This is also how we do it. We have a utility that is run on our clients machines everytime we do an update of the tables and it performs these steps. We move the tables to an archive directory and name the tables with a name/datetime stamp though BEFORE we append from them into the new tables. We have a utility that we run on existing tables and it creates a prg for us that has a procedure to create the table and one to create the indices. That prg is then used in our update programs.
Tracy

>Hmmm,
>So we can, after copying everything to the backup directory:
>Replace all of the client's database/tables with empty, updated database/tables from the library.
>Run the generate client dictionary and structure compare code to log all the changes.
>DOS COPY tables, for no hits. APPEND FROM (and initialize if needed) for any hits in a table.
>
>We have no Database considerations.
>This is like my 1st senario only less complex.
>Have you automated this? done this remotely? Any "Lights out" problems? considerations?
>Thanks,
>Carol Dewar
>
>>Well, what I do is keep an empty structure of the database (+ tables), copy everything to a temp directory, and issue APPEND FROM for each table. Additional considerations: 1) in the case of RI, the parent table has to be copied before the child tables. Otherwise (if the third RI column is "R") you get "trigger failed". 2) to speed things up, I do a detailed comparison of table structures. If the structures are identical, the table is copied with the COPY command.
>>
>>HTH, Hilmar.