clickhouse
https://github.com/yandex/clickhouse
C++
ClickHouse is a free analytic DBMS for big data.
Triage Issues!
When you volunteer to triage issues, you'll receive an email each day with a link to an open issue that needs help in this project. You'll also receive instructions on how to triage issues.
Triage Docs!
Receive a documented method or class from your favorite GitHub repos in your inbox every day. If you're really pro, receive undocumented methods or classes and supercharge your commit history.
C++ not yet supported67 Subscribers
View all SubscribersAdd a CodeTriage badge to clickhouse
Help out
- Issues
- toString over DateTime falsely claims isInjective: uniqExact(toString(dt)) counts DST-fold duplicates, and optimize_truncate_order_by_after_group_by_keys drops explicit ORDER BY tiebreakers
- Segfault in MergeTreeIndexConditionMinMax on LowCardinality(Nullable(String)) with IN condition (23.8.11.29, 23.8.16.42)
- optimize_injective_functions_in_limit_by makes LIMIT n BY return extra rows per key: the pass trusts false isInjective claims (toString over DateTime in a DST fold, toFixedString)
- Variant comparisons disagree with each other: v = '1' cross-type-matches a UInt64 variant while v IN ('1') and uniqExact treat types as distinct, and v = 1 errors while v = '1' works
- Wrong primary-key pruning survives #110344 for `toStartOfInterval` over a `Date32` key: missing rows and phantom `count()` for out-of-range values
- `toDate`/`toDateTime` claim always-monotonic while the conversion wraps at default `date_time_overflow_behavior` — wrong primary-key pruning on `DateTime64`/`Date32` keys
- `enable_identifier_resolve_cache` changes results for cross-shadowing aliases: `WHERE` row counts flip with the cache on/off and with select-list order
- DeltaLake (delta-kernel): mixed-type WHERE comparison silently drops rows — the pushed skip literal is non-accurate-cast to the column type, so stats-based file pruning skips matching files
- Primary key on `mortonEncode`/`hilbertEncode` silently loses rows for values >= 2^32: space-filling-curve index analysis assumes the encode is lossless
- Iceberg v3: filtering on `_last_updated_sequence_number` or `_row_id` silently loses rows whose physical lineage values are NULL (inherited)
- Docs
- C++ not yet supported