2018-03-05-tidb-weekly.md

title	Weekly update (February 26 ~ March 04, 2018)
date	2018-03-05
summary	Last week, we landed 27 PRs in the TiDB repositories, 4 PRs in the TiSpark repositories, and 12 PRs in the TiKV and PD repositories.
tags	TiDB TiKV PD TiSpark

Weekly update in TiDB

Last week, we landed 27 PRs in the TiDB repositories.

Added

• Support stream aggregation on TiKV
• Support long VARCHAR
• Set Range and OutputCounts in Coprocessor Response for streaming API
• Support COMMENT = string in partition definition
• Add restrict and cascade in DropTable

Fixed

• Set have_profiling to NO

Improved

• Remove invalid intervals in building Range
• Resolve locks in a batch
• Extract the same part from DNF's leaves
• Simplify the logic of HashCode
• Improve the performance of decoding decimal
• Test coverage:

• Improve the test coverage in the executor package
• Improve the test coverage in the plan package
• Improve the test coverage in the distsql package

• Panic recover:

• Add the recover mechanism for index lookup reader workers
• Add the recover mechanism for union workers
• Add the recover panic in owner.CampaignLoop

Weekly update in TiSpark

Last week, we landed 4 PRs in the TiSpark repositories.

Added

• Add the index covering optimization
• Add the broadcast join support

Improved

• Add more tests
• Add DDL Dump for using TiDB as Spark MetaStore

Weekly update in TiKV and PD

Last week, we landed 12 PRs in the TiKV and PD repositories.

Added

• Export some struct expressions for other tools
• Log panic stacks
• Update the TiKV version
• Add a Raft message flush metrics

Fixed

• Update the Rust dockerfile
• Fix lowspace metrics
• Check Regions in the background goroutine
• Handle the problem that only one row returns in stream aggregation

Improved

• Rename a confused constant variable
• Make end_point_request_max_handle_secs configurable
• Precreate some labal metrics

New contributor (Thanks!)

• TiKV: Greg Weber