GH-39562: [C++][Parquet] Fix crash in test_parquet_dataset_lazy_filtering #39632
Conversation
TODO complete this description
|
|
|
@github-actions crossbow submit wheel-macos* |
|
|
|
||
| metadata_ = std::move(metadata); | ||
| manifest_ = std::move(manifest); | ||
| original_metadata_ = original_metadata ? std::move(original_metadata) : metadata_; |
There was a problem hiding this comment.
This is ok for be but should we check Subset for this?
There was a problem hiding this comment.
Do you mean FileMetaData::Subset?
There was a problem hiding this comment.
Yeah, I misunderstand this previously, metadata_ is from Subset, but manifest is shared and having lifetime problem here, this looks ok to me
github-actions
bot
added
awaiting committer review
|
This b59082a might have broken the crossbow submissions for committers that don't have their Apache membership set to Public (https://github.com/orgs/apache/people?query=pitrou) |
|
@github-actions crossbow submit wheel-macos* |
|
Revision: 8978cef Submitted crossbow builds: ursacomputing/crossbow @ actions-2f9bd68461
|
|
@github-actions crossbow submit wheel-macos* |
|
Revision: 8978cef Submitted crossbow builds: ursacomputing/crossbow @ actions-b28f7d7852
|
pitrou
changed the title
|
@jorisvandenbossche This seems to have addressed the issue. Should we run the tests another time? |
|
@github-actions crossbow submit -g wheel -g python |
|
Nice! Yes, running once more to be sure is best (it happened before there were no failures in a single run of all wheel builds) |
|
Revision: 8978cef Submitted crossbow builds: ursacomputing/crossbow @ actions-8a0ca7497c
|
|
@raulcd I marked this for 15.0.0. |
…ring (#39632) ### Rationale for this change `ParquetFileFragment` stores a `SchemaManifest` that has a raw pointer to a `SchemaDescriptor`. The `SchemaDescriptor` is originally provided by a `FileMetadata` instance but, in some cases, the `FileMetadata` instance can be destroyed while the `ParquetFileFragment` is still in use. This can typically lead to bugs or crashes. ### What changes are included in this PR? Ensure that `ParquetFileFragment` keeps an owning pointer to the `FileMetadata` instance that provides its `SchemaManifest`'s schema descriptor. ### Are these changes tested? An assertion is added that would fail deterministically in the Python test suite. ### Are there any user-facing changes? No. * Closes: #39562 Authored-by: Antoine Pitrou <[email protected]> Signed-off-by: Antoine Pitrou <[email protected]>
|
After merging your PR, Conbench analyzed the 6 benchmarking runs that have been run so far on merge-commit 4b3de81. There was 1 benchmark result with an error:
There were no benchmark performance regressions. 🎉 The full Conbench report has more details. It also includes information about 1 possible false positive for unstable benchmarks that are known to sometimes produce them. |
…_filtering (apache#39632) ### Rationale for this change `ParquetFileFragment` stores a `SchemaManifest` that has a raw pointer to a `SchemaDescriptor`. The `SchemaDescriptor` is originally provided by a `FileMetadata` instance but, in some cases, the `FileMetadata` instance can be destroyed while the `ParquetFileFragment` is still in use. This can typically lead to bugs or crashes. ### What changes are included in this PR? Ensure that `ParquetFileFragment` keeps an owning pointer to the `FileMetadata` instance that provides its `SchemaManifest`'s schema descriptor. ### Are these changes tested? An assertion is added that would fail deterministically in the Python test suite. ### Are there any user-facing changes? No. * Closes: apache#39562 Authored-by: Antoine Pitrou <[email protected]> Signed-off-by: Antoine Pitrou <[email protected]>
4 participants
Rationale for this change
ParquetFileFragmentstores aSchemaManifestthat has a raw pointer to aSchemaDescriptor. TheSchemaDescriptoris originally provided by aFileMetadatainstance but, in some cases, theFileMetadatainstance can be destroyed while theParquetFileFragmentis still in use. This can typically lead to bugs or crashes.What changes are included in this PR?
Ensure that
ParquetFileFragmentkeeps an owning pointer to theFileMetadatainstance that provides itsSchemaManifest's schema descriptor.Are these changes tested?
An assertion is added that would fail deterministically in the Python test suite.
Are there any user-facing changes?
No.