>>The code to clean up after yourself during closing of a temporary table is no more complex than the code needed to 're-use' a cursor to make it read-write
>
> Thanks for the excellent explanation. I've known something important and new for me. I may revise my old applications to implement index on field to .idx instead of making cursors read-write. I will definitely consider it in a future.

I also omitted the option of using a non-structural CDX - INDEX ON foo TAG foo OF Bar.CDX - only one file is created, which can be an issue with lots of open file handles; the non-structural CDX with lots of tags uses only one file handle, while multiple IDX indexes use one per index.