>>Thanks for all your advise.
>>1)I recalled that in Clipper Programming, it has the flexibility to IGNORE
>>the
>> logically deleted record(as if it was deleted physically). This is good,
>>
>> because one need not to pack (VFP 'packs and reindex' automatically)
>> every time a single record is deleted. Otherwise, it will take a long
>>time
>> to pack and reindex if it is a huge Table.
>
>clipper doesn't have primary or candidate keys. the uniqueness of the key field takes precedence over the deleted/not deleted problem.

Hello

I need an advice. What shell I do in this situation.

From the beginning I had one table with a primary key, but then
I noticed that three fields where seldom used and because I don't
have much disk-space I splitted the table into
four tables. The parent table and three tables with two fields each
(the primary-key from the parent table and one of the seldom used
fields). In this situation my problem is when I delete a record
in one of the three tables and after a while I want to insert a
new record with the same primary-key. That is why I asked
yesterday (How do I use RECALL) if this is a solutien to my problem

SET DELETED OFF
IF SEEK(...)
RECALL
UPDATE ...
ELSE
INSERT ...
ENDIF
SET DELETED ON