Bug #44174 for Crypt-OpenPGP: could dependency on Math::Pari be dropped for Math::BigInt?

Sun Mar 15 17:46:19 2009 dkg [...] fifthhorseman.net - Ticket created

Subject:	could dependency on Math::Pari be dropped for Math::BigInt?
Date:	Sun, 15 Mar 2009 17:45:50 -0400
To:	bug-Crypt-OpenPGP [...] rt.cpan.org
From:	Daniel Kahn Gillmor <dkg [...] fifthhorseman.net>

From a brief review of the source of Crypt::OpenPGP, it looks to me like the dependency on Math::Pari is really only being used for its large integer support. However, Math::Pari is significantly more complicated to build and install than most perl modules (see its RT summary), and large integer support is also available in modern perl distributions from the Math::BigInt module. If Crypt::OpenPGP could be rewritten to Math::BigInt instead of Math::Pari, it would be easier to install and maintain (and could take advantage of new BigInt implementations transparently). --dkg

Download signature.asc
application/pgp-signature 890b

Message body not shown because it is not plain text.

Thu Dec 10 19:28:12 2009 cpan [...] stupidfool.org - Correspondence added

The short answer is yes, almost certainly... we did something similar to this for Crypt::DSA a couple of years ago, and since Math::BigInt supports pluggable backends (in that case, we chose GMP and Pari), it didn't lead to an appreciable performance decrease, in my memory. It'd be a little more complex here, since we'd also need to rewrite Crypt::RSA (and any of its dependencies) to use Math::BigInt. In short, it's not a project that I have a huge desire to do, but I'd welcome patches.

Thu Dec 10 19:28:13 2009 The RT System itself - Status changed from 'new' to 'open'

Tue Dec 07 14:26:42 2010 cpan [...] stupidfool.org - Correspondence added

Crypt::OpenPGP version 1.05, shipped to CPAN yesterday, addresses this, thanks to a patch from Sam Crawley. Note that Crypt::RSA still depends upon Math::Pari, though.

Tue Dec 07 14:26:42 2010 cpan [...] stupidfool.org - Status changed from 'open' to 'resolved'