Project

General

Profile

Bug #7364

Build a full test-suite for fast copy-temp-table

Added by Alexandru Lungu about 1 year ago. Updated about 1 year ago.

Status:
New
Priority:
Normal
Assignee:
Andrei Bălteanu
Target version:
-
Start date:
Due date:
% Done:

0%

billable:
No
vendor_id:
GCD
case_num:
version_reported:
version_resolved:

History

#1 Updated by Alexandru Lungu about 1 year ago

There were many recent changes for FastCopyHelper and some are still pending. However, we don't have such a complete test-suite yet for fast copy-temp-table (except testcases/uast/copy-temp-table).
Please do the full testing report on this after #7351 is completed as there are several functional changes there. Before #7351 is completed, the test-suite can be built.

In fact, I have a starting test-suite, but it should be extended and retested. I will upload it to testcases.
Any issues observed should be repaired immediately:
  • Set-up testing suite and do the tests. Report any current mismatches. [retest after #7351]
  • Extend the testing suite. AFAIK, the scenarios with extent fields are not that well tested.
    • Do fast copy between tables with normalized / denormalized fields (norm with norm / denorm with norm / norm with denorm / denorm with denrom).
    • Do fast copy between tables with datetime-tz and character fields (case sensitive / case insensitive).
    • Try all of the previous combinations with append, replace and loose-mode options.
    • Do more/better tests with before tables.
    • Try with static and dynamic tables (dynamic to dynamic / static to dynamic / dynamic to static / static to static).
  • Check that fast copy is not used in cases like:
    • Unique index schema mismatch
    • Existence of triggers
    • Destination table is already queried by a FOR EACH
    • AFAIK, replace is not done with FastCopyHelper; I will create a separate task for this if it shows up as a bottleneck
  • Extract the final SQL. Make sure that all that scenarios are fast-copy (INSERT INTO SELECT FROM) and not one-by-one (INSERT, INSERT, INSERT, etc.) [after #7351]

#2 Updated by Alexandru Lungu about 1 year ago

  • Assignee set to Andrei Bălteanu

Also available in: Atom PDF