Weekly update in TiDB

Last week, we landed 46 PRs in the TiDB repositories.

Added

• Add required tables of MySQLX .
• Add TOML configuration file support.

Removed

• Remove the performance schema instrumentation.

Fixed

• Fix show create table with foreign key.
• Cast values only for modified columns in the update statement to avoid unnecessary check.
• Fix an OOM issue when analyzing table.
• Fix a cast (date as datetime) error.
• Fix a bug about the DATE literal.

Improved

• Change the FMSketch hash function from “fnv” to “murmur”.
• Add the PARTITION BY KEY grammar support.
• Refine setting/getting the SQL_MODE variable.
• Support the NVARCHAR syntax.
• Refine the parser of localtime and localtimestamp.
• Push down the ANALYZE INDEX statement.
• Support stream aggregation in the new optimization plan.
• Hide the security information in SHOW PROCESSLIST.
• Add a goroutine pool package utility.
• Refactor the following expression/builtin-function evaluation framework:
  • ADDDATE, SUBDATE
  • TIMEDIFF
  • VALUES
  • ROW
  • SETVAR, GETVAR
  • GREATEST, LEAST
  • CONVERT_TZ
  • MAKETIME
  • ADDTIME, SUBTIME

Weekly update in TiSpark

Added

• Add the JDBC Thrift Server / spark-sql support.
• Add the back-off retry policy.
• Add the Meta table prefix support.

Fixed

• Fix the Null decoding error.
• Fixed some Styling issues.

Improved

• Added the Integration test framework.

New contributors (Thanks!)

• Cheng Lian

Weekly update in TiKV

Last week, We landed 34 PRs in the TiKV repositories.

Added

• Balance regions based on the read flow.
• Make the Raft tick configurable for PD.
• Add a classifier interface for namespace.
• Balance by the namespace.

Fixed

• Fix a race condition for threadpool.
• Fix the active region statistics.
• Fix cleaning up data between regions.
• Stop the threadpool before quitting the scheduler.
• Fix the marshal operator.
• Fix the operator timeout issue.
• Fix a bug in cast_str_as_int.
• Wrap InitLogger in sync.Once for thread safety.
• Fix a compiling error.

Improved

• Add tests for the gRPC services.
• Add two queues cache for PD.
• Cache elf sections for backtrace.
• Optimize JSON.
• Skip checking region 0.
• Adjust metrics for the coprocessor.
• Remove the stale peers as soon as possible.
• Enable pin_l0_filter_and_index by default.
• Enhance metrics for the scheduler.
• Optimize the LIKE expression.
• Check the leader request when becoming a follower after receiving message with higher term.
• Use util::time::instant instead of std::time::instant.
• Add metrics for compression ratio at different levels.
• Remove the statistics that may overflow.

New contributors (Thanks!)

• Rain Li
• OuYang Jin