Hi!

In the moment between program figured out that PK is unique and record is saved, another user runs exactly the same process and use the same PK. Because sys(3) is time-depended, there is a high chance to get the same PK and write it to database. To avoid conflicts, you can only lock the entire table until you finish the insert (that is a bad idea).

There are a lot of other solutions to organize PK in VFP database.

In particular, take a look to the FAQ article about auto-increment fields in VFP database at the VFP Forum at the www.tek-tips.com site.

HTH.

>I am trying to generate and capture a unique id using the sys3 command. Once a id is created it checks a table that holds all the previous id's that were generated, if the id is found in the table already, then continue with process until it comes across one that doesnt match and then write to the table. Is this code accomplishing this? I am finding duplicate records in my Keytable.dbf.
>
>PARAMETERS mOdeType, pk_find
>
>*LOCAL m.pk_find
>
>m.pk_find = .F.
>
>IF !USED('KEYTABLE')
> USE (oVar.SysPath) + 'KEYTABLE.dbf' IN 0 SHARED
>ENDIF
>
>SELE KEYTABLE
>GO TOP
>
>*!* SET MULTILOCKS ON
>=CURSORSETPROP("BUFFERING",5)
>
>
>DO WHILE m.pk_find = .F.
> oVar.KeyId = SYS(3)
>
> SELE KEY_ID FROM KEYTABLE ;
> WHERE ALLT(KEYTABLE.KEY_ID) == ALLT(oVar.KeyId) ;
> INTO CURSOR PkIdFound
>
> SELE PkIdFound
> IF RECCOUNT() > 0 && PK ID FOR THIS PARTICULAR TABLE EXIST....LOOP
> m.pk_find = .F.
> LOOP
> ELSE
> m.pk_find = .T.
> DO CASE
> CASE ALLT(UPPER(mOdeType)) == 'UPDATE VTRANS'
> INSERT INTO KEYTABLE (KEY_ID, HTC_PROC) VALUE (oVar.KeyId, 'VTRANS UPDATE')
>
> CASE ALLT(UPPER(mOdeType)) == 'PRINT REPORTS'
> INSERT INTO KEYTABLE (KEY_ID, HTC_PROC) VALUE (oVar.KeyId, 'PRINT REPORTS')
>
> ENDCASE
>
> SELE KEYTABLE
> =TABLEUPDATE(.T.)
> ENDIF
>ENDDO
>
>RETURN