Project

General

Profile

Bug #10689

Wrong temp-table FOR EACH sorting criteria

Added by Teodor Gorghe 10 months ago. Updated 6 months ago.

Status:
Closed
Priority:
Normal
Assignee:
Teodor Gorghe
Target version:
-
Start date:
Due date:
% Done:

100%

billable:
No
vendor_id:
GCD
case_num:
version_reported:
version_resolved:
reviewer:
Constantin Asofiei
production:
No
env_name:
topics:

Index-rules.png (219 KB) Eduard Soltan, 10/09/2025 04:32 AM

book.df (1016 Bytes) Teodor Gorghe, 10/09/2025 08:40 AM

Related issues

Related to Database - Bug #9267: Index Selection in case of a multiple index matches. Test

History

#1 Updated by Teodor Gorghe 10 months ago

  • Related to Bug #9267: Index Selection in case of a multiple index matches. added

#3 Updated by Teodor Gorghe 10 months ago

The following test case fails on FWD because of the unimplemented main index selection by definition:

BLOCK-LEVEL ON ERROR UNDO, THROW.

USING OpenEdge.Core.Assert FROM PROPATH.

CLASS tests.TestTempTableSort:
    DEFINE TEMP-TABLE tt1 NO-UNDO
        FIELD f1 AS INTEGER
        FIELD f2 AS CHARACTER
        FIELD f3 AS CHARACTER
        FIELD f4 AS CHARACTER
        INDEX idx1 f1 f2 f3
        INDEX aidx2 f1 f2 f4.

    @Test.
    METHOD PUBLIC VOID TestIdx1():
        DEFINE VARIABLE i AS INTEGER NO-UNDO INITIAL 1.
        EMPTY TEMP-TABLE tt1.
        CREATE tt1.
        ASSIGN
            tt1.f1 = 1
            tt1.f2 = '1'
            tt1.f3 = '2'
            tt1.f4 = '1'.
        CREATE tt1.
        ASSIGN
            tt1.f1 = 1
            tt1.f2 = '1'
            tt1.f3 = '1'
            tt1.f4 = '2'.
        RELEASE tt1.

        FOR EACH tt1:
            Assert:Equals(STRING(i), tt1.f3).
            i = i + 1.
        END.
    END METHOD.
END CLASS.


╷
├─ JUnit Jupiter ✔
├─ JUnit Vintage ✔
├─ JUnit Platform Suite ✔
└─ FWD Test ✔
   └─ tests.TestTempTableSort ✔
      └─ testIdx1 ✘ Expected: 1 but was: 2

4GL selects the index by the order of definition on temp-table, but FWD selects the index by definition name (by alphabetical order).

#4 Updated by Eduard Soltan 10 months ago

From this list, the only thing left to implement is temp-tables go by definition order

#5 Updated by Teodor Gorghe 10 months ago

  • Assignee set to Teodor Gorghe
  • Status changed from New to WIP

Created task branch 10689a.

#6 Updated by Teodor Gorghe 10 months ago

  • % Done changed from 0 to 100
  • Status changed from WIP to Review
  • reviewer Constantin Asofiei added

Committed revision 16213 on task branch 10689a:
- Improved index selection for temp-table: added implementation temp-tables go by definition order.

#7 Updated by Teodor Gorghe 10 months ago

There is also this test case, which passes on 4GL, but fails on FWD:

BLOCK-LEVEL ON ERROR UNDO, THROW.

USING OpenEdge.Core.Assert FROM PROPATH.

CLASS tests.TestWorkTableQuery:
    @Setup.
    METHOD PUBLIC VOID Setup():
        CREATE book.
        ASSIGN
            book.f1 = 1
            book.f2 = '2'
            book.f3 = '3'
            book.f4 = 'b'.
            book.f5 = '1'.
        CREATE book.
        ASSIGN
            book.f1 = 1
            book.f2 = '2'
            book.f3 = '3'
            book.f4 = 'a'.
            book.f5 = '2'.
    END METHOD.

    @Test.
    METHOD PUBLIC VOID TestMainIndex():
        DEFINE VARIABLE cExpected AS CHARACTER EXTENT 2 NO-UNDO.
        DEFINE VARIABLE i AS INTEGER NO-UNDO INITIAL 1.
        cExpected[1] = 'b'.
        cExpected[2] = 'a'.
        FOR EACH book WHERE book.f1 = 1 AND book.f2 = '2' AND book.f3 = '3':
            Assert:Equals(cExpected[i], book.f4).
            i = i + 1.
        END.
    END METHOD.

    @Teardown.
    METHOD PUBLIC VOID Teardown():
        FOR EACH book:
            DELETE book.
        END.
    END METHOD.
END CLASS.

I think the index selection for this is right (second index must be selected), but on FWD, the second record somehow gets selected first.

#8 Updated by Teodor Gorghe 10 months ago

Teodor Gorghe wrote:

There is also this test case, which passes on 4GL, but fails on FWD:
[...]

I think the index selection for this is right (second index must be selected), but on FWD, the second record somehow gets selected first.

Issue fixed on 10480a, no fix required in this task.

#9 Updated by Constantin Asofiei 9 months ago

Review of 10689a rev 16213:
  • check copyright year and set it to 2025
  • Schema record type ({@link #TEMP_TABLE} or {@link #WORK_TABLE}). - this is not completely right - you can have TABLE or BUFFER here, too. Where TABLE is for a .df table, and BUFFER can be for both. Also, WORK_TABLE AFAIK can not define indexes. Thus breakTieWorkTable should be renamed back to breakTie
  • RecordPhraseData.recordType has no javadoc.

#10 Updated by Teodor Gorghe 9 months ago

Committed revision 16214 on task branch 10689a:
- Addressed code review #10689-9.

Committed revision 16215 on task branch 10689a:
- Fix history entry.

#11 Updated by Constantin Asofiei 6 months ago

  • Status changed from Review to Internal Test

#12 Updated by Teodor Gorghe 6 months ago

Rebased to trunk rev 16358

#13 Updated by Constantin Asofiei 6 months ago

  • Status changed from Internal Test to Merge Pending

Merge to trunk after 10636a

#14 Updated by Teodor Gorghe 6 months ago

  • Status changed from Merge Pending to Test

Branch 10689a was merged to trunk rev. 16386 and archived.

#15 Updated by Constantin Asofiei 6 months ago

  • Status changed from Test to Closed

Also available in: Atom PDF