Re: Is this a job for a fork?

Sean O'Rourke's code for the Wide Finder benchmark does something very similar, you might need to change only a few lines. The idea is to partition the input file into n partitions for n processes, forked via a pipe open. Each one works independently on its part of the file, storing its result into a hash, then passing the hash back to the parent on STDIN using Storable. Then you just need to merge the hashes into one.

Comment on Re: Is this a job for a fork?

Replies are listed 'Best First'.
Re^2: Is this a job for a fork? by BrowserUk (Patriarch) on Jul 13, 2010 at 08:48 UTC
The idea is to partition the input file into n partitions for n processes, Perhaps you can explain how this schema will allow "I'm looking for pairs ie A1:A2, B1:B2 etc. Each member of the pair can exist anywhere in the file. "?	[reply]
Re^3: Is this a job for a fork? by thomas11 (Novice) on Jul 13, 2010 at 09:32 UTC
Yeah, I guess it doesn't. Don't know what I was thinking there. Damn, this was my first post on perlmonks :-( The partitioning solution would only work if you already had a list of potential pairs, which you could then check in parallel chunks.	[reply]