In our import routine, we delete the indexes if we are importing above a certain number of records. Afterwards, we run code similar to what I posted to recreate the indexes. For large numbers of records (50,000 or higher), this can make a significant savings in time. I don't have any exact numbers, but it was enough to need to put that code in.
>Have you tested this to see if it's faster than having the indexes there before the append? (I'm not asking you to do it if you haven't... Just wondering if you had...)
>
>Thanks,
>
>Michelle
>
>
>>You could loop through the indexes in the source table and build an INDEX ON command string for each index. We have similar code that looks something like this:
>>
>>
>>FOR ln_Loop = 1 TO TAGCOUNT()
>>
>> lc_IndexExpression = "INDEX ON " + SYS( 14, ln_Loop ) + " TAG " + ;
>> TAG(ln_Loop)
>>
>> IF !EMPTY( SYS( 2021, ln_Loop ) )
>> lc_IndexExpression = lc_IndexExpression + " FOR " + SYS( 2021, ln_Loop )
>> ENDIF
>>
>> IF DESCENDING( ln_Loop )
>> lc_IndexExpression = lc_IndexExpression + " DESCENDING"
>> ENDIF
>>
>> IF UNIQUE( ln_Loop )
>> lc_IndexExpression = lc_IndexExpression + " UNIQUE"
>> ENDIF
>>
>> IF CANDIDATE( ln_Loop )
>> lc_IndexExpression = lc_IndexExpression + " CANDIDATE"
>> ENDIF
>>
>>ENDFOR
>>
>>
>>Note that this doesn't work for Primary Key indexes that are created with ALTER TABLE <> ADD PRIMARY KEY ...