Fix sparse nullable serialization inconsistency in Tuple subcolumns#91932
Merged
Algunenano merged 8 commits intoClickHouse:masterfrom Dec 12, 2025
Merged
Fix sparse nullable serialization inconsistency in Tuple subcolumns#91932Algunenano merged 8 commits intoClickHouse:masterfrom
Algunenano merged 8 commits intoClickHouse:masterfrom
Conversation
Contributor
|
Workflow [PR], commit [642bad1] Summary: ❌
|
clickhouse-gh
bot
added
the
pr-bugfix
Algunenano
reviewed
Dec 11, 2025
| namespace DB | ||
| { | ||
|
|
||
| bool SerializationInfoSettings::supportsSparseSerialization(const IDataType & type) const |
Member
There was a problem hiding this comment.
Let's find a different name, either for this one or for IDataType(). Otherwise it will be really easy to mess up again.
Not sure 100%, but maybe we could call this SerializationInfoSettings::shouldUseSparseSerialization. Another option is to keep only IDataType and make SerializationInfoSettings a mandatory parameter. WDTY?
Collaborator
Author
There was a problem hiding this comment.
Let's find a different name, either for this one or for IDataType(). Otherwise it will be really easy to mess up again.
Sure.
Another option is to keep only IDataType and make SerializationInfoSettings a mandatory parameter.
I did try that approach, but the required changes were large: every derivative type would need to be updated. Also, the semantics feel a bit off: whether a data type supports sparse encoding shouldn't really be governed by SerializationInfoSettings. That setting should describe limitations of a particular serialization pipeline (e.g. MergeTree serialization, or Native reader/writer), not the capabilities of the type system itself. DataTypeNullable should fundamentally support sparse encoding. It's the specific serialization path that may decide sparse is not allowed under certain SerializationInfoSettings.
Collaborator
Author
There was a problem hiding this comment.
I'll rename to SerializationInfoSettings::canUseSparseSerialization
Member
|
The crash seems related: |
Collaborator
Author
Yes, it looks related. I'm digging into it now. |
Member
|
This also closes #91380. We can add a test later on |
Algunenano
reviewed
Dec 11, 2025
| if (supportsSparseSerialization() && kind == ISerialization::Kind::SPARSE) | ||
| if (settings.canUseSparseSerialization(*this) && kind == ISerialization::Kind::SPARSE) | ||
| serialization = std::make_shared<SerializationSparse>(serialization); | ||
| else if (kind == ISerialization::Kind::DETACHED) |
Member
There was a problem hiding this comment.
Any reason for removing the else except that if you take the previous if you won't take the second one?
Collaborator
Author
There was a problem hiding this comment.
This is definitely an unintended change. It was probably introduced during some local debugging and not fully cleaned up. Good catch!
src/Formats/NativeReader.cpp
Outdated
| /// NativeReader must enable nullable sparse support here. Since it operates on in-memory state, it should | ||
| /// be able to handle all possible serialization variants. | ||
| SerializationInfoSettings settings; | ||
| settings.allowNullableSparse(); |
Member
There was a problem hiding this comment.
Is this necessary only if server >= DBMS_MIN_REVISION_WITH_NULLABLE_SPARSE_SERIALIZATION or it doesn't matter?
Collaborator
Author
There was a problem hiding this comment.
It doesn't matter. The newest reader should be able to handle all serialization variants.
azat
reviewed
Dec 11, 2025
|
|
||
| bool canUseSparseSerialization(const IDataType & type) const; | ||
|
|
||
| void allowNullableSparse() { nullable_serialization_version = MergeTreeNullableSerializationVersion::ALLOW_SPARSE; } |
Member
There was a problem hiding this comment.
Minor, but creating object and call a method on it later looks adhoc, maybe adding separate method to create an object with nullable_serialization_version = MergeTreeNullableSerializationVersion::ALLOW_SPARSE will be better? i.e. createSerializationInfoSettingsWithNullableSparse (or a static function)? And the comment can be put there with an explanation
Collaborator
Author
There was a problem hiding this comment.
Yes. I’ll add a method that builds this setting with the widest serialization capability, so it can be extended in the future, along with proper documentation.
Merged
via the queue into
ClickHouse:master
with commit Dec 12, 2025
e277cb3
127 of 130 checks passed
1 task
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
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 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.
Changelog category (leave one):
Changelog entry (a user-readable short description of the changes that goes into CHANGELOG.md):
Fix a serialization inconsistency between sparse and nullable substreams in Tuple columns that could lead to corrupted parts or crashes during reading. This addresses #91851 . @Algunenano Could you please help check if this can fix the stress test in private repo? @CurtizJ Could you also help take a look please? Thanks!
Documentation entry for user-facing changes