Is it possible to pre allocate space to a DBM-Deep database file?

[SonOfTheLama]

I am using DBM-Deep 1.0013 in a Win32 ActiveState PERL environment. I
wrote a program to process ~800MB text files with 10->100 million ~60
byte lines. I am storing portions of the text in a multilevel array
which I use to compare data from 1 line of text to data from all other
lines of text in a given file. Processing this much data, in this way,
is going to take a while but I’d like to optimize the process as best
I can.

Initially the program reads in the source file & DBM-Deep stores the
relevant data in the database file. I can see the file growing
incrementally as it receives the data and this phase is taking hours
to complete. I know from past experience that growing a file
incrementally can be a costly process from a IO/file system/OS
standpoint & I’m wondering if it would be possible to pre allocate a
large amount of space to the database file prior to populating it with
data so it does not have to grow incrementally?

Any advice would be most appreciated.

Thank you,

[Rob Kinyon]

You most certainly can pad the file with \0. That could possibly help
and, if you want, I'll gladly accept a patch for doing that.

As for what you're doing ... Have you thought about hashing the
various components you're comparing and creating a hash of arrays vs.
a multi-level array? DBM::Deep's manipulation of arrays is extremely
poor. It treats arrays as hashes with numerical keys.

Rob

--
Thanks,
Rob Kinyon