How much faster is this approach than simply replaceing the primary key with RECNO() (or STR(RECNO()) in the deleted record and adding a new record? Foxpro is pretty fast at this stuff! Even if it's slower, it's much less code to maintain and debug and etc ...

Also, if you used a "replace with recno" scheme, you can take the filter off the primary key. I think that an index with a filter cannot be optimzed using rushmore.



>Hi -
>I have a table that is table buffered. (Here it is Optimistic, but it happens with Pesismistic too.) It has a primary key field with a filter on !deleted().
>
>The problem is that during a particular session the user will be adding/deleting/changing multiple records. On the save, I'd like to find the most efficient way to make these updates. What I'd like to do is delete a group of existing records and then recycle them as I add my new group of records.
>
>However, unless I do a tableupdate after the delete, I get uniqueness violations. The tableupdate sort of defeats the purpose of having the table buffering.
>
>Any ideas?
>
>
>Steps to reproduce the problem:
>
>USE MyTable
>SET ORDER TO MyPrimaryKey
>cursorsetprop('buffering',5)
>BROWSE
>
>1. change a record's id to 1
>2. move pointer off record
>2. move back and delete the record
>3. change the next undeleted record's id to 1
>4. move pointer off record
>5. error
>
>TIA -
> ellen