google sparse hash project

alfarid23 has asked for the wisdom of the Perl Monks concerning the following question:

greetings to you, monks.
number of my friends, seasoned c++ programmers, advised that google has couple of a very(most?..) efficient hash implementations open-sourced and ready for use under BSD license, which is using STL.
http://goog-sparsehash.sourceforge.net
So i was thinking to attempt to write a perl module to encapsulate google's algorithms and provide efficient hash functionality for perl both for speed and size. to be concrete, i seek advise whether i should attempt this, and should i use XS vs. SWIG or maybe something entirely different to wrap that monster.
thank you, farid.

p.s.: another g-project, that would be nice to wrap to analyze apache modperl2 threaded is, and possibly improve malloc behaviour: http://goog-perftools.sourceforge.net/

Comment on google sparse hash project

Replies are listed 'Best First'.
Re: google sparse hash project by grinder (Bishop) on May 31, 2006 at 10:17 UTC
The question of hashes came up on the perl5-porters mailing list earlier this month. No definite decision to change the internal hash function was reached, but a number of hash implementations are compared and benchmarked. You might want to see how this Google hash fares. (Assuming we are talking about the issue). Update: it would be better if I posted the links: See http://dev.perl.org/perl5/list-summaries/2006/20060501.html and http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2006-05/msg00261.html • another intruder with the mooring in the heart of the Perl	[reply]
Re: google sparse hash project by Joost (Canon) on May 31, 2006 at 09:25 UTC
Personally, I like XS, but I think SWIG is just as capable. Use whatever feels natural to you :-) Just a remark: the hashing functions in the development version of perl have changed recently. You might find it interesting. See this thread on perl5-porters. "What should it profit a man, if he should win a flame war, yet lose his cool?"	[reply]
Re: google sparse hash project by salva (Canon) on May 31, 2006 at 09:17 UTC
Perl already supports hashes as a native datatype, so there would be little benefit in adding support for another implementation. The extra overhead caused by the wrapping would probably cancel any possible performance gain. Besides that, template based C++ libraries are not good candidates for wrapping because the templates have to be instantianted at compile time and that doesn't match well with the semantics of a dynamic language as Perl.	[reply]
A reply falls below the community's threshold of quality. You may see it by logging in.