I have a database for all the files on a disk. I have several disks. Sometimes I want to rescan one disk but not the others, so I keep each disk's database in its own file, i.e. disk1.db, disk2.db, disk3.db, etc.

is there a quick way to combine all these databases so that I can make one SELECT call for all db's? I know I could pre-process all the db's into one memory db by doing a "select *" on each & then inserting the results into the memory db, but I'm hoping to find something a little quicker/easier/ more efficient/correct.

If it matters, I'm doing this in python2.5 with the sqlite3 module.

Thanks.

If this is a one-off, why don't you just dump the databases and import them into the preferred database?

It's not a one-off. The system is a bit more complex than I alluded to (the "disks" are actually multi-terabyte arrays and the db files may be generated by deamons at regular intervals), but principle is the same - merge two or more dbs together & use the combined db for select statements.

At the end of the day, I may still do what you suggested, as it's the obvious solution (even if done on the fly), but I was hoping for something a little more clever or by design.

I was/am looking at similar problem

See http://souptonuts.sourceforge.net/re..._tutorial.html

I'm not certain if sqlite supports the schema naming convention within the select (MySQL does) so you would do do something like SELECT * FROM disk1db.name, disk2db.name; However I'm guessing that you would then end up with a Cartesian join, can't you just tell I've never done this. Probably because I would have approached the problem differently by adding an extra field in the table that would identify the disk. That way a SELECT * FROM diskdb WHERE disk = disk1 would return the details from one disk and SELECT * FROM diskdb would return everything.

Keeping an extra column to identify disk/host would not solve my problem. What I need to do, ideally, is "attach" (combine) tables/databases/database_files from geographically/logically distributed hosts (so that I don't have to continually transfer large data from the remote hosts to the central host).