Submitted by: @abzalov
Assigned to: @abzalov
Jira_subtask_inward CORE5612
Attachments:
test.7z
core_4731.fbt
Proceeding from the fact that it is necessary to apply adaptive mechanics, and from the fact that in the source code is not used anywhere specifying the plan (perhaps there are reasons for it) - there is hardly a chance that you implement the use of "correct" system indexes in the right places.
Therefore, please remove the restriction on creating indexes for system tables, and disabling/enabling system indexes.
We ourselves will manipulate them at the right time.
In the end, what can be a bad from modification of system indexes, given that DB developers understand what they are doing?
For example, in Oracle there are no restrictions in modifying the system tables - even insert into dual table, not to mention the creation of indexes.
Of course, we are can find a way around this restriction (check_gbak_cheating_insupd, check_gbak_cheating_delete), but why can not you provide a regular way?
Commits: ec86419 bf5614b acb1c28 191a3b6 2bc844e FirebirdSQL/fbt-repository@a4ad37c
====== Test Details ======
todo: update test for CORE4731 ("Prohibit an ability to issue DML or DDL statements on RDB$ tables") - no more restriction on creating user indices for system tables.
Submitted by: @abzalov
Assigned to: @abzalov
Jira_subtask_inward CORE5612
Attachments:
test.7z
core_4731.fbt
Proceeding from the fact that it is necessary to apply adaptive mechanics, and from the fact that in the source code is not used anywhere specifying the plan (perhaps there are reasons for it) - there is hardly a chance that you implement the use of "correct" system indexes in the right places.
Therefore, please remove the restriction on creating indexes for system tables, and disabling/enabling system indexes.
We ourselves will manipulate them at the right time.
In the end, what can be a bad from modification of system indexes, given that DB developers understand what they are doing?
For example, in Oracle there are no restrictions in modifying the system tables - even insert into dual table, not to mention the creation of indexes.
Of course, we are can find a way around this restriction (check_gbak_cheating_insupd, check_gbak_cheating_delete), but why can not you provide a regular way?
Commits: ec86419 bf5614b acb1c28 191a3b6 2bc844e FirebirdSQL/fbt-repository@a4ad37c
====== Test Details ======
todo: update test for CORE4731 ("Prohibit an ability to issue DML or DDL statements on RDB$ tables") - no more restriction on creating user indices for system tables.