init --exit-code-from superset-init
-
-  actions-timeline:
-    needs: [docker-build, docker-compose-image-tag]
-    if: always()
-    runs-on: ubuntu-26.04
-    permissions:
-      actions: read
-    steps:
-      - uses: Kesin11/actions-timeline@7bf79990b7c09f5dfb570ac30b814ca597bd538e # v3.1.1
-        with:
-          expand-composite-actions: true

.github/workflows/pre-commit.yml

@@ -15,7 +15,6 @@ on:
 
 permissions:
   contents: read
-  actions: read
 
 # cancel previous workflow jobs for PRs
 concurrency:
@@ -33,10 +32,6 @@ jobs:
         # rarely differ across patch versions, so 3x per PR is wasteful.
         python-version: ${{ github.event_name == 'pull_request' && fromJSON('["current"]') || fromJSON('["current", "next"]') }}
     steps:
-      - uses: Kesin11/actions-timeline@7bf79990b7c09f5dfb570ac30b814ca597bd538e # v3.1.1
-        with:
-          expand-composite-actions: true
-
       - name: "Checkout ${{ github.ref }} ( ${{ github.sha }} )"
         uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
         with:

.github/workflows/superset-app-cli.yml

@@ -11,7 +11,6 @@ on:
 permissions:
   contents: read
   pull-requests: read
-  actions: read
 
 # cancel previous workflow jobs for PRs
 concurrency:
@@ -41,10 +40,6 @@ jobs:
         ports:
           - 16379:6379
     steps:
-      - uses: Kesin11/actions-timeline@7bf79990b7c09f5dfb570ac30b814ca597bd538e # v3.1.1
-        with:
-          expand-composite-actions: true
-
       - name: "Checkout ${{ github.ref }} ( ${{ github.sha }} )"
         uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
         with:

.github/workflows/superset-docs-deploy.yml

@@ -30,7 +30,6 @@ concurrency:
 
 permissions:
   contents: read
-  actions: read
 
 jobs:
   config:
@@ -60,10 +59,6 @@ jobs:
     name: Build & Deploy
     runs-on: ubuntu-26.04
     steps:
-      - uses: Kesin11/actions-timeline@7bf79990b7c09f5dfb570ac30b814ca597bd538e # v3.1.1
-        with:
-          expand-composite-actions: true
-
       - name: "Checkout ${{ github.event.workflow_run.head_sha || github.sha }}"
         uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
         with:
@@ -81,10 +76,7 @@ jobs:
           distribution: "zulu"
           java-version: "21"
       - name: Install Graphviz
-        uses: awalsh128/cache-apt-pkgs-action@553a35bb8ebd9fcabcb1c9451aa4c98e1b4ca8a9 # v1.6.3
-        with:
-          packages: graphviz
-          version: 1.0
+        run: sudo apt-get install -y graphviz
       - name: Compute Entity Relationship diagram (ERD)
         env:
           SUPERSET_SECRET_KEY: not-a-secret

.github/workflows/superset-e2e.yml

@@ -340,14 +340,3 @@ jobs:
             exit 1
           fi
           echo "playwright-tests result: $RESULT (changes: $CHANGES)"
-
-  actions-timeline:
-    needs: [cypress-matrix, playwright-tests]
-    if: always()
-    runs-on: ubuntu-26.04
-    permissions:
-      actions: read
-    steps:
-      - uses: Kesin11/actions-timeline@7bf79990b7c09f5dfb570ac30b814ca597bd538e # v3.1.1
-        with:
-          expand-composite-actions: true

.github/workflows/superset-extensions-cli.yml

@@ -11,7 +11,6 @@ on:
 permissions:
   contents: read
   pull-requests: read
-  actions: read
 
 # cancel previous workflow jobs for PRs
 concurrency:
@@ -31,10 +30,6 @@ jobs:
       run:
         working-directory: superset-extensions-cli
     steps:
-      - uses: Kesin11/actions-timeline@7bf79990b7c09f5dfb570ac30b814ca597bd538e # v3.1.1
-        with:
-          expand-composite-actions: true
-
       - name: "Checkout ${{ github.ref }} ( ${{ github.sha }} )"
         uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
         with:

.github/workflows/superset-frontend.yml

@@ -196,13 +196,74 @@ jobs:
           docker run --rm $TAG bash -c \
           "npm run build-storybook && npx playwright install-deps && npx playwright install chromium && npm run test-storybook:ci"
 
-  actions-timeline:
-    needs: [report-coverage, lint-frontend, validate-frontend, test-storybook]
-    if: always()
+  bundle-size:
+    needs: frontend-build
+    if: needs.frontend-build.outputs.should-run == 'true'
     runs-on: ubuntu-26.04
+    timeout-minutes: 15
     permissions:
-      actions: read
+      contents: write
+      pull-requests: write
     steps:
-      - uses: Kesin11/actions-timeline@7bf79990b7c09f5dfb570ac30b814ca597bd538e # v3.1.1
+      - name: Checkout Code
+        uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
         with:
-          expand-composite-actions: true
+          persist-credentials: false
+
+      - name: Download Docker Image Artifact
+        uses: actions/download-artifact@3e5f45b2cfb9172054b4087a40e8e0b5a5461e7c # v8
+        with:
+          name: docker-image
+
+      - name: Load Docker Image
+        run: |
+          zstd -d < docker-image.tar.zst | docker load
+
+      # Only ever pull the last recorded data point off the cache, keyed by
+      # run ID -- `restore-keys` prefix-matches the most recently created
+      # entry. Absent on the very first run ever; benchmark-action starts a
+      # fresh history in that case.
+      - name: Restore bundle size history
+        uses: actions/cache/restore@55cc8345863c7cc4c66a329aec7e433d2d1c52a9 # v6.1.0
+        with:
+          path: bundle-size-history.json
+          key: bundle-size-history-${{ github.run_id }}
+          restore-keys: |
+            bundle-size-history-
+
+      - name: Build production bundle with stats
+        run: |
+          mkdir -p ${{ github.workspace }}/superset-frontend/bundle-stats
+          docker run \
+          -v ${{ github.workspace }}/superset-frontend/bundle-stats:/app/superset-frontend/bundle-stats \
+          --rm $TAG \
+          bash -c \
+          "npm i && BUNDLE_SIZE_STATS=true npm run build -- --json=bundle-stats/stats.json"
+
+      - name: Summarize bundle size
+        run: |
+          node superset-frontend/scripts/bundle-size-summary.js \
+            superset-frontend/bundle-stats/stats.json > bundle-size-summary.json
+          rm -rf superset-frontend/bundle-stats
+
+      - name: Track bundle size
+        uses: benchmark-action/github-action-benchmark@52576c92bccf6ac60c8223ec7eb2565637cae9ba # v1.22.1
+        with:
+          tool: customSmallerIsBetter
+          output-file-path: bundle-size-summary.json
+          external-data-json-path: bundle-size-history.json
+          github-token: ${{ secrets.GITHUB_TOKEN }}
+          comment-on-alert: true
+          alert-threshold: "110%"
+          fail-on-alert: false
+          summary-always: true
+
+      # Only the trunk's own bundle size becomes the baseline everyone else
+      # is compared against -- an unmerged PR's numbers (including ones that
+      # regress on purpose to test something) never get persisted.
+      - name: Save bundle size history
+        if: github.event_name == 'push' && github.ref == 'refs/heads/master'
+        uses: actions/cache/save@55cc8345863c7cc4c66a329aec7e433d2d1c52a9 # v6.1.0
+        with:
+          path: bundle-size-history.json
+          key: bundle-size-history-${{ github.run_id }}

.github/workflows/superset-helm-lint-test.yml

@@ -8,7 +8,6 @@ on:
 
 permissions:
   contents: read
-  actions: read
 
 # Serialize runs per PR without cancelling: when a first-time contributor's
 # queued runs are approved together, cancel-in-progress lets an older run
@@ -22,10 +21,6 @@ jobs:
   lint-test:
     runs-on: ubuntu-26.04
     steps:
-      - uses: Kesin11/actions-timeline@7bf79990b7c09f5dfb570ac30b814ca597bd538e # v3.1.1
-        with:
-          expand-composite-actions: true
-
       - name: "Checkout ${{ github.ref }} ( ${{ github.sha }} )"
         uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 # v7.0.1
         with:

.github/workflows/superset-playwright.yml

@@ -54,7 +54,6 @@ jobs:
     permissions:
       contents: read
       pull-requests: read
-      actions: read
     strategy:
       fail-fast: false
       matrix:
@@ -80,10 +79,6 @@ jobs:
         ports:
           - 16379:6379
     steps:
-      - uses: Kesin11/actions-timeline@7bf79990b7c09f5dfb570ac30b814ca597bd538e # v3.1.1
-        with:
-          expand-composite-actions: true
-
       # -------------------------------------------------------
       # Conditional checkout based on context (same as Cypress workflow)
       - name: Checkout for push or pull_request event

.github/workflows/superset-python-integrationtest.yml

@@ -255,14 +255,3 @@ jobs:
             exit 1
           fi
           echo "test-postgres result: $RESULT"
-
-  actions-timeline:
-    needs: [test-mysql, test-postgres, test-sqlite]
-    if: always()
-    runs-on: ubuntu-26.04
-    permissions:
-      actions: read
-    steps:
-      - uses: Kesin11/actions-timeline@7bf79990b7c09f5dfb570ac30b814ca597bd538e # v3.1.1
-        with:
-          expand-composite-actions: true

.github/workflows/superset-python-presto-hive.yml

@@ -158,14 +158,3 @@ jobs:
           verbose: true
           use_oidc: true
           slug: apache/superset
-
-  actions-timeline:
-    needs: [test-postgres-presto, test-postgres-hive]
-    if: always()
-    runs-on: ubuntu-26.04
-    permissions:
-      actions: read
-    steps:
-      - uses: Kesin11/actions-timeline@7bf79990b7c09f5dfb570ac30b814ca597bd538e # v3.1.1
-        with:
-          expand-composite-actions: true

.github/workflows/superset-python-unittest.yml

@@ -101,14 +101,7 @@ jobs:
     if: always()
     runs-on: ubuntu-26.04
     timeout-minutes: 5
-    permissions:
-      contents: read
-      actions: read
     steps:
-      - uses: Kesin11/actions-timeline@7bf79990b7c09f5dfb570ac30b814ca597bd538e # v3.1.1
-        with:
-          expand-composite-actions: true
-
       - name: Check unit-tests result
         env:
           RESULT: ${{ needs.unit-tests.result }}

.github/workflows/superset-translations.yml

@@ -78,10 +78,7 @@ jobs:
 
       - name: Install gettext tools
         if: steps.check.outputs.python == 'true' || steps.check.outputs.frontend == 'true'
-        uses: awalsh128/cache-apt-pkgs-action@553a35bb8ebd9fcabcb1c9451aa4c98e1b4ca8a9 # v1.6.3
-        with:
-          packages: gettext
-          version: 1.0
+        run: sudo apt-get update && sudo apt-get install -y gettext
 
       # Fetch the base ref so we can compare PR-introduced regressions
       # against a fair baseline (also runs babel_update against the base
@@ -156,14 +153,3 @@ jobs:
       - name: Fail if regression detected
         if: steps.regression.outcome == 'failure'
         run: exit 1
-
-  actions-timeline:
-    needs: [frontend-check-translations, babel-extract]
-    if: always()
-    runs-on: ubuntu-26.04
-    permissions:
-      actions: read
-    steps:
-      - uses: Kesin11/actions-timeline@7bf79990b7c09f5dfb570ac30b814ca597bd538e # v3.1.1
-        with:
-          expand-composite-actions: true

superset-frontend/scripts/bundle-size-summary.js

@@ -0,0 +1,78 @@
+#!/usr/bin/env node
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements.  See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership.  The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License.  You may obtain a copy of the License at
+ *
+ *   http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express
+ * or implied.  See the License for the specific language governing
+ * permissions and limitations under the License.
+ */
+
+// Reduces a webpack `--json` stats file down to the handful of headline
+// numbers worth tracking over time, in the flat array format
+// benchmark-action/github-action-benchmark expects for its
+// "customSmallerIsBetter" tool. The full stats file also includes a
+// `modules`/`chunks` graph across ~15k modules, which is enormous and not
+// useful for this purpose, so we only ever read `entrypoints`.
+//
+// Usage: node scripts/bundle-size-summary.js <path-to-stats.json>
+
+const fs = require('fs');
+
+// Entrypoints worth tracking: the two user-facing app shells. `menu`,
+// `preamble`, `theme`, and `service-worker` are small, low-variance
+// infrastructure chunks, not where bundle bloat actually shows up.
+const TRACKED_ENTRYPOINTS = ['spa', 'embedded'];
+
+function entrypointSizeByExt(entrypoint, ext) {
+  return (entrypoint.assets || [])
+    .filter(asset => asset.name.endsWith(ext))
+    .reduce((total, asset) => total + asset.size, 0);
+}
+
+function main() {
+  const statsPath = process.argv[2];
+  if (!statsPath) {
+    console.error('Usage: bundle-size-summary.js <path-to-stats.json>');
+    process.exit(1);
+  }
+
+  const stats = JSON.parse(fs.readFileSync(statsPath, 'utf8'));
+  const { entrypoints } = stats;
+  if (!entrypoints) {
+    console.error(
+      'stats.json has no `entrypoints` key -- was it generated with ' +
+        '`--stats=normal` (or richer)? `minimal`/`errors-only` stats omit it.',
+    );
+    process.exit(1);
+  }
+
+  const results = [];
+  TRACKED_ENTRYPOINTS.forEach(name => {
+    const entrypoint = entrypoints[name];
+    if (!entrypoint) return;
+    results.push({
+      name: `${name} entrypoint (JS)`,
+      unit: 'bytes',
+      value: entrypointSizeByExt(entrypoint, '.js'),
+    });
+    results.push({
+      name: `${name} entrypoint (CSS)`,
+      unit: 'bytes',
+      value: entrypointSizeByExt(entrypoint, '.css'),
+    });
+  });
+
+  console.log(JSON.stringify(results, null, 2));
+}
+
+main();

superset-frontend/webpack.config.js

@@ -738,4 +738,14 @@ const smp = new SpeedMeasurePlugin({
   disable: !measure,
 });
 
+// Emits per-asset/entrypoint sizes via `--json` (the default `stats: 'minimal'`
+// above omits both). Not `normal`/`detailed` stats: those also serialize the
+// full ~15k-module dependency graph, which is hundreds of MB for this app --
+// large enough to exceed Node's max string length when read back with
+// `fs.readFileSync`. Used by scripts/bundle-size-summary.js in CI.
+// e.g. BUNDLE_SIZE_STATS=true npm run build -- --json=stats.json
+if (process.env.BUNDLE_SIZE_STATS) {
+  config.stats = { all: false, assets: true, entrypoints: true };
+}
+
 module.exports = smp.wrap(config);

superset/daos/dashboard.py

@@ -397,7 +397,8 @@ class DashboardDAO(BaseDAO[Dashboard]):
             md["color_namespace"] = data.get("color_namespace")
 
         md["expanded_slices"] = data.get("expanded_slices", {})
-        md["refresh_frequency"] = data.get("refresh_frequency", 0)
+        if "refresh_frequency" in data:
+            md["refresh_frequency"] = data["refresh_frequency"]
         md["color_scheme"] = data.get("color_scheme", "")
         md["label_colors"] = data.get("label_colors", {})
         md["shared_label_colors"] = data.get("shared_label_colors", [])

tests/unit_tests/dao/dashboard_test.py

@@ -24,6 +24,7 @@ from superset.connectors.sqla.models import Database, SqlaTable
 from superset.daos.dashboard import DashboardDAO
 from superset.models.dashboard import Dashboard
 from superset.models.slice import Slice
+from superset.utils import json
 from tests.unit_tests.conftest import with_feature_flags
 
 
@@ -117,3 +118,53 @@ def test_set_dash_metadata_preserves_soft_deleted_members(
     )
     # And the position slot kept its UUID rather than being nulled.
     assert positions["CHART-trashed"]["meta"]["uuid"] == str(trashed_chart.uuid)
+
+
+def test_set_dash_metadata_preserves_refresh_frequency(session: Session) -> None:
+    """set_dash_metadata must not reset refresh_frequency when absent from data.
+
+    Regression test for #42116: ``data.get("refresh_frequency", 0)`` would
+    unconditionally overwrite the existing value with 0 whenever the caller
+    did not include ``refresh_frequency`` in the data dict.
+    """
+    Dashboard.metadata.create_all(session.get_bind())
+
+    dashboard = Dashboard(
+        dashboard_title="refresh_test_dash",
+        json_metadata=json.dumps({"refresh_frequency": 30}),
+    )
+    db.session.add(dashboard)
+    db.session.flush()
+
+    # Simulate a save that does NOT include refresh_frequency
+    # (e.g. changing only the title via the PropertiesModal).
+    DashboardDAO.set_dash_metadata(dashboard, {"color_scheme": "superset"})
+
+    md = json.loads(dashboard.json_metadata)
+    assert md["refresh_frequency"] == 30, (
+        "refresh_frequency should be preserved when not present in data"
+    )
+
+
+def test_set_dash_metadata_updates_refresh_frequency_when_present(
+    session: Session,
+) -> None:
+    """set_dash_metadata must update refresh_frequency when it IS in data."""
+    Dashboard.metadata.create_all(session.get_bind())
+
+    dashboard = Dashboard(
+        dashboard_title="refresh_test_dash_2",
+        json_metadata=json.dumps({"refresh_frequency": 30}),
+    )
+    db.session.add(dashboard)
+    db.session.flush()
+
+    # Simulate a save that explicitly sets refresh_frequency to 0.
+    DashboardDAO.set_dash_metadata(
+        dashboard, {"refresh_frequency": 0, "color_scheme": "superset"}
+    )
+
+    md = json.loads(dashboard.json_metadata)
+    assert md["refresh_frequency"] == 0, (
+        "refresh_frequency should be updated when present in data"
+    )

tests/unit_tests/result_set_test.py

@@ -571,3 +571,53 @@ def test_stringify_values_non_serializable_dict_falls_back_to_str() -> None:
     # Must not raise — falls back to str()
     result = stringify_values(data)
     assert result[0] == str({"key": _Unserializable()})
+
+
+def test_empty_result_set_preserves_column_metadata() -> None:
+    """
+    Test that column metadata is preserved when query returns zero rows.
+
+    When a query returns no data but has a valid cursor description, the
+    column names and types from cursor_description should be preserved
+    in the result set. This allows downstream consumers (like the UI)
+    to display column headers even for empty result sets.
+    """
+    data: DbapiResult = []
+    description = [
+        ("id", "int", None, None, None, None, True),
+        ("name", "varchar", None, None, None, None, True),
+        ("created_at", "timestamp", None, None, None, None, True),
+    ]
+
+    result_set = SupersetResultSet(
+        data,
+        description,  # type: ignore
+        BaseEngineSpec,
+    )
+
+    # Verify column count
+    assert len(result_set.columns) == 3
+
+    # Verify column names are preserved
+    column_names = [col["column_name"] for col in result_set.columns]
+    assert column_names == ["id", "name", "created_at"]
+
+    assert result_set.columns[0]["type"] == BaseEngineSpec.get_datatype(
+        description[0][1]
+    )
+    assert result_set.columns[1]["type"] == BaseEngineSpec.get_datatype(
+        description[1][1]
+    )
+    assert result_set.columns[2]["type"] == BaseEngineSpec.get_datatype(
+        description[2][1]
+    )
+
+    # Verify the PyArrow table has the correct schema
+    assert result_set.table.num_rows == 0
+    assert len(result_set.table.column_names) == 3
+    assert list(result_set.table.column_names) == ["id", "name", "created_at"]
+
+    # Verify DataFrame conversion works
+    df = result_set.to_pandas_df()
+    assert len(df) == 0
+    assert list(map(str, df.columns)) == ["id", "name", "created_at"]