Weixin-Xu opened a new pull request, #13379:
URL: https://github.com/apache/gluten/pull/13379
## What changes are proposed in this pull request?
This PR fixes scan metadata handling and table-format integration in the
Bolt backend.
- Preserve Substrait row-index column types and exclude partition columns
from the physical file schema, while retaining scan outputs and column
assignments.
- Register Delta change data feed expansion during optimization,
controlled by a Bolt-specific session setting.
- Fall back to Spark when a Delta scan reads files with actual deletion
vectors. Keep DV-free scans eligible for native execution, including scans
whose partition filters exclude files containing deletion
vectors.
- Honor `spark.gluten.paimon.native.source.enabled` when the Bolt Paimon
offload rule executes, preserving native offload by default.
- Add the missing Bolt TPC-H suite base and correct classpath resource
references in the Delta and Hudi suites, following the existing Velox default
test behavior.
Add focused regressions for scan metadata, Delta CDF configuration,
deletion-vector fallback, and Paimon source configuration.
Changes are scoped to Bolt implementation and tests.
## How was this patch tested?
These changes were validated as part of the integrated Bolt CI candidate
using Bolt revision `071e54764f887dfb4337260e440f2f12f6ab82f5` with the
recorded engine fixes applied.
Regression coverage includes:
- Row-index column mapping, invalid column-count rejection, and
partition-column exclusion from the physical file schema.
- Delta CDF optimizer registration, output identity, and session
configuration changes.
- A Delta DELETE that creates an actual deletion vector, followed by exact
result checks and verification of Spark fallback. DV-free and partition-pruned
scans remain native.
- Paimon source configuration defaults, explicit disabling, and changes
across SQLConf contexts.
Integrated validation included the affected complete Spark reference cases
and an additional Spark 4.1 Delta reactor run covering four DELETE suites.
After rebasing this topic onto Apache Gluten main, patch-equivalence
checks passed. All 13 changed files retained their original contents and modes,
and `git diff --check` passed.
The standalone rebased branch has not been rebuilt or runtime-tested
independently. Native Delta deletion-vector reading remains unsupported; this
PR preserves correctness through Spark fallback.
## Was this patch authored or co-authored using generative AI tooling?
Generated-by: Codex (GPT-6)
--
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]
