diff --git a/CHANGELOG.md b/CHANGELOG.md index 693d47e..2c84893 100644 --- a/CHANGELOG.md +++ b/CHANGELOG.md @@ -9,7 +9,7 @@ v0.3.1 (Unreleased) * **SummingMergeTree with multiple sum columns**: `HasSummingMergeTreeEngine("A", "B")` now generates valid DDL (`SummingMergeTree((A, B))`). Previously it emitted a comma-separated argument list (`SummingMergeTree(A, B)`), which ClickHouse rejects with `NUMBER_OF_ARGUMENTS_DOESNT_MATCH`. Single-column and no-column usage are unaffected. ### Dependencies -* **`ClickHouse.Driver` 1.3.0 → [1.4.0](https://github.com/ClickHouse/clickhouse-cs/releases/tag/1.4.0)** across the provider and both test projects (the pins must move together or restore fails with `NU1605`). No provider code changes were required. Two driver behaviour changes are worth knowing: query responses are now compressed with **zstd** instead of gzip by default (decoded transparently, no action needed), and reading a column from `ClickHouseDataReader` with no current row now throws `InvalidOperationException` instead of returning a value. The provider's own materialization always positions the reader first, so the latter only affects code that uses the underlying `ClickHouseConnection` directly. +* **`ClickHouse.Driver` 1.3.0 → [1.5.0](https://github.com/ClickHouse/clickhouse-cs/releases/tag/1.5.0)** across the provider and both test projects (the pins must move together or restore fails with `NU1605`). No provider code changes were required. Two behaviour changes from driver [1.4.0](https://github.com/ClickHouse/clickhouse-cs/releases/tag/1.4.0) are worth knowing: query responses are now compressed with **zstd** instead of gzip by default (decoded transparently, no action needed), and reading a column from `ClickHouseDataReader` with no current row now throws `InvalidOperationException` instead of returning a value. The provider's own materialization always positions the reader first, so the latter only affects code that uses the underlying `ClickHouseConnection` directly. Driver 1.5.0 adds two changes that apply to the provider: a `DateTime`/`DateTime64` column whose `Fixed/UTC±HH:MM:SS` timezone has a minute or second field above 59 (for example `Fixed/UTC+05:60:00`, which servers before ClickHouse 26.10 read as `+06:00`; 26.10 rejects the name) now reads with the correct wall-clock time, not as UTC; and `ClickHouseBulkCopySerializationException` from an insert now has `RowIndex`, the zero-based position of the failed row in the driver batch. The driver starts a new batch every `InsertOptions.BatchSize` rows (100,000 by default), and `SaveChanges` makes one insert call for each table and column set, so the index is not a position in the EF batch or in the input sequence. The new native TCP client in 1.5.0 has no ADO.NET layer, so the provider does not use it. v0.3.0 --- diff --git a/RELEASENOTES.md b/RELEASENOTES.md index af2af0e..dc9eec8 100644 --- a/RELEASENOTES.md +++ b/RELEASENOTES.md @@ -9,7 +9,7 @@ v0.3.1 (Unreleased) * **SummingMergeTree with multiple sum columns** now produces valid DDL. Configuring more than one sum column (`HasSummingMergeTreeEngine("A", "B")`) previously emitted `SummingMergeTree(A, B)`, which ClickHouse rejects with `NUMBER_OF_ARGUMENTS_DOESNT_MATCH` — the engine takes a single optional parameter that must be a tuple of columns. Multiple columns are now wrapped in a tuple (`SummingMergeTree((A, B))`); single-column and no-column usage are unchanged. ### Dependencies -* The provider now builds against **`ClickHouse.Driver` [1.4.0](https://github.com/ClickHouse/clickhouse-cs/releases/tag/1.4.0)**, up from 1.3.0, and picks up that release's substantial read-path allocation and GC reductions for free. Upgrading needs no changes to your code or configuration. Two driver-level behaviour changes are worth being aware of: query responses now come back **zstd**-compressed rather than gzip by default — the driver decodes this for you, so it is invisible unless you were inspecting raw responses — and reading a column value from a `ClickHouseDataReader` that has no current row (before the first `Read()`, or after `Read()` has returned `false`) now throws `InvalidOperationException` rather than returning something arbitrary. EF Core always positions the reader before materializing, so that only matters if you drop down to the raw `ClickHouseConnection` yourself. +* The provider now builds against **`ClickHouse.Driver` [1.5.0](https://github.com/ClickHouse/clickhouse-cs/releases/tag/1.5.0)**, up from 1.3.0, and picks up the substantial read-path allocation and GC reductions from driver [1.4.0](https://github.com/ClickHouse/clickhouse-cs/releases/tag/1.4.0) for free. Upgrading needs no changes to your code or configuration. Two driver-level behaviour changes from 1.4.0 are worth being aware of: query responses now come back **zstd**-compressed rather than gzip by default — the driver decodes this for you, so it is invisible unless you were inspecting raw responses — and reading a column value from a `ClickHouseDataReader` that has no current row (before the first `Read()`, or after `Read()` has returned `false`) now throws `InvalidOperationException` rather than returning something arbitrary. EF Core always positions the reader before materializing, so that only matters if you drop down to the raw `ClickHouseConnection` yourself. Driver 1.5.0 also fixes a timezone read: a `DateTime` or `DateTime64` column declared with a `Fixed/UTC±HH:MM:SS` timezone whose minute or second field is above 59 (such as `Fixed/UTC+05:60:00`, which ClickHouse before 26.10 reads as `+06:00`; 26.10 rejects the name) now comes back with the correct wall-clock time instead of as UTC. And when an insert fails because a value cannot be serialized, `ClickHouseBulkCopySerializationException` now has a `RowIndex` property next to `Row`: the zero-based position of the failed row in the batch the driver sent. A driver batch holds up to 100,000 rows by default, and `SaveChanges` sends each table separately, so the index can differ from the row's position in your own list. v0.3.0 --- diff --git a/src/EFCore.ClickHouse/EFCore.ClickHouse.csproj b/src/EFCore.ClickHouse/EFCore.ClickHouse.csproj index f0e349d..a85a22a 100644 --- a/src/EFCore.ClickHouse/EFCore.ClickHouse.csproj +++ b/src/EFCore.ClickHouse/EFCore.ClickHouse.csproj @@ -26,7 +26,7 @@ - + diff --git a/test/EFCore.ClickHouse.FunctionalTests/EFCore.ClickHouse.FunctionalTests.csproj b/test/EFCore.ClickHouse.FunctionalTests/EFCore.ClickHouse.FunctionalTests.csproj index 6e1a118..a3f4a10 100644 --- a/test/EFCore.ClickHouse.FunctionalTests/EFCore.ClickHouse.FunctionalTests.csproj +++ b/test/EFCore.ClickHouse.FunctionalTests/EFCore.ClickHouse.FunctionalTests.csproj @@ -9,7 +9,7 @@ - + all diff --git a/test/EFCore.ClickHouse.Tests/EFCore.ClickHouse.Tests.csproj b/test/EFCore.ClickHouse.Tests/EFCore.ClickHouse.Tests.csproj index fdc0d6d..f9f2834 100644 --- a/test/EFCore.ClickHouse.Tests/EFCore.ClickHouse.Tests.csproj +++ b/test/EFCore.ClickHouse.Tests/EFCore.ClickHouse.Tests.csproj @@ -9,7 +9,7 @@ - +