* workig commit №1
* working commit №2
* first version of ArithmeticOperationsInAgrFuncOptimize (min, max, sum)
* fix conflicts
* fix №2
* attempt №3
* Description + little mistake
* fix style + expanded optimization + add performance test
* style + merge with new version
* one more branch
* bug fix + merge
* just merge
* some code style and logic fixes
* little changes
* tests
* test(2)
* style
* seems to have fixed everything that was required
* bug fix
* style
* build fix
* fix typo
* fix clang-10 warning
* fix clang-10 warning
* nolint
* add new files to ya.make
* extract all functions from Any
* add last new lines.
* extract all functions from Any
* add last new lines.
* build + requested changes
* code style
* build + little perf test + anyLast
* minor changes
* bug fix
* minor changes
* another minor changes
* fix + experement
* experement 2.0
* experement 3.0
* experement 3.1
* experement 4.0
* last experement(think so)
* just another attempt to fix UB
* minor changes
* I think I won
* ya.make
* fix requested changes
Co-authored-by: Artem Zuikov <chertus@gmail.com>
The following error reported:
FAILED: src/Interpreters/tests/hash_map
...
ld.lld: error: undefined symbol: typeinfo for DB::AggregateFunctionCount
>>> referenced by typeinfo:308 (../contrib/libcxx/include/typeinfo:308)
>>> InterpreterSelectQuery.cpp.o:(std::__1::enable_if<is_pointer_v<DB::AggregateFunctionCount*>, DB::AggregateFunctionCount*>::type typeid_cast<DB::AggregateFunctionCount*, DB::IAggregateFunction>(DB::IAggregateFunction*)) in archive src/libdbms.a
collect2: error: ld returned 1 exit status
And also this one with gold linker:
FAILED: programs/clickhouse-odbc-bridge
... src/AggregateFunctions/libclickhouse_aggregate_functions.a ... src/libdbms.a ...
../contrib/libcxx/include/typeinfo:308: error: undefined reference to 'typeinfo for DB::AggregateFunctionCount'
Due to order?
Introduced by 394fb64a9cda376ca8dfa8bac08a4fbcfa9bf3bf/#11661 I guess.
After #11325 trailing whitespaces has been removed for data skipping
indicies, and it may be different, if you have multiple skip indices,
and in this case new server will not load such tables, because metadata
will be different.
Fix this by re-parse metadata in zookeeper.
