sadpandajoe commented on code in PR #45601:
URL: https://github.com/apache/superset/pull/45601#discussion_r4344255934
##########
.github/workflows/caches.js:
##########
@@ -35,6 +35,25 @@ const assetsConfig = {
};
// Multi-layer cache definition
+// A pg_dump of the metadata + examples database the Playwright jobs load with
+// `superset load_examples`. Keyed on everything that shapes that data: the
+// example loaders and their YAML configs, migrations, the CLI commands that
+// run, the CI config, and the E2E workflow itself (it pins the Postgres image
+// the dump has to match).
+const playwrightTestdataConfig = {
+ path: ['/tmp/superset-playwright-testdata'],
+ hashFiles: [
+ `${workspaceDirectory}/superset/examples/**/*`,
+ `${workspaceDirectory}/superset/migrations/**/*`,
+ `${workspaceDirectory}/superset/cli/examples.py`,
+ `${workspaceDirectory}/superset/cli/test.py`,
+ `${workspaceDirectory}/tests/integration_tests/superset_test_config.py`,
+ `${workspaceDirectory}/.github/workflows/superset-e2e.yml`,
+ `${workspaceDirectory}/.github/workflows/bashlib.sh`,
+ ],
Review Comment:
Agreed—the key also misses other code that shapes the loaded rows, e.g.
`superset/connectors/sqla/models.py` (`fetch_metadata` fills
`table_columns.is_dttm`) and the Python dependency pins. A PR changing that
logic restores the old rows, and `db upgrade`/`init` won't regenerate them, so
Playwright runs against stale fixture data. Should the hash cover those paths,
or should the dump be rebuilt on every `superset/**` change?
--
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]
