ErikBPF opened a new pull request, #6112:
URL: https://github.com/apache/datafusion-comet/pull/6112
Closes #6102.
The TPC-H and TPC-DS dataset caches were keyed on
`hashFiles('.github/workflows/pr_build_linux.yml')`, so any workflow edit
regenerated both datasets and left entries that a later run can never
restore.
The generators were also unpinned: `GenTPCHData` cloned `tpch-dbgen` HEAD and
the `tpcds-kit` checkout had no `ref:`.
This pins both generators and keys the datasets on their inputs:
- TPC-H key is the hash of `GenTPCHData.scala`; the generator checks out a
fixed
`tpch-dbgen` commit before applying the stdout patch.
- TPC-DS key embeds the pinned `tpcds-kit` commit, which the checkout names
explicitly.
- Both caches are split into `actions/cache/restore` plus a save step
guarded to
`main`, matching the cache-write policy used for other large caches.
`check_tpc_dataset_caches()` in `dev/ci/check-ci-config.py` asserts the key
shape, the pinned `ref:`, and that the TPC caches are no longer bare
`actions/cache@vN`.
Testing: the new guard fails on the previous revision with five findings and
passes after the change (`CI config checks passed`); `dev/ci/check-suites.py`
also passes.
--
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]
