Update java client version

[mzitnik]

Summary

Checklist

Delete items not relevant to your PR:

• Unit and integration tests covering the common scenarios were added
• A human-readable description of the changes was provided to include in CHANGELOG
• For significant changes, documentation in https://github.com/ClickHouse/clickhouse-docs was updated with further explanations or tutorials

[windsurf-bot]

Other comments (20)

• spark-3.5/clickhouse-spark/src/main/scala/com/clickhouse/spark/read/ClickHouseReader.scala (71-72) The compression codec parameter has been commented out in the class definition, but there's still a reference to it in the commented code on line 72. Either remove the commented code entirely or update the implementation to handle compression with the new API.
• clickhouse-core/src/main/scala/com/clickhouse/spark/client/NodeClient.scala (176-183) The syncInsert method creates an empty ByteArrayInputStream (`is`) to pass to the deserializer after a successful insert, but this stream contains no data. The deserializer will likely fail when trying to process this empty stream. You should use the actual response data from the InsertResponse instead.
• gradle.properties (16-16) The new Maven snapshots repository URL `https://central.sonatype.com/repository/maven-snapshots/` appears to be incorrect. This is a web UI URL, not a repository URL that can be used for dependency resolution.

  The correct URL format should be something like https://s01.oss.sonatype.org/content/repositories/snapshots/ (the previous value) or https://oss.sonatype.org/content/repositories/snapshots/.

  mavenSnapshotsRepo=https://s01.oss.sonatype.org/content/repositories/snapshots/
  

• spark-3.3/clickhouse-spark/src/main/scala/com/clickhouse/spark/read/ClickHouseReader.scala (44-44) The compression codec parameter is commented out on line 44, but it seems the code might still need this parameter. The call to `nodeClient.queryAndCheck()` on line 72 has a comment suggesting the codec parameter was removed, but this might cause issues if compression is still needed elsewhere in the code.
• spark-3.3/clickhouse-spark/src/main/scala/com/clickhouse/spark/read/ClickHouseReader.scala (74-74) The `totalBlocksRead` method now returns a hardcoded `0L` instead of getting the actual blocks read from the response. This will cause inaccurate metrics reporting. If the new API version doesn't support this metric, consider finding an alternative way to track blocks read or document this limitation.
• spark-3.3/clickhouse-spark/src/main/scala/com/clickhouse/spark/read/format/ClickHouseBinaryReader.scala (109-109) Converting a String to bytes without specifying an encoding could lead to platform-dependent results. Consider checking if the value is already a byte array first, or specify an explicit encoding:

        case BinaryType => value match {
          case bytes: Array[Byte] => bytes
          case str: String => str.getBytes("UTF-8")
          case _ => value.toString.getBytes("UTF-8")
        }
  

• build.gradle (248-250) The PR changes the main dependency from `clickhouse-jdbc` to `client-v2` in the clickhouse-core project, but `clickhouse-jdbc` is still being used in the clickhouse-core-it project. This could lead to version conflicts or unexpected behavior if both libraries have different APIs or behaviors. Consider updating the test dependency to use the same client library as the main code.
• spark-3.5/clickhouse-spark/src/main/scala/com/clickhouse/spark/read/format/ClickHouseBinaryReader.scala (109-109) Casting a String to bytes for BinaryType might cause encoding issues if the value is already binary data. Consider checking if the value is already a byte array first.
• gradle.properties (26-27) You're changing from a stable release version (`0.6.3`) to a SNAPSHOT version (`0.9.2-SNAPSHOT`) of the ClickHouse JDBC driver. Using SNAPSHOT dependencies in production code can lead to build instability and unpredictable behavior, as these versions can change without notice.

  Is there a specific reason to use a SNAPSHOT instead of a stable release? If you need features from 0.9.x, consider waiting for a stable release or at least document why a SNAPSHOT is necessary.

• clickhouse-core/src/main/scala/com/clickhouse/spark/client/NodeClient.scala (175-179) There's a redundant creation of ByteArrayInputStream. You read all bytes into `payload` and then create a new ByteArrayInputStream with the same payload. You can simplify this by using the payload array directly in a single ByteArrayInputStream.
• spark-3.4/clickhouse-spark/src/main/scala/com/clickhouse/spark/write/ClickHouseWriter.scala (254-254) The comment `// , codec` doesn't provide any useful information. If this parameter was removed due to an API change in the Java client, it would be better to add a more descriptive comment explaining why the parameter was removed.
• spark-3.4/clickhouse-spark/src/main/scala/com/clickhouse/spark/read/ClickHouseReader.scala (44-44) The compression codec parameter has been removed from the queryAndCheck method call, but the codec field is just commented out rather than properly removed. If compression is no longer needed with the updated Java client, consider removing the field entirely rather than commenting it out.
• clickhouse-core/src/main/scala/com/clickhouse/spark/client/NodeClient.scala (161-165) The `readAllBytes` implementation is inefficient as it reads one byte at a time and builds an array using Stream. Consider using Java's built-in `InputStream.readAllBytes()` method (available since Java 9) or a more efficient buffered reading approach.
• spark-3.4/clickhouse-spark/src/main/scala/com/clickhouse/spark/read/format/ClickHouseBinaryReader.scala (109-109) Casting a String directly to bytes for BinaryType might cause encoding issues if the value is already binary data. Consider checking if the value is already a byte array first.
• spark-3.5/clickhouse-spark/src/main/scala/com/clickhouse/spark/write/ClickHouseWriter.scala (256-256) The commented out `codec` parameter should be removed since it's no longer needed with the updated Java client version.
• spark-3.5/clickhouse-spark/src/main/scala/com/clickhouse/spark/read/format/ClickHouseBinaryReader.scala (89-89) This commented out code can be removed as it's not needed - the proper check is implemented on the next line.
• spark-3.3/clickhouse-spark/src/main/scala/com/clickhouse/spark/read/ClickHouseReader.scala (76-76) The method to get bytes read has changed from `resp.getSummary.getReadBytes` to `resp.getReadBytes()`, but there's still a commented reference to the old method. Consider removing the comment to avoid confusion for future developers.
• spark-3.3/clickhouse-spark/src/main/scala/com/clickhouse/spark/read/format/ClickHouseBinaryReader.scala (89-89) This commented out code should be removed as it's redundant with the active implementation on the next line that uses `isInstanceOf`.
• spark-3.4/clickhouse-spark/src/main/scala/com/clickhouse/spark/read/format/ClickHouseBinaryReader.scala (89-89) This commented out line should be removed as it's not needed and the implementation is already using `isInstanceOf` in the next line.

        case IntegerType if value.isInstanceOf[java.lang.Long] =>
  

• spark-3.4/clickhouse-spark/src/main/scala/com/clickhouse/spark/read/ClickHouseReader.scala (71-71) There's an inline comment `// , codec` on line 71 that appears to be a reminder of a removed parameter. Consider removing this comment or replacing it with a more descriptive explanation of why the parameter was removed.

💡 To request another review, post a new comment with "/windsurf-review".

[windsurf-bot]

There's a commented-out parameter and explanatory comment that should be removed before merging. If the codec parameter is no longer needed in the method call due to API changes, we should just remove it cleanly without leaving commented code.

[windsurf-bot]

The totalBlocksRead method now returns a hardcoded 0L instead of getting the actual value from the response. This will cause the BLOCKS_READ metric to always report 0, which could be misleading for monitoring and debugging purposes.

[windsurf-bot]

The totalBlocksRead method now returns a hardcoded 0L instead of getting the actual block count from the response. This will cause inaccurate metrics reporting. Consider updating this method to retrieve the block count from the new QueryResponse API.

[CLAassistant]

Thank you for your submission! We really appreciate it. Like many open source projects, we ask that you all sign our Contributor License Agreement before we can accept your contribution.
2 out of 3 committers have signed the CLA.

✅ mzitnik
✅ ShimonSte
❌ Shimon Steinitz

---

Shimon Steinitz seems not to be a GitHub user. You need a GitHub account to be able to sign the CLA. If you have already a GitHub account, please add the email address used for this commit to your account.
_{You have signed the CLA already but the status is still pending? Let us recheck it.}