BLOB fields did not work as grouping keys at all, they were assigned
value NULL for any value, be it NULL or not. The fix is in the
rowaggregation.cpp in the initMapping(), a switch/case branch was added
to handle BLOB field copying there.
Also, TEXT columns did not distinguish between NULL and empty string in
the grouping algorithm, now they do. The fix is in the equals()
function, now we specifically check for isNull() equality between
values.
This fixes discrepance with the server, which assigns DATE type to
last_day()'s result.
Now we also assigns DATE result type and, also, use proper
dataconvert::Day data structure to return date.
Tests agree with InnoDB.
Also, this patch includes test for MCOL-5669, to show we fixed it.
The fix is simple: enable subtotals in single-phase aggregation and
disable parallel processing when there are subtotals and aggregation is
single-phase.
* feat(PrimProc): MCOL-5950 Improve disk-based aggregation finalization
Iterate over the rows in the plain vector of RGData instead of
iterating over the hashmap. This reduces the complexity and speeds
up finalization (by up to the twice in the certain cases)
* replace magic constant with muggle constant
* chore(build): refactor main CMakeLists.txt
* chore(build): fix boost version for packages, set clang-20 only for amd and arm
* chore(build): boost 4 sm
* chore(build): boost dep for rowgroup
* chore(build): toolset for boost
* chore(build): suppress clang warnings for boost
* chore(ci, build): use ASAN for unittest on ubuntu 24.04 only, added custom cmake flag option for bootstrap, custom params for new and existing pipelines
* chore(build): sort bootstrap flags
* chore(CI): remove publish pkg step, adding clickable link instead to publish steps, fix customenv
The purpose of this changeset is to obtain list of partitions from
SELECT_LEX structure and pass it down to joblist and then to
CrossEngineStep to pass to InnoDB.
* move GROUP_CONCAT/JSON_ARRAYAGG storage to the RowGroup from
the RowAggregation*
* internal data structures (de)serialization
* get rid of a specialized classes for processing JSON_ARRAYAGG
* move the memory accounting to disk-based aggregation classes
* allow aggregation generations to be used for queries with
GROUP_CONCAT/JSON_ARRAYAGG
* Remove the thread id from the error message as it interferes with the mtr
