d3c6045f63
*: replace compareDatum by compare ( #30421 )
2021-12-06 20:41:56 +08:00
7755d25aba
*: Integrate SkyWalking-eyes to check license headers in CI and fix licenses headers ( #27198 )
2021-08-16 18:52:00 +08:00
55b3e1110c
*: reduce the allocation caused by Datum.Copy ( #15219 )
2020-03-09 17:38:46 +08:00
d9e02b76dd
expression: remove redundant "errors.Trace()" in "expression" package ( #8609 )
2018-12-07 12:20:31 +08:00
32b1dbd8d5
*: rename "github.com/pkg/errors" to "github.com/pingcap/errors" ( #8136 )
...
We import "github.com/pkg/errors" in the code, and actually put
the "github.com/pingcap/errors" there in the vendor path.
That's a dirty hack, and prevent the introduce of Go module.
2018-11-01 16:09:07 +08:00
cb03f2bec1
move from juju/errors to pkg/errors ( #7151 )
2018-09-12 15:42:20 +08:00
06021c7cfe
remove types.Row interface ( #7170 )
2018-07-27 13:45:03 +08:00
eec1c524dc
util: refine Row.GetDatum when tp is MyDecimal ( #6087 )
2018-03-19 17:11:07 +08:00
fc6376981d
*: remove GetSessionVars() in expression evaluation ( #5683 )
2018-01-22 11:04:32 +08:00
2550969cae
aggregation, plan: split the presentation and evaluation layers of aggregation functions ( #5635 )
2018-01-15 15:48:03 +08:00
6b6afb6c55
expression, plan: support builtin aggregation function 'bit_or' ( #5145 )
2017-12-19 05:51:08 -06:00
8b15cb52b7
*: move StatementContext to its own package. ( #5177 )
2017-11-22 00:11:14 +08:00
5c76863a79
plan, expression: remove useless error return ( #5085 )
2017-11-13 18:40:58 +08:00
62433663e5
util/types: change types path ( #5007 )
...
* util/types: change types path
2017-11-04 10:37:14 -05:00
ed7f68ca9b
*: Add Row interface ( #4859 )
2017-10-23 11:36:28 +08:00
6f84392d40
*: optimize SortExec ( #4622 )
...
* *: change key of orderByRow from "key []types.Datum" to "key []*types.Datum"
* address comment
2017-09-25 19:50:26 +08:00
25b5a2ce18
*: Refactor aggregation ( #4605 )
...
Aggregation results were stored in aggregation functions, so one map is created for each function to store different groups’ results, which requires much memory.
The aggregation functions are refactored and are now stateless, and the maps are kept in executor (executor.HashAggExec and mocktikv.aggregateExec) so that only one map is needed to store all the functions’ results, which reduced the memory usage.
2017-09-22 19:41:01 +08:00
2762c44161
*: move aggregate functions to an individual package "aggregation" ( #4549 )
2017-09-18 15:07:35 +08:00
