index does not match the table

[EinTerraner]

This error occurs after several runs of the exchange procedure. Sometimes it fires after 2 runs or even after 10 rounds.



I have a table, sorted by group-number, line-number and subtext-number (like "1.7.3") all are numeric
The result should look something like this
before moving -> after moving ->
To move a group (or a row/sub-row), you just have to swap the corresponding numbers. I have the following code for this.

WITH THIS && MyGrid.DataMove(Tc_Line as Object)
** Use a mirror-cursor because .RecordSource might have a active filter
    USE DBF(.RecordSource) ALIAS(.TmpAlias) IN SELECT(.TmpAlias) SHARED AGAIN 
** Lo_Line.BEDGRUP = Destination
** Tc_Line.BEDGRUP = Source
    SELECT * FROM (.TmpAlias) ;
        INTO CURSOR (.SrcAlias) READWRITE ;               && fetch source group
        WHERE Lo_Line.BEDGRUP == BEDGRUP
    SELECT * FROM (.TmpAlias) ;
        INTO CURSOR (.DstAlias) READWRITE ;              && fetch target group
        WHERE Tc_Line.BEDGRUP == BEDGRUP
    REPLACE BEDGRUP WITH Lo_Line.BEDGRUP ;        && Rename target group to source group
        IN (.DstAlias) ;
        FOR .t.
    REPLACE BEDGRUP WITH Tc_Line.BEDGRUP ;        && Rename source group to target group
        IN (.SrcAlias) ;
        FOR .t.
    BLANK IN (.TmpAlias) ;                                            && wipe both groups
        FOR INLIST(BEDGRUP, Lo_Line.BEDGRUP, Tc_Line.BEDGRUP)
    Lc_Done = .t.

and to reinsert those two xchanged groups/Lines

** Replace moved Group back to table
** recycle blank lines.
** otherwise table becomes large
    IF Lc_Done
        GO TOP IN (.SrcAlias)
        DO WHILE !EOF(.SrcAlias)
            SELECT(.SrcAlias)
            SCATTER NAME Lo_Line MEMO
            SELECT(.TmpAlias)
            LOCATE FOR EMPTY(BEDNUMM)
            IF !FOUND(.TmpAlias)
                APPEND BLANK IN (.TmpAlias)
            ENDIF
            GATHER NAME Lo_Line MEMO
            SKIP IN (.SrcAlias)
        ENDDO

        GO TOP IN (.DstAlias)
        DO WHILE !EOF(.DstAlias)
            SELECT(.DstAlias)
            SCATTER NAME Lo_Line MEMO
            SELECT(.TmpAlias)
            LOCATE FOR EMPTY(BEDNUMM)
            IF !FOUND(.TmpAlias)
                APPEND BLANK IN (.TmpAlias)
            ENDIF
            GATHER NAME Lo_Line MEMO
            SKIP IN (.DstAlias)
        ENDDO
    ENDIF

any idea what i'm doing wrong?

 

[Scott24x7]

What indexes exist on the table? Are they compound? Are they creating a unique index?

How big is the table?
Also, why are you doing it this way?

Can you explain what you are trying to achieve (what your business logic is in this case)?

 

[EinTerraner]

What indexes exist on the table? Are they compound? Are they creating a unique index?

there actually two tags. native table. not in a DB

* structure
* DOCID C(8)
* BEDGRUP,BEDPOSI,BEDSUBP N(3)
index on DOCID+trans(bedgrup, "@L 999")+;
               trans(bedposi, "@L 999")+;
               trans(bedsubp, "@L 999") ;
        tag ID_EDIT ;
        for !empty(bedgrup)
index on DOCID+trans(bedgrup, "@L 999")+;
               trans(bedposi, "@L 999")+;
               trans(bedsubp, "@L 999") ;
        tag ID_GSUM ;
        for bedposi=999

How big is the table?

not really, but might contain up to 500 rows, splitted in 5-30 groups (BEDGRUP) and in the group several positions (BEDPOSI) and within the positions s few detail-lines (BEDSUBP)

Also, why are you doing it this way?

the code above is only a snippet from a whole CASE-construct. here is a rough functional description

case SWAP BEDGRUP 2/3
case move BEDPOSI 5 to BEDGRUP 4 as BEDPOSI 3
case import BEDGRUP 4,5,6 between BEDGRUP 2/3
case insert BEDPOSI before/after current BEDPOSI or end of BEDGRUP
case delete any BEDGRUP or BEDPOSI or BEDSUBP

Within these CASE blocks, it is first checked whether the process is valid. For example, the first group cannot be moved up or the last existing group cannot be moved down.
Then the renumbering and space release are prepared and acknowledged with Lc_DONE = .t.

Can you explain what you are trying to achieve (what your business logic is in this case)?

there are more needs for this usage.
1.) Correct the recorded order of the groups/positions/detail lines
2.) Import entire groups/positions with detail lines from another document (pre-defined specifications)
3.) Insert a new group/position/detail line

2.) Record several groups with n positions (+ n detail lines) from an already recorded but not completed processing description for resubmission. And they can be inserted in the middle of the document. A restructuring of the group order may also be necessary here.

the next problem we have. the user is creating new documents offline. far away from a stable internet-connection and other civilisation facilities (far out in the deep forest). so he's creating some new documents without a valid DOCID. This will be given, as soon he might reconnect to the office-computer. usually only towards the weekend (Thursday or Friday)