So I had this question as well.

However I am also interested on how you can make a hash of arrays memory efficient.

• I am adding arrays to a hash.
• I am adding several arrays to one hash key
• I am doing this to then compare the contents of arrays within one key for differences.
• My processing will use over 84566 Hash Keys
• My Processing will capture over 94269 arrays (Each with 30 values)

push @{ $eventHash {$EventScheduleReference} }, [@tmpxmlarray];
[download]

However this uses all my available memory and the program terminates.

I have tried using the following to avoid RAM usage

tie (%eventHash, 'DB_File', 'eventHash.db', O_CREAT|O_RDWR, 0644) ||
        die ("Cannot open eventHash.db");
[download]

However memory use seems to be the same and the program terminates. Also..interestingly when using strict refs...under a tie'd condition, an attempt to push to a duplicate key causes the following error:

Can't use string ("ARRAY(0x11614e8)") as an ARRAY ref while "strict re
+fs" in use
[download]

Any ideas on how to force the Hash onto Disk and/or use less RAM to store the hash of arrays? I assume I am only storing ARRAY references in the Hash and that the data in the arrays are still being stored semi-anomalously in RAM somewhere....although I can see the array data in the tie'd file...this might just be due to the way a tie uses random memory junk during file creation.