Hi

Whatever the above you use, you need to have a concept. Databases are doing this type of joins with two basic algorithms: sort merge or nested loop.

The nested loop works the following way:

• Loop around one of the table.
• Set up the key made of your 3 columns.
• Loop around the second table and merge the rows each time the key match.
• Move to the next row of the first table and repeat the process.

I do not recommend this algorithm because it has an awful performance impact. Actually it is one of two ways to take down a server for a couple of hours or weeks!

The second way to do it is called sort merge:

• Sort file A on all 3 columns of your key in ascending way. Eventually use the sort command of the OS. It will use less RAM and be probably faster.
• Sort file B exactly the same way.
• Retrieve the first row of file A.
• Check if the keys of the first row of file B match. They should.
• Work yourself down file B as long as the keys match.
• When the keys in B do not match any more, move to the next row of A.
• Start at the same position in B and work yourself down as long as keys match.
• ...

The above assumes one of both file has only one row per composite key ( your 3 columns). If this is the case you will process both files only once and your files are merged. If you have a so called many to many relation between the keys of your two files, you will have to work down file B several times.

I recommend you start merging the two smallest files first, then you sort the result and add the next file with a new merge.

Eventually search CPAN or the net for sort merge examples or modules.

K