Subject: | consider moving Crypt::RSA to Math::BigInt? |

Date: | Fri, 11 Dec 2009 18:24:51 -0500 |

To: | bug-Crypt-RSA [...] rt.cpan.org, Salvatore Bonaccorso <salvatore.bonaccorso [...] gmail.com>, Benjamin Trott <cpan [...] stupidfool.org>, Vipul Ved Prakash <mail [...] vipul.net> |

From: | Daniel Kahn Gillmor <dkg [...] fifthhorseman.net> |

Hi Vipul--
Would you consider patches (if i could provide them) that would make it
so that Crypt::RSA used Math::BigInt instead of Math::Pari?
Math::BigInt can use either GMP or Math::Pari as a backend, so a change
like this could make Crypt::RSA a bit more versatile.
My ultimate reason for this is that Math::Pari has proven very difficult
to get into binary-distributed operating systems like debian because of
upstream's preference for building from source everywhere.
And i'd really like to see Crypt::RSA become available in debian.
I'm Cc'ing Salvatore Bonaccorso here, who has indicated an intent to
package Crypt::RSA for debian, if we can arrange to have the
dependencies sorted out [0].
Also, i'm CC'ing Benjamin Trott (upstream of Crypt::OpenPGP, which
relies on Crypt::RSA, has indicated a willingness to accept patches for
Crypt::OpenPGP to move to Math::BigInt if i can get Crypt::RSA to do the
same [1]. He also had some experience in porting Crypt::DSA to
Math::BigInt several years ago.
So what do you say? i'm willing to take a stab at the work, if a change
like this is something you'd consider accepting. I suppose if you're
dead set against it, forking is an option, but i'd much prefer to have
your guidance and oversight on an undertaking like this.
What do you think? is this doable? is it a good idea? is it madness?
Are there a further dependency chains we'd need to follow (e.g.
Crypt::Random, etc) to be able to use a Math::BigInt-based Crypt::RSA?
Regards,
--dkg
[0] http://bugs.debian.org/532839
[1] https://rt.cpan.org/Public/Bug/Display.html?id=44174

Message body not shown because it is not plain text.