Loading
Commits on Source 24
-
Andrzej Such authored
-
Andrzej Such authored
-
Andrzej Such authored
-
Andrzej Such authored
-
Andrzej Such authored
-
Andrzej Such authored
-
Michał Kudela authored
-
Michał Kudela authored
-
Marcin Sobczyk authored
-
Marcin Sobczyk authored
-
-
-
Add functions in C to operation id computations Use new functions in sql-serializer Use new C functions in hive_fork_manager sql scripts - Warning: additional wrrapers added because the functions are used by indexes (part of data schema), and 'C' functions are part of runtime deployment, so they are unknow during data installation
-
operation_id
-
-
-
hive.operation_id is used now to compute stable id for sorting keyauth changes. The computed id guarantees that change caused by a harfork are lates in a hardfork block.
-
Wrappers where introduced to be use be indexes during instalation data schema and runtime must use exactly the same functions to allow planner to use indexes, what was not implemented when wrapper were added.
-
-
-
postgres by defualt sets COST to 100 for plpgsql functions, and 1 for calling C functions, because the wrappers only directly call C functions it is better to inform planner about their low cost.
-
Marcin authored
-
Marcin authored
indexed which are using expressions needs to be ANALYZE-d, otherwise planner will chose wrong executions plans, because there are nostatistics for the expressions
-
Konrad Botor authored