Weekly update in TiDB

Last week, we landed 31 PRs in the TiDB repositories.

Added

• Use etcd to elect DDL job leader.

• Support Load Data with specified column list.

• Add the following builtin functions:

  • umcompress and uncompressdlength
  • convert_tz
  • period-diff

• Support the Json type and Json data encoding/decoding.

• Add a Jenkins CI in Pull Request.

• Add the EvalDuration and EvalTime interface for expressions.

• Support Index Lookup Join in new planner.

Fixed

• Add default value for the Password Validation system variables.

• Fix the issue of retrying with no limitation when committing primary key failed.

• Fix the issue of context cancellation triggering onSendFail and dropping cache: context cancel doesn't mean cache out of data.

• Fix a bug in Sort Merge Join.

• Correct comment mistakes.

• Reprocess SQL statement when meeting the infoschema change error.

• Update an error code to MySQL standard error code.

• Consider schema changing when retrying prepared statement.

Improved

• Refator range calculation related code.

• Consider task type when building the physical plan: to make cost more precise.

• Support more SQL grammar for Lock Options.

• Rename SupportRequestType to IsRequestTypeSupported to improve readability.

• Support more SQL grammar for Compress Options in the AlterTable statement.

• Change the schema of statistic table.

Weekly update in TiKV

Last week, We landed 16 PRs in the TiKV repositories.

Added

• Support table scan with DAG in coprocessor.
• Add Jenkins support, see 643, 644, 1835, 1836.
• Write Raft log synchronously to ensure data safety.

Fixed

• Use IEC size for size output.
• Enlarge max duration for scheduler histogram metrics.

Improved

• Use SmallGroupFirstQueue in endpoint scheduler.
• Fetch snapshot lazily to reduce the cost of creating unnecessary snapshot.
• Create WriteBatch with capacity to avoid reallocating.