Fix array types coercion: preserve child element nullability for list types #17306
Conversation
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
sgrebnov
commented
Aug 23, 2025
sgrebnov
force-pushed
the
sgrebnov/upstream-fix-type-coercion
branch
from
8b478f8 to
bf70047
Compare
sgrebnov
added a commit
to spiceai/datafusion
that referenced
this pull request
Aug 25, 2025
… types (#96) UPSTREAM NOTE: This was submitted upstream and should be available in DF50 apache#17306
Member
Author
|
@alamb, @jayzhan211 - Could you please take a look at this change and see if we can merge it? Without this change, we encounter the |
Member
Author
Jeadie
pushed a commit
to spiceai/datafusion
that referenced
this pull request
Sep 9, 2025
… types (#96) UPSTREAM NOTE: This was submitted upstream and should be available in DF50 apache#17306
Jeadie
pushed a commit
to spiceai/datafusion
that referenced
this pull request
Sep 12, 2025
… types (#96) UPSTREAM NOTE: This was submitted upstream and should be available in DF50 apache#17306
peasee
pushed a commit
to spiceai/datafusion
that referenced
this pull request
Oct 27, 2025
… types (#96) UPSTREAM NOTE: This was submitted upstream and should be available in DF50 apache#17306
peasee
added a commit
to spiceai/datafusion
that referenced
this pull request
Oct 27, 2025
* fix: Ensure only tables or aliases that exist are projected (#52) fix: More dangling references (#54) UPSTREAM NOTE: This PR was attempted to be upstreamed in apache#13405 - but it was not accepted due to the complexity it brought. Phillip needs to figure out what a good solution that solves our problem and can be upstreamed is. * Support for metadata columns (`location`, `size`, `last_modified`) in ListingTableProvider (#74) UPSTREAM NOTE: This PR was attempted to be upstreamed but was not accepted. Needs to be applied manually apache#15181 * Infer placeholder datatype for `Expr::InSubquery` (#80) UPSTREAM NOTE: Upstream PR has been created but not merged yet. Should be available in DF49 apache#15980 * Infer placeholder datatype after `LIMIT` clause as `DataType::Int64` (#81) UPSTREAM NOTE: Upstream PR has been created but not merged yet. Should be available in DF49 apache#15980 * Do not double alias Exprs UPSTREAM NOTE: This was attempted to be fixed with apache#15008 but was closed This is the tracking issue on DataFusion: apache#14895 Do not double alias Exprs * Add prefix to location metadata column (#82) UPSTREAM NOTE: This will not be upstreamed as is. * Infer placeholder types for CASE expressions (#87) UPSTREAM NOTE: This has not been submitted upstream yet. * Expand `infer_placeholder_types` to infer all possible placeholder types based on their expression (#88) UPSTREAM NOTE: This has not been submitted upstream yet. * Fix `Expr::infer_placeholder_types` inference to not fail (#89) UPSTREAM NOTE: This has not been submitted upstream yet. * cherry-pick parquet patch (#94) * Fix array types coercion: preserve child element nullability for list types (#96) UPSTREAM NOTE: This was submitted upstream and should be available in DF50 apache#17306 * Expand `infer_placeholder_types` to infer all possible placeholder types based on their expression (#88) UPSTREAM NOTE: This has not been submitted upstream yet. * do not enforce type guarantees on all Expr traversed in infer_placeholder_types (#97) * Use UDTF function args in `LogicalPlan::TableScan` name (#98) * use UDTF function args in LogicalPlan::TableScan name * update test snapshots * Implement timestamp_cast_dtype for SqliteDialect (#99) * Use text for sqlite timestamp * Add test * Custom timestamp format for DuckDB (#102) * Revert "cherry-pick parquet patch (#94)" This reverts commit d780cc2. * Support ExprNamed arguments to Scalar UDFs (#104) * support ExprNamed until 17379 ships * add same exprnamed lifting to udtf * resolve projection against `ListingTable` table_schema incl. partition columns (#106) * fix: Ensure ListingTable partitions are pruned when filters are not used (#108) * fix: Prune partitions when no filters are defined * fix: Backport for DF49: * review: Address comments * FileScanConfig: Preserve schema metadata across serde boundary (#107) * FileScanConfig: preserve schema metadata across serde boundary * add test * Merge conflict fixes UPSTREAM NOTE: this should not be upstreamed. This contains conflict fixes from various cherry-picks and differences in v50. * update arrow-rs fork UPSTREAM NOTE: this should not be upstreamed --------- Co-authored-by: Phillip LeBlanc <phillip@leblanc.tech> Co-authored-by: Kevin Zimmerman <4733573+kczimm@users.noreply.github.com> Co-authored-by: sgrebnov <sergei.grebnov@gmail.com> Co-authored-by: jeadie <jack@spice.ai> Co-authored-by: Jack Eadie <jack.eadie0@gmail.com> Co-authored-by: Viktor Yershov <krinart@gmail.com> Co-authored-by: Viktor Yershov <viktor@spice.ai> Co-authored-by: David Stancu <david@spice.ai>
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
3 participants
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Which issue does this PR close?
Fixes #17305
What changes are included in this PR?
PR updates
array_valid_typesimplementation to preserve nullability information for nested fields. It follows up on the change/improvement in the linked item to restore the previous behavior:#15149 (comment)
Are these changes tested?
Yes, the existing unit test was expanded to cover nullability.
Are there any user-facing changes?
No. This only prevents schema mismatch errors during execution, as described in the linked bug report.