sadpandajoe commented on code in PR #45588:
URL: https://github.com/apache/superset/pull/45588#discussion_r4369961306
##########
.github/workflows/update-country-maps.yml:
##########
@@ -0,0 +1,184 @@
+name: Update Country Maps
+
+# Regenerates plugins/plugin-chart-country-map's GeoJSON files, countries.ts
+# and docs/data/countries.json from upstream Natural Earth + community geo
+# data, by running the same notebook `npm run update-maps` runs locally.
+#
+# Runs weekly and on workflow_dispatch, but only opens or updates a PR when
+# the regenerated output actually differs from what's on `master` - most
+# weeks this is a no-op. Unlike a translation-string catch-up, nothing here
+# auto-merges: a border/region change is the kind of thing that needs an
+# actual human look at the rendered maps. Community geo sources are pinned
+# to a commit where practical; the live WHO ArcGIS query can't be pinned
+# (it's a live service, not a repo), so a run can still legitimately pick up
+# an unreviewed upstream change there. The job itself fails (separately from
+# whether a PR was opened) if the notebook hits a cell error, so a broken
+# upstream source doesn't just quietly go green.
+on:
+ schedule:
+ - cron: "0 7 * * 1" # Mondays 07:00 UTC, an hour after
sync-translation-catalogs.yml
+ workflow_dispatch: {}
+
+permissions:
+ contents: read
+
+concurrency:
+ group: update-country-maps
+ cancel-in-progress: false
+
+jobs:
+ update-country-maps:
+ if: github.event_name != 'schedule' || github.repository ==
'apache/superset'
+ runs-on: ubuntu-26.04
+ permissions:
+ contents: write
+ pull-requests: write
+ steps:
+ - name: "Checkout master"
+ uses: actions/checkout@3d3c42e5aac5ba805825da76410c181273ba90b1 #
v7.0.1
+ with:
+ ref: master
+ persist-credentials: true
+
+ - name: Setup Python
+ uses: actions/setup-python@5fda3b95a4ea91299a34e894583c3862153e4b97 #
v7.0.0
+ with:
+ python-version: "3.11"
+
+ - name: Install notebook dependencies
+ run: |
+ pip install jupyter nbconvert ipykernel geopandas shapely matplotlib
requests
Review Comment:
The pins cover the top-level packages, but `pyogrio`, `pandas`, `numpy` and
`pyproj` still resolve to whatever is newest on each run. `geopandas` 1.x
writes the GeoJSON through `pyogrio`, whose wheel bundles GDAL, so a release
there can change coordinate precision or serialization and open a map-change PR
with no upstream data change. Could the full resolved set be pinned, e.g. a
constraints file passed with `-c`, so the diff only moves with the data?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
